DEV Community: EvvyTools

Why the H2 Outline Is the Section of the Content Brief Most Likely to Be Ignored

EvvyTools — Mon, 15 Jun 2026 09:54:56 +0000

If you have ever compared a content brief's H2 outline against the published article, you have probably seen this: the headings are not the same. A section got dropped. Two got merged. A new one appeared. The article is still on-topic, the writing is still fine, but the structure does not match the brief.

This is the single most common form of brief drift, and it is almost always blamed on the writer being undisciplined. It is not. The H2 outline is the section of the brief structurally most likely to be ignored, and the reasons are predictable.

Headings without word counts are categories, not constraints

A brief that lists H2s as titles only - "What is X," "Why it matters," "How to do it" - has given the writer category labels, not constraints. Within a section labeled "Why it matters," the writer can produce 80 words or 800 and either looks like they hit the brief.

So the writer optimizes for what they care about: rhythm. They balance section lengths by feel. The section they like gets longer. The section they do not get shorter. The brief had no answer to "how long should this be," so the writer's instincts answered it.

Word counts per heading are the fix, and they have to be tight enough to matter. "150-200 words" constrains. "200-500 words" does not.

Headings without a stated purpose get reinterpreted

"What is X" is a category. Without a one-line purpose - "definitional grounding for readers who do not know the term" - the writer is going to reinterpret it. They might use the section to argue why X is important, or to compare X to Y, or to set up the rest of the article. All are reasonable. None match what the brief actually meant.

A heading with a stated purpose is a contract. A heading without one is a Rorschach test. Two writers reading the same outline will produce structurally different drafts, and neither of them is being lazy.

Outline-by-headings ignores the article's center of gravity

Every article has a center of gravity - the section that does most of the load-bearing work. In a how-to, it is the steps. In a comparison, it is the side-by-side. In a definitional explainer, it is the definition plus implications.

A brief that lists eight equal-weight H2s has not told the writer where the article's center of gravity is. So the writer picks one on instinct. Sometimes they pick the section closest to their own expertise. Sometimes they pick the section they have a strong opinion about. Sometimes they spread the load evenly across all eight, which produces a thin-everywhere article.

Word counts per heading also solve this. If "How to do X" is 700-900 words and every other section is 150-300, the writer knows where the article actually lives.

The brief shows H2s, not the order they will be read in

Writers do not read briefs in section order. They scan for the headings table because it is the most concrete thing on the page. The intro framework, the source URLs, the CTA, the no-list - all of those get read later, after the writer already has a mental model of the article from the H2s alone.

So if the H2 outline carries the constraints but not the purpose statements or word counts, the writer's mental model gets locked in around an under-specified structure. When they hit the source URLs and CTA paragraph in the brief, they are working backward to fit them into the structure they have already committed to.

The fix is to put the H2s last in the constraints page, after the intro framework, source URLs, and CTA. The writer reads the constraints in order, then sees the headings as the final assembly instructions. The mental model forms with all the constraints already in mind.

Photo by cottonbro studio on Pexels

The outline is treated as fixed, but it should not always be

This one is the inverse of the first four. Briefs sometimes specify the H2 structure so rigidly that the writer cannot fix obvious problems. If the brief says "Section 4 must be 'Mistakes to avoid'" and the article actually has nothing useful to say in that section, the writer either fills it with filler or fights with the brief.

A surviving brief is explicit about which headings are negotiable and which are not. "Section 4 must address common pitfalls; you can rename or restructure if a better frame appears" is a constraint with room. "Mistakes to avoid - exact heading" is a constraint without room. Both are valid. Mixing them up creates friction.

Google's helpful content guidelines push toward structure that serves the reader, not structure that satisfies the brief. A brief that locks down headings in places where reader intent has shifted is going to produce articles that read as ranking-driven rather than reader-driven. The Schema.org Article specification is a useful background read if you want the structured-data side of why heading hierarchy matters as a signal.

What a working H2 outline looks like in the brief

A working H2 outline in the brief has three columns: heading title, target word count range, purpose in one phrase. It is positioned last in the constraints page, after intro framework, sources, and CTA. Headings that must stay exactly as written are flagged with a "fixed" annotation; the rest are negotiable.

Here is the same outline in working form:

Heading	Words	Purpose	Flexibility
What is X	200-250	Definitional grounding	Title flexible
Why it matters now	250-300	Stakes for the reader	Title flexible
How to do X step by step	700-900	The actual guide	Fixed heading
Common pitfalls	200-250	Pitfalls and edge cases	Title flexible
Wrap and CTA	150-200	Drive to action	Fixed heading

The writer reads that and has zero room to invent structure. They also have permission to rename three of the five headings if a better frame appears mid-draft. Both are necessary.

A free Content Brief Builder handles this format natively - the H2 outline ships with word count ranges and a purpose column on every brief it generates. The full longer piece on what survives the draft is on the EvvyTools blog and walks through the other four sections that drift along with the H2 outline.

How to audit the H2 outlines in your current briefs

Pull the last ten briefs your team produced. For each one, mark the H2 outline section. Then ask three questions:

Are there word count ranges on each heading? If no, every heading is a category, not a constraint.
Is there a one-line purpose statement on each heading? If no, two writers will reinterpret the heading differently.
Is the article's center of gravity (the section that should carry the most weight) clearly the longest target range? If no, the writer will distribute weight on instinct, usually evenly.

If any of those three answers is "no" on more than half of the briefs, the outline format is the leverage point. Fix the template once. Subsequent briefs all inherit the fix.

Common pushback from writers

When you tighten the outline format, expect writers to push back at first. "Word counts per heading constrains my voice." They do not - the ranges are wide enough to leave room for paragraph rhythm, and they constrain only the section length, not the prose inside it.

"Purpose statements make me feel like I am filling in a worksheet." Not really - the purpose statement just tells the writer what the section is for, which the writer already had to figure out on their own before. The brief just supplies the answer.

The Wikipedia entry on content marketing is a good reference for the underlying principle: consistent structure across a content series compounds for topical authority, and the structure has to come from somewhere. If it does not come from the brief, the writer invents it, and consistency suffers.

The format the outline should ship in by default

Every brief generated by your team's tooling should default to: heading title, word count range, purpose statement in one phrase, and a flexibility flag (fixed or negotiable). Anything less is leaving the writer to invent constraints, which is when drift starts.

Briefs that ship in this format produce drafts that come back close enough to the intended structure that the editorial pass is a polish. Briefs that ship without it produce drafts that need restructuring, and the cost compounds across the content series.

The H2 outline does not have to be the section that drifts. It is the section that drifts because most briefs hand the writer category labels and call it an outline. Hand them constraints with purpose statements, and the structure holds.

How to Set Up a Content Brief Template That Holds Across Multiple Writers

EvvyTools — Mon, 15 Jun 2026 09:54:54 +0000

Most content brief templates fall apart at the second writer. The first writer who used it learned how to read it. The second one reads it cold and sees a wall of bulleted attributes, no signal about which ones matter most, and starts inventing structure to fill the gap. By the fifth article, the template has been informally re-interpreted three different ways.

Here is a step-by-step for setting up a brief template that survives more than one writer using it.

Step 1: Split constraints from context, visually and structurally

Open the template in whatever tool your team uses (Google Docs, Notion, Coda, Airtable). The first page is constraints. The second page is context. They should look different - different headings, different visual weight, a horizontal rule between them.

Constraints get a numbered list with bolded section labels. Context gets a normal table or bullet structure. The point is that a writer scanning the brief for the first time should see immediately that page one is "things you cannot change" and page two is "things to read once."

Stack the constraints in this order:

Intro framework (4-5 sentences as a structure, not the actual intro)
Heading hierarchy with word count ranges
Source URLs with one-line notes
CTA paragraph (drafted in the brief)
No-list (what not to write)

The order matters. Writers read top to bottom, and intros are the part most likely to drift.

Step 2: Pre-fill the intro framework as four bullets

In the template, do not write "open with a hook." Write:

- Sentence 1: Name the reader's specific frustration in their own words
- Sentence 2: Confirm it is a real problem (one stat or named example)
- Sentence 3: State what this piece will give them and what it will not
- Sentence 4: Transition into the first H2 without recapping

The four bullets stay the same on every article in the template. Per-article, the brief writer fills in which frustration and which example. The writer reads four bullets and writes four sentences. There is no room to negotiate.

Step 3: Attach word count ranges to every heading

The template's heading table has three columns: heading title, target word range, purpose in one phrase.

Heading	Word range	Purpose
Intro	150-200	Frame the problem
What is X	200-300	Definition
Why it matters now	250-350	Stakes
How to do X	700-900	The actual guide
Common mistakes	200-300	Pitfalls
Wrap and CTA	150-200	Drive to action

The ranges add up to the article's target word count, give or take 100 words. Per-article, the brief writer customizes the headings but not the ranges. The writer sees the ranges and knows where the article actually lives.

Step 4: Build a source URL bank the brief pulls from

In a separate tab or table, keep a source bank with these columns: source name, URL, one-line note, topic tags. When writing a per-article brief, pull three to five URLs from the bank into the sources section.

A source bank pays back hard. The brief writer stops Googling for citations. The writer pastes vetted URLs. The editor's fact-check drops to a five-minute pass. And the Google Search Central blog point about consistent, authoritative sourcing across a content series gets built in by default.

Vet the bank quarterly. URLs go stale. Schema.org's Article type page is unlikely to move, but a blog post slug from 2022 might. Keep the bank to homepage-level URLs and Wikipedia entries where possible.

Step 5: Write the CTA paragraph inside the template

The CTA section of the template should be a full paragraph, not a bullet. Per content type and funnel stage, draft it once:

For a top-of-funnel guide:

If you want the same brief structure prebuilt for your next article instead of writing one from scratch, you can use this free content brief builder by EvvyTools to generate one with the framework, headings, and sources already populated.

For a comparison post:

Most of the patterns above show up across teams running ten posts a week or more. Tools that ship the structure prebuilt - including this free content brief builder by EvvyTools - let you stop writing the constraints layer manually.

The writer pastes the right paragraph and moves on. The team's CTAs stay consistent. The anchor text stays consistent (important for SEO). No one writes filler.

Step 6: Add a no-list per article

The no-list is the most underused section of every brief template. Add a final constraint section called "Do not write this," with one or two bullets per article.

Examples:

Do not use "content brief template" in the H1 or meta title (that query belongs to the landing page)
Do not list more than five tools by name (this is not a roundup)
Do not open with a personal anecdote (audience is content ops managers, not solopreneurs)

The no-list takes two minutes to write and saves a redraft. Without it, the writer pattern-matches to whatever the closest article they have written looks like, which is usually wrong.

Step 7: Pilot the template with two writers, not one

The most expensive mistake is rolling out a template the original brief writer can read because they wrote it. Hand the template to two writers cold. Have them draft from it. Compare outputs.

If the two drafts diverge structurally, the template is leaving room for interpretation in some section. Find the section and tighten it. Repeat until two cold writers produce structurally similar drafts from the same brief.

The Wikipedia entry on content marketing has been edited by hundreds of contributors and stays internally consistent because the section structure is enforced. Your template is doing the same job at a smaller scale.

What the template is actually doing

The template is not making the writer better. It is removing the places where the writer has to invent structure. Invention is where drift happens. Strip those out and the brief survives across writers.

There is a longer guide on this with examples on the EvvyTools blog - it covers the underlying problem (where writers drift) which makes the seven steps above easier to internalize. The Content Marketing Institute also has a body of work on editorial workflows worth reading once you have your template piloted.

The template is a one-time setup with ongoing returns. Build it once, vet the source bank quarterly, run the two-writer pilot before deploying. Then briefs stop drifting and you stop rebuilding the same article five times.

Step 8: Version the template and review it quarterly

The template is not a one-time artifact. Audience changes. SEO standards change. Source URLs go stale. A template that was right in 2024 is not necessarily right in 2026.

Schedule a quarterly review. Look at the last twenty articles produced from the template. Mark the sections that were rewritten most often on the editorial pass. Those are the sections where the template is leaving room for interpretation. Tighten them in the next version.

Also audit the source bank. Pull every URL with a simple link checker, drop the dead ones, and refresh the topic tags as your content series expands. The Schema.org Article reference is unlikely to move, but other URLs do.

Version the template explicitly - v1.0, v1.1, v2.0 - so writers and editors know which version a brief was generated from. That makes it possible to attribute drift to a template revision rather than to a writer, which keeps the conversation about format instead of about discipline.

What the template earns you over a year

A team running ten briefs a week through a constraints-first template versus a loose one will see meaningful differences in editorial throughput by quarter two. Editorial pass time per article drops noticeably. Briefs themselves take fifteen minutes longer to write but save thirty to forty-five minutes per article on the edit.

Across a year, that is several hundred hours of editorial time recovered. The template is one of the highest-leverage one-time setups in a content operation.

Why the Avalanche Method Wins on Interest But Often Loses on Follow-Through

EvvyTools — Sun, 14 Jun 2026 09:55:15 +0000

If you ask a developer how to pay down multiple debts, the answer is almost always "avalanche, obviously." Highest APR first, lowest total interest, math is right there, case closed.

The math is right. The follow-through usually isn't. And in a multi-year payoff, follow-through is most of what determines whether you actually hit the date.

This is about why the obvious mathematical answer isn't always the right answer for the same person who came up with it, and what to do about that.

The avalanche promise

Avalanche says: pay the minimum on every debt, then put every extra dollar on the debt with the highest APR. When that one clears, roll its payment onto the debt with the next-highest APR. Repeat.

Mathematically, this minimizes total interest paid. Always. Snowball can never beat it on dollars. If the only thing you cared about was paying the smallest total of dollars over the life of the payoff, avalanche is the answer and there is nothing to discuss.

The full math is in the Federal Reserve consumer education materials if you want the formal derivation, but you can verify it yourself with any multi-debt model in five minutes.

Photo by Vitaly Gariev on Pexels

Why engineers in particular get burned

Engineers are the demographic most likely to choose avalanche and most likely to abandon it nine months in. Three reasons.

First, the highest-APR debt is usually the largest balance. Credit cards have higher APRs than car loans and student loans, and your worst-rate card is often the one with the biggest balance because that's the one that's been accruing interest fastest. So month 9 of avalanche looks identical to month 1 of avalanche: still chipping at the same big card, no debt has cleared, no sense of progress.

Second, engineers underestimate the cost of zero positive feedback over multi-year horizons. A two-year payoff with no debt clearing for the first 18 months is psychologically brutal. Even very disciplined people start asking whether the math is worth it around month 11, and the question itself is dangerous because once you start asking, you sometimes answer no.

Third, engineers tend to over-trust the model. "I ran the numbers, avalanche saves $1,400 over snowball, end of discussion." But avalanche only saves $1,400 if you actually follow it for the entire payoff window. If you drop out in month 14 and start splitting payments emotionally, you might end up paying more total interest than snowball would have, because the strategy you didn't follow is worth zero.

When avalanche actually works

Avalanche works when you have one of these conditions:

The interest savings is large enough to motivate you on its own. If running both strategies through a multi-debt model shows avalanche saves $2,500+ in interest and 8+ months, the math has real weight and you'll probably stick with it.

Your largest balance is not your highest-APR debt. If your worst-rate card is a $1,200 balance and your $15,000 debt is a 6% car loan, avalanche kills the $1,200 card in month 2 or 3, you get a quick win, and the strategy holds.

You have a partner or accountability mechanism. Avalanche during the long middle months benefits enormously from someone else looking at the schedule and saying "yes, this is still working, you're on track." Without that, the long middle is the danger zone.

When snowball actually wins

Snowball wins when one of those conditions doesn't hold and the cost of avalanche failing is higher than the cost of snowball succeeding.

A snowball payoff that you actually complete in 38 months is better than an avalanche payoff that you abandon in month 14. The interest savings on avalanche are zero if you don't follow it.

The honest answer is: pick the one you'll actually stick with. Run both, look at the dollar difference and the month difference. If avalanche saves you 3 months and $400, snowball is fine. If avalanche saves you 11 months and $2,800, fight for avalanche.

The hybrid that engineers actually adopt

The version of avalanche most disciplined people end up running in practice is "avalanche, except if the next-targeted debt's APR is within 2 percentage points of a smaller-balance debt, kill the smaller one first." You get most of the interest savings and one or two psychological wins along the way.

You can model this in any multi-debt planner by manually re-ranking your debts. The Debt Payoff Planner lets you swap strategies in place, so you can see what the hybrid actually costs versus pure avalanche.

In my experience, the cost of one hybrid swap is something like $80-$200 in interest and a couple of weeks of payoff date, which buys you a clean psychological win that materially raises the probability you finish.

How to know which one you are

Ask yourself this: if I told you the schedule was 36 months, and the first 14 months would show no debt fully clearing, do you believe yourself when you say you'd stay on plan?

If yes, run avalanche. The math is on your side, and the math will pay you.

If you hesitate even a little, snowball or hybrid will probably get you further than pure avalanche. The schedule that finishes is worth more than the schedule that's theoretically optimal.

A worked example for engineers

Three debts: a $6,200 credit card at 24.99%, a $1,800 credit card at 19.99%, and a $11,000 personal loan at 8.5%. Minimums total around $410. You can put $750 a month toward debt.

Pure avalanche. Targets the 24.99% card first. The big card clears around month 23. First win comes well over a year and a half in. Total payoff in roughly 41 months. Total interest: about $5,800.

Pure snowball. Targets the $1,800 card first. First clear in month 8. Second clear (the 24.99% card) in roughly month 29. Total payoff in roughly 43 months. Total interest: about $6,300.

Hybrid (snowball through the small card, then avalanche). First clear in month 8. Pivots to the 24.99% card. Total payoff in roughly 41 months. Total interest: about $5,900.

The hybrid gives you 95% of the avalanche savings and the snowball's early win. For most people, that combination is the highest-probability path to actually finishing.

You can verify these numbers yourself in any multi-debt model. The exact dollars depend on the minimums and how the model handles them, but the rank ordering of the three approaches will be stable.

What to do if you've already abandoned a previous attempt

A common situation: someone tried avalanche eighteen months ago, lost momentum somewhere in month 10, and has been paying inconsistent amounts since. The schedule is now meaningless because the inputs aren't accurate.

Restart from zero. Don't try to "get back on" the old plan. Pull current balances today, pick a strategy (probably hybrid or snowball, given the prior abandonment is a signal), and build a fresh schedule. The old schedule is sunk cost; only the new one matters.

This is also a good moment to bring in an accountability mechanism if you didn't have one before. The National Foundation for Credit Counseling has free counselors who serve roughly that function for the price of a single call.

Re-running every quarter

Whatever you pick, re-run the model every three months. APRs drift, balances change, life happens. The model is only useful if it reflects current reality. The full walkthrough of how to do this is at How to Build a Debt Payoff Plan That Hits a Real Date if you want the longer version.

For the rest of the personal finance toolkit (emergency fund sizing, refinance break-even, retirement contribution impact), see what's available at EvvyTools. The Consumer Financial Protection Bureau is the regulatory reference for any of the broader debt-management questions a calculator can't answer.

The right answer to "avalanche or snowball" is "whichever one you'll actually finish." Run both, pick honestly, schedule the quarterly re-checks.

How to Set Up a Multi-Debt Payoff Schedule in a Spreadsheet

EvvyTools — Sun, 14 Jun 2026 09:55:14 +0000

If you are the kind of person who would rather build a model than trust a black box, this is how you set up a debt payoff schedule in a spreadsheet so you understand exactly what the math is doing.

The end state is a sheet where you enter your debts, an APR, a minimum, a total monthly payment, and a strategy, and you get back the month-by-month schedule that shows when each debt clears.

This walkthrough uses Google Sheets, but every formula works identically in Excel or LibreOffice Calc.

Photo by Tima Miroshnichenko on Pexels

The data layout

Start with a small table at the top of the sheet for your inputs. Five columns:

Debt	Balance	APR	Minimum	Strategy Rank
Card A	4200	24.99	95	1
Card B	1800	19.99	45	2
Car loan	8400	6.49	285	3

The "Strategy Rank" column is what makes this flexible. For avalanche, you rank by APR descending (highest APR is rank 1). For snowball, you rank by balance ascending (smallest balance is rank 1). You can re-rank in 30 seconds to compare strategies.

Add one cell somewhere obvious for total monthly payment. Call it TotalPayment. Add another for extra payment above minimums; call it ExtraPayment. These two are the levers you'll play with.

The monthly schedule

Below the input table, build a schedule. Each row is a month. Each debt gets three columns: starting balance, payment made, ending balance.

For month 1, the starting balance of each debt is the input balance. The payment for the rank-1 debt is its minimum plus the entire ExtraPayment. The payment for every other debt is just its minimum.

Interest for the month accrues on the starting balance. The formula is:

interest = starting_balance * (APR / 12 / 100)

The ending balance is:

ending_balance = starting_balance + interest - payment

If ending_balance goes negative or hits zero, the debt is paid off this month. The "leftover" payment (the amount you tried to pay over what was owed) needs to roll onto the next debt.

For month 2, the starting balance of each debt equals the prior month's ending balance. The payment logic is the same as month 1, but: if any debt cleared in month 1, the rank-1 debt this month is the next-ranked surviving debt, and its payment now equals its minimum plus ExtraPayment plus the cleared debt's old minimum.

The roll-up formula that does the work

The roll-up is the entire reason a payoff strategy beats paying minimums forever. The formula in spreadsheet form is:

target_payment_this_month =
  this_debt_minimum
  + ExtraPayment
  + SUM(minimums of all cleared lower-ranked debts)

The SUM(minimums of all cleared...) term is what most homebrew spreadsheets get wrong. They handle the first debt clearing fine, then forget to roll up when the second debt clears. You can implement it with a SUMIFS that checks which debts have ending_balance = 0 in the prior month.

A safer pattern: maintain a small helper column per month showing which debt is currently being targeted, and compute the target payment from that. It's more verbose but easier to debug when month 17 looks weird.

Sanity checks before you trust the output

Three checks I run on every schedule I build:

Total of all payments across all months should equal total of all original balances plus total interest paid. If it doesn't, money is being created or destroyed somewhere.

The month any debt clears should match the month the running balance hits zero on the per-debt running total. If the schedule says debt A clears in month 14 but the per-debt running balance is still $230 in month 14, the roll-up trigger is off by a month.

Switching strategy ranks should change the total interest but not the total principal paid. Principal is principal regardless of order. If switching from avalanche to snowball changes the total principal paid, there's a formula bug.

These three checks catch about 90% of the mistakes I see in homebrew payoff sheets.

Handling variable APRs

If you want to model a future APR change (you expect the Fed to hike, or you're modeling what a balance transfer does), add an APR column per debt per month rather than a single APR per debt. The interest formula then reads the row's APR rather than the input table's APR.

This is overkill for most plans, but useful if you want to stress-test how much rate sensitivity your schedule actually has. The Federal Reserve publishes quarterly consumer credit summaries that are a decent benchmark for what "realistic" rate movement looks like quarter to quarter.

Adding lump sums and snowflakes

Two extra columns make this model substantially more useful for real life.

Lump-sum column. A monthly column where you can enter any one-off windfalls (tax refund, bonus, gift). The schedule adds the lump-sum amount to that month's target-debt payment. Suddenly you can model "what if I throw the $2,400 tax refund at the highest-APR card in April."

Snowflake column. Smaller irregular payments that don't deserve to be a "lump sum" but should still be tracked. Reselling something for $40, a small side gig payday, a rebate check. These add up; without modeling them, your schedule is pessimistic versus what you'd actually achieve.

Both are simple SUM additions to the target-debt payment line. The schedule recomputes downstream months automatically.

Where the model is honest about its limits

A few things a spreadsheet (or any planner) cannot model precisely.

Missed payments. The schedule assumes every payment lands. If you miss one, the late fee and the interest catch-up create a downstream effect the model won't show until you re-enter the actual current balance.

Credit limit decreases. If you have a credit card that's part of the plan and the issuer drops your limit such that your balance is now over-limit, you might have an over-limit fee or a credit-utilization hit. The model doesn't know about either.

APR resets. Promotional APR ends (a balance transfer's 0% period rolls off), or a deferred-interest medical financing plan crosses the deferral boundary. The model treats APRs as flat unless you build the change in explicitly.

These are the moments where you re-pull current balances and re-run from scratch.

Validating against a battle-tested tool

Before you trust your spreadsheet, validate the numbers against an established calculator. The free debt payoff planner by EvvyTools takes the same inputs (balances, APRs, minimums, total payment, strategy) and outputs the same schedule shape. If your sheet's debt-free date and total interest match the planner's output to within a few dollars, your formulas are right. If they're off by hundreds, hunt the bug.

For the broader walkthrough of how to think about debt payoff planning generally, the longer guide is at How to Build a Debt Payoff Plan That Hits a Real Date.

Engineers who want even more rigor can build the same model in Python with pandas and a small loop. The math is identical; the speed is irrelevant for a 60-month schedule with 5 debts. Use whichever tool you'll actually look at every quarter.

The Consumer Financial Protection Bureau publishes reference rules for how credit card minimums are calculated by issuer category, which is useful if you want your minimum formulas to be more realistic than a flat percentage. For the underlying interest math, Khan Academy's compound interest section is the cleanest free walkthrough I know.

Why bother building it yourself

Three reasons engineers I know give for going the spreadsheet route over a hosted calculator:

You learn what assumptions the calculator is making, which makes you a better consumer of the calculator.
You can model your specific scenario (lumpy income, planned balance transfers, expected raises) more flexibly than most online tools allow.
You actually look at it again because you built it. The schedule becomes a thing you maintain, not a printout you lose.

For most people, a hosted planner is the better answer because they will never re-open the spreadsheet. Pick whichever you'll actually use in 90 days.

You can browse the rest of the EvvyTools finance tools if you want a parallel one already built for adjacent questions (emergency fund sizing, mortgage affordability, retirement contribution impact). They use the same model logic this spreadsheet would.

How to Evaluate a New Employer's 401(k) Plan in 15 Minutes

EvvyTools — Sat, 13 Jun 2026 09:52:32 +0000

The benefits packet at a new job lands with twenty PDFs and a 90-day enrollment window. Most people open the health plan PDF, eyeball the dental coverage, and skip the 60-page 401(k) summary plan description on the assumption that all 401(k) plans are basically the same.

They are not basically the same. The spread between a good 401(k) plan and a mediocre one is, over a 30-year career, somewhere between $150,000 and $400,000 in retirement account value on the same contribution rate. That is worth 15 minutes of reading.

Here is the 15-minute version, the four things to check, and what to do once you have answers.

Photo by Eugenia Remark on Pexels

What You Are Looking For

A 401(k) plan is mostly a wrapper around an investment menu, with three or four parameters around it. The good news is the parameters that matter are short and easy to find in the plan documents. The summary plan description is the canonical source, and federal law requires it to be in plain language. The Department of Labor's plan participant guide walks through what the SPD is supposed to contain.

The four things to check, in order:

The employer match formula.
The vesting schedule.
The investment menu and the expense ratios on the index funds.
The administrative fees baked into the plan.

If those four numbers look reasonable, the rest of the plan probably is too. If they do not, the plan is going to cost you over time, and the right move might be to contribute only up to the match and put additional retirement savings into an IRA outside the plan.

Check 1: The Match Formula

The match formula will be one of three shapes:

Dollar-for-dollar up to X percent. The employer matches 100 percent of your contributions up to X percent of your salary. This is the strongest match.
50 cents on the dollar up to X percent. The employer matches 50 percent of your contributions up to X percent of your salary.
Tiered formula. The first chunk of your contribution gets matched at one rate, additional contributions at a lower rate. Often written as "100 percent of the first 3 percent, 50 percent of the next 2 percent," which is one of the most common formulas in the U.S. workforce.

The headline number to compute: at the contribution rate where the match maxes out, what percentage of your gross salary does the employer add. A 100 percent match up to 6 percent is worth 6 percent of salary. A 50 percent match up to 6 percent is worth 3 percent of salary. The difference, over 30 years, is large.

The rule of thumb: contribute at least enough to capture the full match. Anything below that is leaving compensation on the table. The Bureau of Labor Statistics publishes annual data on what typical employer matches actually look like across industries, useful as a reality check.

Check 2: The Vesting Schedule

Your own contributions are always 100 percent vested immediately. The employer's match is a different story. The vesting schedule determines how much of past employer matches you get to keep if you leave before a certain anniversary.

There are two flavors:

Cliff vesting. Zero percent until year N, then 100 percent. N is capped by federal law at 3 years for cliff schedules.
Graded vesting. A rising percentage each year, capped by federal law at 6 years to reach 100 percent.

The number to know: how many years until you hit 100 percent vested. If the answer is more than two and you are someone who switches jobs every two years, the headline match number is misleading. A 100 percent match that takes 4 years to fully vest is worth a lot less to a 22-month-tenure employee than the same match with immediate vesting.

Photo by Kindel Media on Pexels

Check 3: The Investment Menu

Pull up the plan's investment menu. Look for two things:

Is there at least one broad-market index fund. A total US stock market index, an S&P 500 index, or a total world index. If the answer is yes, you have a low-cost, diversified core option, which is most of what you need.

What is the expense ratio on that index fund. This is buried in the fund's fact sheet but it is always there. Acceptable: under 0.10 percent (10 basis points). Mediocre: 0.10 to 0.30 percent. Bad: above 0.30 percent for an index fund.

For comparison, the same index funds in a retail brokerage IRA usually run 0.03 to 0.04 percent at places like Vanguard or Fidelity. A 0.50 percent expense ratio drag in a 401(k), compounded over decades, costs roughly 15 percent of the final balance compared to a 0.04 percent expense ratio outside the plan. This is the math the Securities and Exchange Commission keeps hammering on in its investor education materials, and it is worth taking seriously.

If the plan has a good low-cost index option, the menu is fine. If it does not, contribute up to the match and put additional savings into an IRA at a brokerage where the expense ratios are lower.

Check 4: The Administrative Fees

Some plans, especially at smaller companies, layer an administrative fee on top of the investment expense ratios. This shows up as a percentage of your balance deducted annually (a "wrap fee"), or as a flat dollar amount. The fee disclosure is required to be in the plan documents under federal regulations the Department of Labor publishes.

The number to look for: total annual plan fees as a percentage of your balance. Add the index fund expense ratio plus any wrap fee. Reasonable: under 0.5 percent total. Acceptable: 0.5 to 1 percent. Problematic: above 1 percent total.

A 1 percent annual fee drag, compounded, eats roughly a quarter of the final balance over a long career. Plans that high are usually small-employer plans where the administrative cost gets pushed onto participants. They are not common at large companies, but they exist.

The 15-Minute Process

Open the summary plan description. Search the PDF for these strings:

"match" or "matching contribution" → the formula.
"vesting" → the schedule.
"investment options" or "designated investment alternatives" → the menu.
"administrative" or "recordkeeping" → the fees.

Twelve minutes of reading, three minutes of arithmetic. The decision then falls into one of three buckets:

Plan is good. Index funds under 0.1 percent, total fees under 0.5 percent, reasonable match. Maximize the match, and seriously consider contributing more than the match if you have the cash flow.

Plan is okay but expensive. Match is good, fund menu is okay, fees are 0.5 to 1 percent. Contribute up to the match. Additional savings go into an IRA outside the plan.

Plan is bad. Fees above 1 percent, no decent index option, weak match. Contribute up to the match anyway (the match is still a 50 to 100 percent return on the contribution), then route everything else outside the plan.

Run the Projection

Once you have the match formula and the fund expense ratios in hand, run the projection. The free retirement calculator at EvvyTools handles this directly: enter your starting balance (zero, or the balance you are rolling in from the old plan), the contribution rate, the match formula, and a realistic net-of-fees return assumption. The result is a year-by-year projection that makes the impact of the plan choice concrete.

For the broader question of what to do with the 401(k) at the old employer when you start at the new one, https://evvytools.com/blog/what-happens-to-401k-when-you-change-jobs/ walks through the four real options. For the rest of the personal finance calculators that pair with the rollover decision, https://evvytools.com has the full set.

Fifteen minutes of reading a benefits packet is one of the highest-leverage uses of time at any new job. The plan is going to compound for decades. Knowing what is actually inside it changes the contribution rate you set on day one.

Why the 60-Day Indirect 401(k) Rollover Is a Tax Landmine for Job-Hopping Engineers

EvvyTools — Sat, 13 Jun 2026 09:50:33 +0000

Engineers job-hop. The numbers from places like Stack Overflow's annual developer survey put the average tenure in the two to three year range, which means a lot of people in this field will, over the span of a career, leave behind four to seven different 401(k) accounts at four to seven different employers. Most of the literature on what to do with those accounts is fine. There is one specific landmine that catches people anyway, and it tends to catch the same kind of person every time.

The landmine is the indirect rollover, specifically the 60-day version, and it is the kind of process failure that an engineering brain will rationalize into existence even when the direct rollover is sitting right there. This piece is about why it happens, what the cost is, and how to make sure it does not happen to you the next time you switch jobs.

Photo by Willfried Wende on Pexels

What an Indirect Rollover Actually Is

A 401(k) rollover can happen one of two ways:

Direct rollover (also called a trustee-to-trustee transfer): you fill out the paperwork, the old plan sends the money directly to the receiving institution. The check, if there is one, is made out to "[Receiving Brokerage] FBO [Your Name] IRA". The money never lands in your bank account. No withholding, no 60-day deadline, no risk of accidentally turning it into a taxable distribution.

Indirect rollover: the old plan sends a check made out to you personally. You have 60 days from the day you receive the check to deposit it into an IRA or another qualified plan. If you miss the 60 days, the entire balance is treated as a regular distribution: taxable income at your marginal rate, plus the 10 percent early withdrawal penalty if you are under 59 and a half.

The kicker on the indirect rollover, even if you make the 60-day deadline: the old plan is required by federal law to withhold 20 percent for federal taxes when it sends the check. That 20 percent does not just disappear, but you have to make it up from your own pocket to roll over the full original balance, or else the missing 20 percent gets treated as a partial distribution that is taxable on its own. The full mechanics are documented by the IRS in Publication 575, which is one of the more readable government tax publications, in the sense that "more readable" is a relative term.

Why This Specifically Catches Engineers

The engineering pattern of failure on this is consistent enough to be worth naming. It tends to go:

The 401(k) statement says "leave us a forwarding address." Engineer puts off the rollover for six months while learning the new job. Stack of mail piles up.
The old plan, after some threshold of inactivity, force-distributes the balance. Smaller balances are at the highest risk; most plans will mail an indirect rollover check rather than chase the account holder forever.
Engineer comes home to a check for thousands of dollars made out to them personally. Brain recognizes this as "money, deposit it in checking, deal with it later."
Later turns out to be more than 60 days. Now the whole balance is taxable.

Or, the variant where the rollover is intentional but mishandled:

Engineer requests the rollover, paperwork asks for receiving institution. Engineer has not opened the new IRA yet, leaves the field blank, expects to "figure it out when the check arrives."
Plan defaults to mailing the check to the participant. Direct trustee-to-trustee transfer never happens.
60-day clock starts. Engineer opens the IRA in week three, gets distracted by a launch in week six, misses the deadline.

The common thread is the same kind of "I will handle it asynchronously" instinct that makes engineers good at their jobs and bad at financial paperwork. The 60-day deadline is non-negotiable, has limited self-certification exceptions documented by the Treasury Department, and does not care that you were debugging a production incident the week the deadline hit.

What the Cost Actually Looks Like

Run the numbers on a worked example. Engineer at 32, $35,000 balance in an old 401(k), 24 percent federal marginal bracket, 6 percent state rate.

Miss the 60-day deadline:

Federal tax: $35,000 × 24 percent = $8,400
Early withdrawal penalty: $35,000 × 10 percent = $3,500
State tax: $35,000 × 6 percent = $2,100
Total immediate cost: $14,000, leaving $21,000 net.

That is the obvious cost. The less obvious one is the foregone growth on the full $35,000. Assuming a 7 percent average return over 30 years until retirement at 62, the $35,000 would have grown to roughly $266,000. The $21,000 left over, if reinvested in a taxable brokerage account where dividends and capital gains get taxed annually instead of compounding tax-free, grows to something more like $130,000 to $145,000 over the same horizon depending on the drag.

Net cost of the missed deadline, over a career: somewhere around $120,000 to $140,000 in foregone retirement value, on what was a $35,000 balance to begin with.

This is the kind of calculation the 401(k) Calculator is built for. Run it once with the rolled balance intact, run it again with the cashed-out net deposited fresh, and the gap between the two end balances is the real cost.

Photo by cottonbro studio on Pexels

How to Make Sure the Rollover Stays Direct

The defensive process is short:

Open the receiving IRA first. Before you call the old plan. The IRA needs an account number that you can hand to the old plan administrator. Vanguard, Fidelity, and Schwab all open these accounts online in under thirty minutes.

On the rollover form, specify direct trustee-to-trustee transfer. Read the form. There is usually a box that distinguishes direct from indirect. Tick the direct one. If the form does not have that distinction, call the plan and ask which option triggers a direct transfer in their system.

Make sure the check, if there is one, is made out to the receiving brokerage. The correct payee is "[Brokerage] FBO [Your Name] IRA". A check made out to you personally is the indirect path, no matter what anyone calls it.

Specify zero federal withholding. Direct rollovers should have zero withholding. If the form forces you to enter a withholding percentage, that is the signal to call the plan and confirm the rollover is actually being processed as direct.

Set a calendar reminder if the money is in transit. Track the rollover for the two to four weeks it usually takes. If a check arrives at your address instead of going to the brokerage, you have caught the problem early enough to redirect it.

The Programming Analogy That Sticks

If you think of a 401(k) rollover as a transactional operation, the direct rollover is a true atomic transfer: money leaves one account and lands in another, with no intermediate state where it sits in your hands. The indirect rollover is a two-phase commit with a 60-day timeout and no automatic rollback. If the second phase fails for any reason, the system does not retry. It just records the failure as a permanent tax event.

You would not design a financial system that way on purpose. Federal tax law did. The fix is to stay in the path that is actually atomic, which means doing the paperwork correctly the first time.

For the full breakdown of the four real options when changing jobs, including the cases where leaving the money at the old plan or rolling into the new 401(k) wins on fees alone, see https://evvytools.com/blog/what-happens-to-401k-when-you-change-jobs/. For the rest of the personal finance calculators that pair with the rollover decision, https://evvytools.com has the full free set.

The job change is already chaotic. The 401(k) decision does not have to be.

Seven Free Text Quality Tools That Pair Well With AI Content Detection

EvvyTools — Fri, 12 Jun 2026 09:55:00 +0000

AI content detection is one signal among several worth checking when you are reviewing writing. A piece can score clean on a detector and still be unclear, full of dead phrasing, or stuffed with cliches. A piece can score 80% AI and still be valuable original work that happened to land in the statistical pattern.

The honest workflow uses detection as one input among multiple, and combines it with other free tools that evaluate different dimensions of text quality. Here are seven that pair well with detection in a routine content review.

Photo by Nataliya Vaitkevich on Pexels

1. Hemingway Editor for Readability

The Hemingway Editor scores text on grade level and flags adverbs, passive voice, complex sentences, and weak phrasing. It is one of the older free tools in this space and still one of the most useful for a quick read of prose mechanics.

Why this is first: most text problems are mechanical, not substantive. A passage rated at grade 14 with 30% passive voice is not connecting with the reader regardless of whether the content underneath is original or AI-assisted. Hemingway catches the mechanical issues before substance becomes the right conversation.

2. Grammarly for Mechanical Cleanup

Grammarly catches the grammar, punctuation, and obvious-word-choice errors that slip past most writers on their own work. The free tier covers the basics, which is what you need for a quality screen.

A passage with twelve grammar errors and a 70% AI detection score has two problems. Fix the mechanics first, then read the substance, then decide what to do about the detection question.

3. Readability Analyzer Without an Account

For a quick readability check without signing up for anything, WebFX hosts a free readability calculator that runs Flesch-Kincaid, Coleman-Liau, and SMOG scores against pasted text. The output is roughly comparable across tools and gives you a defensible reading level number.

Useful when you are screening submissions intended for a specific audience reading level and need to confirm the writing actually lands there.

4. EvvyTools AI Content Detector

The free AI content detector by EvvyTools breaks down detection into the underlying signals: sentence uniformity, vocabulary diversity, AI phrase density, and hedging frequency. The composite score is reported separately from the sub-scores, so you can see what is actually driving the flag rather than treating the headline number as the whole story.

This matters for the screening workflow because most detector false positives are driven by one specific signal (often phrase density on formal writing). When you can see which sub-score is high, you can decide whether that specific signal matters for your situation.

5. CopyScape for Originality

Copyscape checks pasted text against published content on the web to identify lifted or near-duplicate passages. The free version is limited but useful for spot-checks on individual paragraphs.

Detection and originality are different questions. A piece can be 100% original and 80% AI. A piece can be 0% AI and substantially plagiarized. Running both checks lets you separate the two issues and respond to each appropriately.

6. Reverso for Translation Detection

If you suspect a submission was machine-translated rather than originally written in English, Reverso can help. It shows how phrases are used in context across translated texts, which makes translated phrasings visible in a way that a single read might miss.

Machine-translated text scores high on AI detectors for the same reasons original AI output does (uniformity, common vocabulary), so this is a useful diagnostic for separating "this was AI-generated" from "this was translated from another language by a machine."

Photo by COPPERTIST WU on Pexels

7. ProWritingAid for Style Patterns

ProWritingAid goes deeper than Hemingway on style analysis, surfacing patterns like overused words, sentence start variety, sticky sentences, and pacing. The free tier limits document length but is enough for routine screening.

The "overused words" report often catches the same vocabulary patterns that AI detectors flag as AI signal. If ProWritingAid is flagging "delve," "in conclusion," and heavy hedging, the detector is likely to flag the same piece, but for a more interpretable reason.

How These Fit Together

In a routine content review, the workflow that uses these tools effectively looks something like this:

Hemingway for mechanical screen (readability, passive voice, complex sentences)
Grammarly for grammar pass
EvvyTools for AI detection with sub-score breakdown
CopyScape for originality check
ProWritingAid for deeper style analysis

If a piece fails one tool but passes the others, you have a specific, named problem. If it fails multiple, the issue is structural and the conversation with the writer is different.

Treating any single one of these tools as a verdict produces the same false-positive problems that have damaged trust in AI detection specifically. Treating them as a coordinated panel of screening signals is how editorial workflows actually use tools without burning honest writers.

A Note on Setup and Time

Adopting all seven tools at once is overkill for most teams. A more practical onboarding is to start with two and add tools as gaps become obvious:

Week one: Hemingway and AI detection. These two together catch most of the mechanical and statistical problems. Run every submission through both. Get used to reading their output side by side.

Month one: add Grammarly and CopyScape. These extend the screen into grammar and originality. The marginal time per piece stays low because the tools mostly run in parallel.

Quarter one: add ProWritingAid and one of the readability tools. These are for deeper analysis on pieces that pass the first screen but still feel off.

Most teams that try to adopt a seven-tool screen on day one bounce off the workflow within a few weeks. A staged adoption that starts with two and grows produces sticky habits.

What Each Tool Will Not Catch

Worth being honest about what this combined toolkit still misses:

Factual errors. None of these tools fact-check. A piece can pass all seven and still contain incorrect claims. For factual screening, you need a separate workflow with sources.

Voice mismatch. A piece can pass every quality screen and still be wrong for your publication's voice. That is editorial judgment, not tool output.

Strategic AI use. A writer who used AI to outline, then wrote the prose by hand, may pass every screen. The detection tools cannot distinguish between "outlined with AI" and "drafted without AI," and the prose tools cannot see the outline at all.

Subject expertise. A piece can be technically clean and substantively shallow. The tools cannot judge whether the writer actually knows the subject.

These are the parts of editing the tools do not replace. The tools clear the mechanical and statistical floor; the substantive work still belongs to the editor.

Adopting the Toolkit Without Process Bloat

A risk with multi-tool screening workflows is that the process becomes the work. Editors spend more time running tools than reading prose. This is usually a symptom of treating tool output as the editorial work rather than as input to it.

The healthy framing: tool output is data. Editorial judgment is the work. Tools that take 10 minutes per piece to run and 2 minutes to interpret are useful. Tools that take 2 minutes to run and 20 minutes to interpret are usually overkill for the marginal value they add.

For most teams, a five-tool screen runs in under 15 minutes per piece if the tools are integrated. That is sustainable for normal volumes. If your screen is taking 45 minutes per piece, something is wrong: either the tools are too slow, the integration is too manual, or you are treating the tools as the answer rather than as the input.

The Detection Question in Context

The right way to think about AI detection in 2026 is as one signal among many that a piece deserves closer human attention. The score is real information. It is not the verdict the marketing copy implies.

For deeper coverage of how detection scores are generated, why they disagree, and how to read them honestly, the EvvyTools guide on detector disagreement walks through the mechanics. Additional writing-quality tools at the EvvyTools tools directory cover the broader pre-publish review surface beyond detection.

A combined toolkit produces better screening than any single tool, and protects you from the documented false-positive risks of leaning on detection alone.

How to Screen Writer Submissions for AI Content Without Burning Honest Writers With False Positives

EvvyTools — Fri, 12 Jun 2026 09:54:56 +0000

If you commission writing, you have probably hit the AI question. Did the writer use AI? How much? Does it matter? And the central problem: you cannot tell from reading, and the detector tools have failure modes that make them dangerous to use as evidence.

This is a workflow problem, not a tool problem. The right answer is not "find a better detector." It is to build a screening process that uses detector scores for what they are actually good at - flagging text that deserves a closer look - and avoids what they are bad at - rendering verdicts on individual writers.

Here is a workflow that gets the value out of detector tools without producing the false-positive damage that has gotten institutions in trouble.

Photo by www.kaboompics.com on Pexels

Step 1: Use Multiple Detectors, Not One

A single detector score is one tool's interpretation of statistical features that other tools weight differently. Running the same submission through two or three independent detectors and comparing results tells you whether the signal is consistent.

If three detectors return 80%+, the AI signal is strong and well-supported across different methodologies. If they return wildly different scores (one at 85%, another at 30%), the text is in the disagreement zone, and any single tool's confidence is misleading.

For a workflow, pick three: one that emphasizes phrase pattern matching, one that emphasizes statistical uniformity, and one that emphasizes hedging. Run every submission through all three before drawing any conclusion.

Step 2: Establish a Per-Writer Baseline

Every writer's natural prose scores somewhere on the detector spectrum. A writer who produces tightly edited, conventional copy may naturally score 50% AI on every piece, including the ones written by hand five years ago. A writer with a looser style may naturally score 20%.

When you onboard a new writer, get them to send two or three samples of writing you know was produced before AI tools were widely available, or work they produced live in front of you in a conversation. Run those through your detectors to establish the writer's baseline.

A submission scoring 85% from a writer whose baseline is 50% is suspicious. A submission scoring 85% from a writer whose baseline is 80% is normal.

This step solves the biggest single source of detector false positives: that some writing styles naturally score high regardless of AI involvement.

Step 3: Read the Sub-Scores, Not the Composite

The single number a detector reports is a weighted average of underlying signals. The signals are what is actually informative. A composite score of 75% might be driven by:

High AI phrase density ("delve into," "in conclusion," "furthermore")
Low sentence-length variation (mechanical uniformity)
Heavy hedging across declarative statements
Generic vocabulary throughout

Each of these is a different problem with a different appropriate response. The first is a vocabulary habit that can be addressed in revision. The second often reflects good editing rather than AI. The third may be a tone choice. The fourth could mean the writer is unfamiliar with the subject matter or that they are using a model at default settings.

Tools that expose the sub-scores let you reason about what the flag actually represents. The AI Content Detector on EvvyTools breaks the underlying signals out individually so you can see which factor is driving the result.

A workflow based on composite scores will produce false positives. A workflow that examines sub-scores produces more nuanced flags.

Step 4: Handle Flags as a Conversation, Not a Verdict

When a piece is flagged, the next step is not rejection. It is a conversation with the writer about the piece.

Useful questions:

Walk me through how you researched this. What sources did you use?
Did you draft any part of this with AI assistance? If so, what parts?
Can you show me your notes or earlier draft?

Most writers will answer these questions honestly. Writers who used heavy AI assistance will usually disclose it when asked directly, especially if your guidelines made it clear that disclosure rather than concealment is what matters. Writers who did the work by hand will be able to walk you through their process in specific detail.

This conversation does what a detector cannot do: it surfaces intent and process, not just statistical patterns.

Photo by Vladislav Anchuk on Pexels

Step 5: Make Your Policy Explicit

If you do not have a written policy about AI use in submissions, write one. Vague policies produce inconsistent application and selective enforcement, both of which damage writer relationships.

A useful policy covers:

Whether AI use is allowed (with bounds), disallowed entirely, or allowed with disclosure
What "AI assistance" means specifically (grammar tools? outlining? full drafts? polishing?)
How submissions will be screened
What happens when a piece is flagged (conversation, not automatic rejection)
What disclosure looks like when AI was used as a tool

The Writers Guild and several publishing trade groups have published model policies that work as starting points. Adapting one to your specific situation is faster than writing from scratch. The Modern Language Association has also published guidance specifically for academic and publishing contexts that handles the disclosure question well.

A clear policy lets writers self-select into work they want to do without ambiguity, and lets you enforce consistently when issues arise. Vague policies are also the source of most disputes: writers reasonably argue that they could not have known what was prohibited if you did not write it down. Writing the policy is cheap, and it pays back every time a question comes up.

Step 6: Track Outcomes Over Time

A screening workflow that catches false positives but never measures them is not actually catching false positives. It is just catching some pieces and assuming the catch was correct.

A simple outcome log helps: every time a piece is flagged, record what happened next. Was the writer able to explain the work? Did revision resolve it? Did the writer leave? Did the piece get published as-is? Over time, the log reveals what the screen is actually doing.

If 95% of flagged pieces resolve with conversation and revision, the screen is working well. If 90% of flagged writers leave, the screen is producing false-positive damage that you have not noticed. The numbers are how you know.

Reference material from the Association for Computational Linguistics on detection accuracy and false-positive rates is useful as a calibration check. If your screen produces flag rates significantly higher than published baselines on similar content, the threshold may need adjusting.

Step 7: Calibrate for Known Failure Modes

A few categories of writer score high on detectors regardless of AI involvement. Build awareness of these into your screening:

Non-native English writers consistently false-flag at higher rates
Formal academic prose scores high because the style is constrained
Tightly edited marketing copy scores high because editing converges on patterns
Translated text scores high because translation introduces uniformity

If your writer pool includes any of these categories, a high score is much less informative than for a casual conversational writer. Adjust your threshold accordingly.

A Tool, Not a Judge

Detection tools fit naturally into a content review workflow as a screening signal. They surface text that deserves a closer human look. They are not built to render verdicts on individual writers, and the documented harm from treating them as such is real.

The right workflow uses them at the front of the funnel to triage attention, not at the back as decision-making evidence. With a per-writer baseline, sub-score awareness, multi-detector confirmation, and a clear policy framework, the tools earn their keep without producing false-positive damage.

For deeper coverage of why detector scores disagree and what they actually measure, the longer EvvyTools guide on detector disagreement walks through the underlying statistics in detail. Additional writing-quality tools at the EvvyTools tools directory pair well with detector workflows for broader pre-publish review.

Quick Workflow Recap

Multiple detectors, not one
Per-writer baseline before drawing conclusions
Sub-scores over composite scores
Conversation when flagged, not verdict
Written policy so expectations are clear
Outcome tracking so you know whether the screen is working
Known false-positive categories get extra context

This workflow produces the same value detector tools were sold to deliver, without the false-positive damage that has cost institutions trust with their own writers.

Why Your Cold Email Open Rate Drops on Mobile Preview Panes

EvvyTools — Thu, 11 Jun 2026 09:53:12 +0000

Most cold-email diagnostics treat the preview pane as if the reader were looking at a full desktop client with three lines of preview text and a wide subject column. The data does not support that assumption anymore. Somewhere north of sixty percent of cold emails are opened (or not opened) on mobile, and the mobile preview pane has different rules than the desktop one.

If your open rate drops when you switch sending volume from a few dozen a day to a few hundred, there is a good chance you are losing the mobile pane.

Photo by Kerde Severin on Pexels

The mobile preview is shorter than you think

On most mobile mail clients, the subject line is truncated around forty to fifty characters, and the preview text is two lines max. The reader makes the open-or-delete decision based on what fits in that window.

This means several things at once:

A subject line that reads well at seventy characters on desktop reads as a half-sentence on mobile.
Anything you put in your first line that is template-y ("Hi NAME, I hope this finds you well") is the entire preview the reader sees.
A subject that depends on the second half to make sense (e.g., "Quick question, do you have a few minutes to talk about ...") loses its meaning when the rest is hidden.

The fix is to write subject lines and openers as if the reader will only see the first forty characters of each. If both parts read as a person sending a real message in that constrained window, you keep the open. If either reads as a template, you lose it.

What the reader actually sees

A useful exercise: open your own draft on a phone in vertical orientation. Look at the preview pane. Read only what is visible without expanding.

Most senders are surprised by how little signal is there. The subject is half a sentence. The preview is the salutation and maybe the first half of the first real sentence. The signature, the value prop, the CTA, the personalization in the third paragraph are all invisible at this stage. The reader sees only what fits.

This is why mobile preview panes systematically reward short, specific subject lines and openers, and systematically punish long template-style ones. The reader on the train does not have the patience to expand the preview to read more. They tap the trash icon or move on.

Three rewrites that recover the mobile open rate

Specific changes that tend to move the number:

Rewrite 1: cut the subject line by half

If your subject is over fifty characters, it is almost certainly losing mobile opens to truncation. Find a way to say the same thing in five or six words. "Quick question about your onboarding email sequence" becomes "Question on your onboarding flow." Same meaning, fits the preview, reads as natural.

Rewrite 2: kill the salutation as the first visible line

"Hi [Name], I hope this finds you well, I wanted to reach out about..." consumes the entire mobile preview before the value prop starts. The fix is to either drop the salutation entirely (acceptable in outreach), put the salutation on its own line with no body content, or write the first sentence so the salutation does not eat the preview.

Better: "Hi [Name] - quick question about your onboarding flow. Saw the post from your head of product last week and had a specific suggestion." The reader sees the first sentence in the preview, knows the email is about a specific artifact, and decides to open.

Rewrite 3: write for the truncated subject

Test the subject by truncating it manually at forty characters. Does it still make sense? "Quick question about your data pipeline" still makes sense at forty characters. "Following up on our conversation last Tuesday at the conference about ..." does not.

If the truncated version reads as gibberish, rewrite until it reads as a complete thought.

The deliverability layer matters too

If your sending domain has a reputation problem, even a perfect mobile-friendly draft will go to spam. Run your sending domain through MXToolbox for SPF, DKIM, DMARC, and blacklist status. Cross-check at Spamhaus. For a one-shot deliverability score on a specific test send, Mail Tester returns a 0-to-10 rating based on what a real spam filter would see.

These three free tools together cover most of the technical deliverability problems senders run into before they ever get to content optimization.

What scoring tools catch that the eye misses

Reading your own subject line on your own phone is the cheapest test you can run. But there are two failure modes the eye misses reliably: spam-trigger density in long subject lines, and unconscious template patterns that slip into your phrasing after you write a few similar emails in a row.

The free cold email scorer by EvvyTools runs the subject line and body against the seven dimensions of outreach diagnostics and returns a score with specific notes on the lines that drag it down. It catches the things that are obvious once pointed out but invisible from inside your own draft: an opener that pattern-matches a template, a subject that does not survive truncation, a CTA that asks for too much.

For background on what each dimension means and how to read the score, the longer guide How to Diagnose a Failing Cold Email Before You Hit Send walks through all seven. Other writing and copywriting utilities live in the tools directory on the same site.

A small habit that compounds

The senders who land in the inbox consistently treat the preview pane as the unit of work, not the full email. They write the subject line to survive truncation, they write the first line to deliver value in the preview window, and they trust that the rest of the body can do the real selling once the open has been earned.

Read your next draft on your phone before sending. If the preview pane does not earn the open by itself, the rest of the email does not get the chance.

That is the entire game for the mobile reader. Most senders never run this check, which is why most senders see their open rate drop when their list moves from desktop to mobile. The fix is fifteen seconds long and you can run it on every draft.

Run it before the next batch. The numbers move.

A few notes on screen sizes

The truncation thresholds I gave above (roughly forty to fifty characters for subjects, two lines for preview text) are averages across the major mobile mail clients. Specific clients vary. The Gmail mobile app shows a few more characters than the Apple Mail app on iPhone in portrait. Outlook Mobile shows fewer than either. Reading at landscape orientation adds about a third more characters, but most people on the move are reading in portrait.

The practical implication: design for the most restrictive case. If your subject and opener earn the open on a portrait iPhone showing forty characters, they will earn it on every other client too. If they only work at a hundred characters, you are losing the readers who matter most (the busy ones on the move).

The dark-mode wrinkle

One small thing worth knowing: dark-mode rendering in some clients can change how preview text contrasts with the background. If your draft includes any HTML formatting (bold, color, background fills) that depends on light-mode assumptions, the preview can render as low-contrast gibberish in dark mode.

Plain-text cold emails avoid this problem entirely, which is one more reason to send cold outreach as plain text rather than as styled HTML. The plain-text version reads the same in both modes, on every client, at every screen size. One fewer thing to debug.

Worth fifteen seconds of attention before sending any batch at volume.

How to A/B Test Cold Email Subject Lines Without Burning Your List

EvvyTools — Thu, 11 Jun 2026 09:51:08 +0000

You want to know which subject line gets more opens. Easy in theory: write two, send half your list each, compare. In practice, this is where most cold email tests go off the rails. You run the test at the wrong sample size, you do not control for the day of the week, you keep both losing variants in rotation, and three weeks later your list is half-burnt and you still have no signal.

A clean A/B test on subject lines does not require a six-figure tool. It requires a small, disciplined process you can run with whatever sending stack you already have.

Photo by Min An on Pexels

Step 1: Pick two genuinely different subjects

The first mistake is testing variants that are too similar. "Quick question about your onboarding" versus "Question about your onboarding flow" is not an A/B test. It is two near-identical subject lines, and the variance you will see is noise, not signal.

A real test changes the pattern, not just the wording. Test a specific question against a visible-artifact reference. Test a numerical specific against a forward-able routing question. Test five words against seven words and a number. The bigger the structural difference between A and B, the more useful the result.

If you cannot articulate one sentence on why A might perform differently from B, you have not designed a test, you have made a typo.

Step 2: Split the test list, not the live list

Hold back a slice of your list specifically for testing. A hundred recipients is the minimum for any signal; two hundred is more reliable. Split that slice fifty-fifty into A and B groups.

Do not run the test on the whole list. If A wins, you have burnt half your prospects on the loser. If neither wins clearly, you have burnt all of them on undifferentiated outreach. A dedicated test slice means you keep the rest of the list available for whichever variant wins.

The most useful tool for this kind of list segmentation is whatever sending platform you already use. Almost all of them support segments. If yours does not, segment by exporting and uploading two lists; the manual version is fine for a hundred-recipient test.

Step 3: Send on the same day at the same hour

Day-of-week and hour-of-day have a measurable effect on open rates. A test that sends A on Tuesday morning and B on Wednesday afternoon is comparing four variables at once: subject, day, hour, and recipient pool. You will get a number, but you will not learn anything about subject lines from it.

The fix is to send both variants in the same fifteen-minute window. Schedule them in advance. Verify they actually went out at the same time before you start counting opens.

Step 4: Wait at least 48 hours before counting opens

Open rates do not stabilize for at least a day, sometimes two. People open emails on commute, at lunch, after dinner, the morning after. A snapshot at four hours is misleading; a snapshot at twenty-four hours is closer but still moving.

Wait forty-eight hours minimum. Seventy-two if the test is not urgent. Count the opens at that point and lock the numbers in. Re-checking the next week will surface a few more late opens, but the test is over.

If your sending tool gives you both open rate and reply rate, count both. Open rate tells you which subject earned the click; reply rate tells you which subject set up the body for conversion. They do not always agree.

Step 5: Define the win threshold before you read the data

This is the step most senders skip. They run the test, look at the numbers, and rationalize whichever variant won. "A had a 28 percent open rate and B had 26 percent, so A wins." With a hundred-recipient test slice, that two-point gap is noise, not signal. Two more recipients opening one variant or the other would flip the result.

A useful rule of thumb: at a sample size of one hundred per variant, you need at least a five-percentage-point gap to call it. At two hundred per variant, four points. At five hundred, two points. Below those thresholds, declare the test inconclusive and run it again with a new pair.

You can compute the actual significance with any free statistical A/B test calculator if you want the formal number. For most cold email tests, the rule of thumb is enough.

Step 6: Ship the winner, archive the loser

Once you have a winning variant, send it to the rest of the list. Archive the loser; do not keep it in rotation, do not save it for next time. Cold email testing is incremental: each round produces a new champion that you then test the next variant against.

The discipline is to never test more than one thing at a time. Subject lines first, then openers, then CTAs, then length. If you change three things at once and the numbers move, you cannot tell which change did it.

Step 7: Re-test the winner against fresh ideas

The subject line that won this month will not necessarily win next quarter. Lists fatigue. Inbox providers adjust their filters. The reader's pattern recognition for "this is a sales template" updates.

Plan to re-test your champion against new variants every couple of months. The cadence keeps the open rate from drifting downward as your subject style gets predictable to the recipient.

Tools that help the workflow

You can run this entire process with a spreadsheet, two saved-template versions, and your sending tool's existing segmentation. No special software required. But there are two specific tools worth knowing about:

For technical deliverability checks before sending the test (so a domain reputation issue does not confound the result), MXToolbox and Spamhaus are the standard free tools. Run your sending domain through both before any volume test.

For scoring each subject line variant against known cold-email patterns before you send them, the Cold Email & Outreach Scorer at EvvyTools rates a subject and body across seven dimensions and points at the lines that are likely to fail. Useful as a pre-test filter: if both variants score below 60, the test is going to be a comparison of two losers, and you might want to write a third candidate first.

For a longer read on what those seven dimensions are, the guide How to Diagnose a Failing Cold Email Before You Hit Send walks through all of them. Other writing utilities live in the EvvyTools tools directory.

What not to test

A few things that look like A/B test material but produce noisy or useless data:

Sender name capitalization (one capital, all lowercase, etc.). The effect size is too small for any reasonable sample to detect.
Emoji versus no emoji. The effect is so different across audiences that one test result does not generalize.
Subject line punctuation. Same problem; effect size is too small.

Save the test budget for changes that actually move the number: subject line structure, opener pattern, CTA specificity, value-prop framing. Those are where the wins live.

Run small tests often

The senders who improve their open rate continuously are not the ones who run one big test a year. They run a small test every two weeks, on a hundred-recipient slice, with one variable changed, and they ship the winner to the next batch.

The math compounds. A small, real improvement every two weeks, sustained for a year, is the difference between a campaign that earns the open and one that goes to the trash. Most senders never set up the discipline. The ones who do separate themselves quickly.

Build the test loop once. Run it forever. The numbers move.

5 Free Physics Calculators That Every Engineering Student Should Bookmark

EvvyTools — Wed, 10 Jun 2026 09:54:52 +0000

Engineering coursework runs on quick calculation tools. Slide rules and TI-89s are nostalgia at this point. The actual day-to-day reality is keeping a few browser tabs open with calculators that handle units, give you the answer in three seconds, and let you sanity-check the homework you just spent an hour deriving by hand.

This is a short list of free, no-account-required calculators that hold up. None of them require a subscription. None of them require you to dodge ads or sign-up walls. They are simple, fast, and they do not lie about their answers.

Photo by cottonbro studio on Pexels

What "useful" actually means in this context

Most physics calculator sites either bury the input fields under twelve banner ads or quietly drop unit conversions that change the answer by orders of magnitude. The bar for this list is:

Loads in under two seconds on a normal connection.
Lets you type in units freely (g vs kg, ml vs cm3, etc.) and handles the conversion correctly.
Shows the formula it used, not just the answer, so you can spot if it interpreted your input wrong.
Does not require account creation, payment, or a download.

That filter cuts most of the search results for "free physics calculator" immediately. The five below all clear the bar.

1. Density Calculator (EvvyTools)

A density-mass-volume solver with a built-in database of more than 100 materials. You can solve for any of the three given the other two, and the unit picker accepts the mix of metric and imperial you actually see in lab work.

The piece that elevates it above a plain m = p * V solver is the mystery-material identifier. You measure a sample, type the density, and it returns a ranked list of which materials in the database match within a tunable tolerance. That is the feature you use when you have an unknown sample in front of you and want to figure out what it is.

Live at https://evvytools.com/tools/math-science/density-calculator/. There is also a longer write-up of how to actually use it on unknown materials if you want the full procedure.

2. WolframAlpha (Wolfram Research)

The general-purpose computational engine. For physics, it shines on the symbolic side: enter a formula with named variables and it will solve, differentiate, plot, and convert units in one go. Good for derivations and quick sanity checks of homework problems where you want the closed-form answer.

The free tier covers most undergraduate work. The paid tier unlocks step-by-step solutions, which is genuinely useful when you are stuck on a derivation. Hosted by Wolfram Research.

3. Engineering Toolbox

Not a single calculator but a sprawling reference and calculator collection. Material properties (density, thermal conductivity, modulus), fluid mechanics solvers, HVAC calculations, and a few hundred other tools. Some are dated in styling but the numbers are sound and the formulas are shown.

Particularly useful for material property lookups when you are checking a measurement against published values. Live at https://www.engineeringtoolbox.com/.

4. NIST Chemistry WebBook

For chemistry-adjacent physics work (thermochemistry, phase changes, gas properties), the NIST Chemistry WebBook is the authoritative reference. It is not a calculator in the polished-UI sense, but it is the source of the numbers other calculators pull from.

If you are checking a heat-capacity value, a vapor pressure, or an enthalpy of formation, this is where you go. It is government-funded, no account required, and the data has been peer-reviewed for decades.

5. CalcTool (Various)

A clean general-purpose calculator collection with physics, electronics, mechanics, and finance categories. The physics tools cover the textbook formulas (kinematics, work-energy, thermodynamics, electromagnetism) with reasonable input validation.

Worth bookmarking specifically for the projectile motion, simple harmonic motion, and electric circuit solvers, which are the calculations that come up most often in introductory courses. Live at https://www.calctool.org/.

How to use a list like this without becoming dependent

A calculator is a checking tool, not a learning tool. The pattern that works for engineering students is:

Work the problem by hand first, deriving the formula and plugging in numbers.
Use a calculator to check the final number.
If they disagree, find your error. If they agree, you are done.

Skipping step 1 hollows out your intuition. Skipping step 2 leaves silly arithmetic mistakes in your final answer. Doing both is the path through.

For material identification problems specifically (the ones where "what is this thing?" matters more than "what is the formula?"), the EvvyTools density tool is the one to keep open. The built-in database means you do not have to look up reference values separately, and the mystery-material identifier saves a step on every unknown sample.

For the underlying procedure on how to take a clean density measurement at home with cheap equipment, see the full guide from EvvyTools. It covers the measurement gotchas (meniscus reading, air bubbles, plated samples) that are the difference between a useful answer and a confusing one.

The longer point

The reason this list is short is that most physics calculators are noise. Search results are clogged with sites that load slowly, lock the answer behind a sign-up, or fail silently on unit conversions. Five good tools, bookmarked and used carefully, are worth more than fifty marginal ones.

If you are not sure where to start, the EvvyTools tools directory has the broader math and science calculator collection. The density tool is what got me writing this list. The rest are the ones I kept reaching for in the months since.

What I left off the list, and why

A few tools that probably belong on a longer version of this list, and the reason I left them off here:

Symbolab. Strong on calculus derivations and integration walkthroughs. Useful for math homework. Left off because the physics-specific calculators are thin compared to Wolfram, and many features require an account these days.

Desmos. The graphing calculator is excellent for visualizing equations and exploring function behavior. Left off because it is a graphing tool more than a calculator. If you have not used it, the Desmos graphing calculator is worth a separate bookmark; for plotting curves and exploring function shape it is unmatched. But for "calculate this physics quantity" it is the wrong tool.

Octave (online). Free MATLAB-clone available in a browser. Useful for numerical methods coursework, especially linear algebra and matrix-heavy problems. Left off because it is closer to a programming environment than a calculator. If your coursework involves matrices larger than 3x3, learn it.

Python with NumPy in a notebook. Same idea as Octave but with a much bigger ecosystem. Most engineering programs now teach this in introductory courses. Worth learning early; it scales from "homework calculator" to "real engineering tool" without changing notation.

The reason these are in a footnote rather than the main list is that they are not "free physics calculators" in the bookmark-and-go sense. They are tools that require some setup or some learning curve before they pay off. Worth the investment for any engineering student, but they belong in a different list.

How to evaluate a new physics calculator

If you find one that is not in this list and want to know whether it is worth adding to your bookmarks, the same filter still applies:

Does it load fast and work without an account?
Does it handle units explicitly, and can you mix metric and imperial in the input?
Does it show the formula it used, or at least let you click through to see the math?
Is there a clear note about which physical principle it is solving (so you do not accidentally use a non-relativistic kinematics solver on a problem that needs special relativity)?
Does it work on mobile, because half the time you check homework on your phone?

Anything that passes all five is a candidate for the bookmark bar. Most that you find will fail at least one. That is fine; it just means they are not on the short list.

The five above all pass. The EvvyTools density tool in particular is the one I use most often because the material-lookup feature does the part of the work I cannot do in my head, which is comparing a measured value against a hundred reference materials. The arithmetic is trivial; the lookup is the actual feature.

If you have a favorite that I left off, I am genuinely curious. The shape of the list of "tools engineering students actually use" probably shifts every few years as new tools appear and old ones quietly go behind paywalls. The Engineering Toolbox calculator index is one place to find calculators by category if you are hunting for a tool that does something specific.

Why Density Is the First Physical Property to Measure When Identifying an Unknown Material

EvvyTools — Wed, 10 Jun 2026 09:54:14 +0000

Every workshop, lab, and maker space runs into this problem eventually. You have a chunk of something. It might be aluminum. It might be a zinc alloy that just looks like aluminum. It might be one of the cheaper white metals that fail under load when you assume they are aluminum.

You could send it for spectroscopy. You could try a spark test. You could grind off a corner and hit it with acid. But before any of that, there is a much cheaper test that knocks out 80 percent of the candidates: measure the density.

This post is about why density should be your first measurement, what it actually rules out, and where the limits are.

Photo by Tara Winstead on Pexels

Density is a fixed property of the material itself

Mass changes if you cut the sample in half. Volume changes if you machine away surface material. But mass divided by volume does not change, as long as the sample is solid and uncontaminated. That ratio depends only on what the material is made of and how its atoms pack together. It does not care about shape, size, or surface finish.

That is what makes it a useful identification tool. Most other simple tests (color, magnetism, hardness) give you partial information that often overlaps between candidates. Density gives you a single number that places the material on a one-dimensional axis.

For the underlying definition and how the unit is derived, the Wikipedia entry on density covers the basics. The unit you actually use day to day is grams per cubic centimeter, which is numerically the same as grams per milliliter.

How much it actually narrows the field

Common engineering materials span a wide density range:

Polyethylene: 0.91 g/cm3
Most engineering plastics: 1.0 to 1.4 g/cm3
Aluminum alloys: 2.6 to 2.9 g/cm3
Titanium: 4.5 g/cm3
Zinc: 7.14 g/cm3
Iron and most steels: 7.7 to 8.0 g/cm3
Brass: 8.4 to 8.7 g/cm3
Copper: 8.96 g/cm3
Silver: 10.49 g/cm3
Lead: 11.34 g/cm3
Gold: 19.32 g/cm3

A measured density of 2.7 g/cm3 rules out everything except aluminum and a couple of low-density alloys. A reading of 7.85 g/cm3 puts you in steel territory and rules out copper, brass, and the heavy metals entirely.

That is a huge win for almost no time invested. A spectrometer rental might cost a few hundred dollars and require shipping the sample. A scale and a cylinder cost forty bucks total and give you the answer in five minutes.

What it actually feels like to do

The simplest method is water displacement. Weigh the dry sample. Put water in a graduated cylinder, record the level. Drop the sample in, record the new level. Volume of sample is the change in water level. Density is mass divided by volume.

The math is trivial but the rounding errors are not. A small mistake on a small sample swings the answer between candidates. The free density calculator by EvvyTools handles the arithmetic, the unit conversions, and the lookup against a database of common materials all at once. Type in mass and volume, get the density and a ranked list of materials whose published density matches.

The full procedure (and a discussion of where measurement noise becomes a problem) is in the longer guide on identifying unknown materials by density.

Why this matters for engineering work specifically

If you are sourcing materials, salvaging hardware, or recycling production scrap, density catches a few specific failure modes that are easy to miss:

Substitution. A supplier sends you bars labelled as aluminum 6061. The density should be 2.70. If the bars come in at 2.85, you have an aluminum-zinc alloy or contaminated metal, not 6061. The density test runs in five minutes; you can flag the shipment before anyone machines a part out of it.

Plating reveal. A "solid brass" fitting that reads at 8.96 is not solid brass; it is copper. The plating hid the difference. Density on the bulk catches plated-on-base-metal substitution that visual inspection misses.

Porosity in castings. A cast iron part with internal porosity has a measurably lower density than the same alloy cast properly. If you are inspecting incoming castings and a few read 7.5 instead of 7.85, those are the ones to X-ray before they go into production.

Composite identification. Filled plastics (glass-filled nylon, carbon-filled PEEK) have densities that shift predictably with the filler fraction. A density reading often tells you the fill percentage without a tear-down.

Where it stops being enough

Density gets you a long way, but it is not a unique fingerprint:

Alloy families overlap. Brass, bronze, and many copper alloys sit close enough together that density alone cannot pick the winner. Most stainless steels sit at 7.9 to 8.0 g/cm3, which overlaps mild steel. You need a second test (magnetism, conductivity, hardness) to separate them.

Plating is the obvious trap. Density measures the bulk material. If the sample is plated, you read the bulk and miss the surface. Strip a corner first if you suspect plating.

Voids and inclusions. Anything with internal porosity, voids, or trapped air reads lower than it should. This is sometimes what you want to find (porosity QC), but it can fool a one-shot identification if you are unaware.

For deeper identification, organizations like NIST and the ASTM standards library publish the reference data that lets you compare your measurement against documented values for specific alloy grades.

The workflow most people end up with

After running this enough times, the routine settles into something like:

Eyeball: what could it plausibly be? Make a short list.
Magnet: stick to it strongly, weakly, or not at all? Cuts the list further.
Density measurement: which candidates on the remaining list match your reading?
If two or more match, pick a tiebreaker test (conductivity for metals, melt or burn test for plastics).
Done.

You only reach step 4 occasionally. Step 3 is the workhorse, and it costs you a five-minute setup with hardware you already own (or can buy for under fifty bucks).

That is the case for density as the first physical property to measure. It is cheap, fast, and rules out more candidates per minute spent than any other simple test. The free density calculator by EvvyTools is what most people end up using for the arithmetic, because doing the unit conversions by hand on every reading is a path to mistakes.

For the full step-by-step guide including measurement tolerances, common errors, and three worked examples, see the EvvyTools blog post on identifying unknown materials. It covers the measurement gotchas that turn a clean 5-minute test into a wrong answer if you skip them.

A note for engineers who deal with materials professionally

For incoming inspection work, the density measurement is not a replacement for a certificate of conformance or for spectroscopic verification on critical material. It is a screening tool. The use case is: a shipment arrives, you want to check that what is in the box matches the paperwork before anyone machines a part from it, and you do not have a spectrometer in the building.

In that scenario, a density measurement on one sample per heat or per lot catches gross substitution (wrong family of alloy entirely). It does not catch grade substitution within a family (6061 vs 6063 aluminum, for instance, both at ~2.70 g/cm3). For those cases the only reliable check is composition analysis, and you should specify that as part of the incoming inspection plan if grade matters to the application.

The other place where density is genuinely valuable is in salvage and reverse-engineering work. You have a part of unknown provenance, you want to specify a replacement, and density narrows the possibility space enough that a follow-up test (hardness, conductivity, magnetic response) confirms the choice. Composition does not have to be exact; you just have to be close enough that the mechanical and thermal properties are comparable.

For project work where the requirement is "make this part out of something that performs like the original," the density-plus-magnetism-plus-hardness triple usually gets you close enough. The materials properties data on Matweb is the public reference most engineers actually use to cross-check candidate materials once density has narrowed the list.

So: density first, because it is cheap and rules out the most candidates per minute. The rest of the tests are for after density has done its job. That is the principle. The tools are just convenience.