DEV Community: QuikTools Hub

How Many Days Until Christmas 2026? (Exact Countdown)

Umar — Tue, 27 Oct 2026 16:37:10 +0000

Published: 26 October 2026 · by QuikTools Hub

There are 60 days until Christmas Day, Friday 25 December 2026, counted from Monday 26 October 2026. That is 8 weeks and 4 days, or 44 business days, or 1 month and 29 days — and the Days Until Calculator on this site gives the live count for any date you pick.

The countdown in weeks, business days and months

Sixty days breaks down three ways, depending on what you are counting. In weeks it is 8 weeks and 4 days. In business days — Monday to Friday only — it is 44 weekdays, counted from 27 October up to and including Christmas Day itself. In calendar months it is 1 month and 29 days: one full month from 26 October to 26 November, plus 29 days to 25 December. Months here are calendar months, not 30-day blocks, which is why the figure does not divide neatly.

The 60 days add up in three parts. October has 5 days left after today: 27, 28, 29, 30 and 31 October. November contributes all 30 of its days. December contributes 25, up to and including Christmas Day itself. Five plus 30 plus 25 is 60 — the same total the calculator returns, and a useful check if you ever want to verify a countdown by hand.

Days until the dates around Christmas

Christmas Eve comes first: Thursday 24 December 2026 is 59 days away, or 43 business days. New Year's Day follows a week later — Friday 1 January 2027 is 67 days away, or 49 business days, which makes it 9 weeks and 4 days. Between now and Christmas Day there are 16 weekend days to get through: 8 Saturdays and 8 Sundays. Christmas Day itself is a Friday, so it is not one of them.

The gap between the dates is the part people usually misjudge. Christmas Eve is only 1 day before Christmas Day, so anything left for the 24th has almost no slack — travel, food shopping and last deliveries all land on the same day. New Year's Day is 7 days after Christmas, which leaves a full week between the two holidays for visits, rest and preparation. If you are counting for a deadline rather than a holiday, the business-day figures matter more: 43 weekdays to Christmas Eve and 49 to New Year's Day, before any public holidays you need to remove by hand.

How the count works

Both dates are converted to UTC midnights before subtraction. That matters because local clock times shift with daylight saving: in spring a day can be 23 hours long and in autumn 25, and naive timestamp subtraction would occasionally report a day too many or too few. Business days count Monday to Friday only, and public holidays are not subtracted, since holidays vary by country — if your deadline spans one, remove that day by hand.

Count down to your own date

Christmas is the classic countdown, but the same arithmetic works for anything: a birthday, a wedding, an exam, a product launch, or a date in the past you want to measure. Enter any date and the calculator returns the days, weeks, business days and months, with the years-months-days figure for anniversaries. It runs entirely in your browser, with nothing uploaded and no sign-up.

The same tool runs backwards for dates in the past. A wedding anniversary, a work anniversary, or the day a business opened can all be measured the same way — in days, weeks and months since the date, with the years-months-days figure reading most naturally for long spans. Ten years, 3 months and 12 days means more to most people than a raw day count, but the raw count is there too when a milestone is counted in days.

One more use for the count: working backwards from a deadline. If cards must be posted 5 business days before Christmas, the last posting day is 18 December — 44 business days minus 5 leaves 39 weekdays from today to post. The calculator does not set the deadline for you, but it makes the arithmetic behind one exact.

Try it: open the Days Until Calculator, enter 25 December 2026, and check the countdown — then change the date to your own event and watch every figure update.

Frequently asked questions

How many days are there until Christmas 2026?

Counted from Monday 26 October 2026, there are 60 days until Christmas Day, Friday 25 December 2026. For the live count on any other day, use the Days Until Calculator.

How many weekends are there until Christmas?

16 weekend days fall between now and Christmas Day: 8 Saturdays and 8 Sundays, counted from 26 October 2026 up to Christmas Day. Christmas Day itself is a Friday, so it is not one of them.

What day of the week is Christmas Day 2026?

Friday 25 December 2026. Christmas Eve is Thursday 24 December, and New Year's Day 2027 is also a Friday.

How many business days are left until Christmas?

44 weekdays, Monday to Friday, counted from 26 October 2026 up to and including 25 December 2026. Public holidays are not subtracted — if your deadline spans one, remove it by hand.

How do I count down to my own event?

Open the Days Until Calculator, enter any date, and it returns the days, weeks, business days and months — for future dates and past anniversaries alike.

Random Number Generator With No Repeats: Fair Draws, Raffles and Lottery Lines

Umar — Tue, 27 Oct 2026 16:35:33 +0000

Published: 23 October 2026 · by QuikTools Hub

Most random draws have a quiet flaw for everyday use: nothing stops the same number coming up twice. Draw six numbers between 1 and 49 one at a time and there is nothing unusual about getting 17 twice - each draw is independent, and the generator has no memory of what it picked a second ago. For a lottery line, a raffle or a list of winners, a duplicate is useless or worse. The fix is a setting called No repeats, and on the Random Number Generator it does exactly what the name says: every number in your draw is unique. This article explains what changes under the hood, then works through the three jobs people use it for most.

What "no repeats" changes

Statisticians call the two approaches sampling with replacement and sampling without replacement. A plain generator samples with replacement: the full range is available again for every pick, so repeats are allowed. Ticking No repeats switches to sampling without replacement - each number, once drawn, is removed from the pool, the way lottery balls are not put back in the machine between picks. The generator here does it literally: it draws, checks the result against the numbers already in your list, and discards any duplicate and draws again until the list is full of unique values.

How likely is a repeat if you leave the setting off? Much likelier than intuition suggests. This is the same effect behind the birthday paradox, where 23 strangers in a room are enough for a better-than-even chance that two share a birthday. The general rule is that collisions arrive around the square root of the range: drawing from 1 to 100, a repeat is already more likely than not after roughly a dozen draws. So a teacher picking students at random, or a giveaway drawing ten winners from a hundred entries, will hit a duplicate quickly - and whoever runs the draw then has to improvise a rule on the spot, which is exactly what a fair draw should avoid.

One hard limit: your range must contain at least as many numbers as your Count. A 1-49 range can supply at most 49 unique numbers. Ask for more and the generator stops with an error rather than quietly repeating one.

Running a lottery line

For a classic 6/49 game, set Min to 1 and Max to 49, set Count to 6, and tick No repeats. Generate, and you get six different numbers - a valid line, in the order drawn. Use the Copy button if you want to keep it. If your game has a separate bonus or power ball drawn from its own pool, that one is a separate draw with its own range, and it is allowed to match one of the main numbers, so generate it separately with No repeats off.

One thing worth saying plainly, because a lot of lottery content online implies otherwise: a no-repeats line is not luckier than any other line. Every set of six unique numbers from 1 to 49 has exactly the same chance of being drawn. The setting does not improve your odds; it simply produces a line that follows the rules of the game, with no number wasted by appearing twice. Treat it as a way to pick numbers without agonising over birthdays and anniversaries, not as a strategy.

Running a raffle or giveaway fairly

Fairness in a small draw is mostly procedure, and most of it happens before anyone clicks Generate. Number your entries first: if 137 people entered, they are entries 1 to 137, in a list anyone can inspect. Announce the rules before the draw - how many winners, and what happens if a drawn entry turns out to be ineligible - then draw with the Random Number Generator set to your entry range, Count set to the number of winners, and No repeats ticked, so the same entry cannot win twice. Do it on a shared screen or a livestream if the audience is remote; a draw people can watch is a draw people trust.

Publish the numbered entry list and the rules before drawing, not after.
Draw all winners in one go with No repeats on, and read them out in the order drawn.
If a drawn entry is ineligible - a duplicate signup, a staff member, someone who already won - do not quietly re-run the whole draw. Draw one replacement entry publicly, state why, and keep the rest of the original winners standing.
Post the winning numbers where entrants can see them afterwards, along with the entry list they were drawn from.

The replacement rule matters more than it looks. Re-rolling everything because one entry was invalid throws away valid winners and looks, from the outside, like drawing until you get a result you prefer. A single public replacement draw keeps the process defensible.

Classroom and team picks

The same mechanics cover most classroom and office jobs. Number the class list or team list, then draw. For a presentation order, draw as many numbers as there are presenters with No repeats on: the list you get is the running order. For splitting people into groups, draw the full list of numbers in one go and deal them out in order - first chunk to group one, next chunk to group two - which gives every group a genuinely mixed set rather than the front row plus friends.

Secret-santa style assignments need one extra idea. The generator draws numbers; the assignment is a mapping you make from those numbers. Draw one number per person, in list order, with No repeats on so nobody is assigned twice. If someone draws their own number, you need a rule decided in advance - the usual one is to swap that person with the next drawn entry, or redraw only that pairing while everyone watches. What does not work is quietly regenerating the whole thing until it looks right, for the same reason as the raffle: the process stops being checkable.

How the randomness works here

The integers on this page come from crypto.getRandomValues, the cryptographic random source built into your browser - the same source the Password Generator uses, put to a different job. That distinction matters. The ordinary Math.random function in JavaScript is built for speed and convenience, and its output is predictable enough that it should never decide anything of value. A cryptographic source is designed so its output cannot be guessed from previous output. The generator also maps those random values into your range carefully, rejecting the rare values that would skew the odds, so every number in the range is equally likely - no number is favoured because of how the arithmetic falls out.

It also all happens on your device. The ranges, entry counts and results are never uploaded, stored or placed in the page URL, which is a reasonable property for a draw whose entry list might be a class roster or a customer list.

Limits and honesty

This is a good tool for games, classrooms, social giveaways, team draws and picking a lottery line for fun. It is not a certified draw. Regulated lotteries, licensed prize promotions and anything with legal or significant financial stakes need an audited, certified random process with a paper trail - and, as the tool page itself warns, browser randomness should not be the sole source for regulated or high-stakes draws. If money or lawyers are involved, that warning is the whole answer.

Within those limits, the draw is as fair as the procedure around it. The generator lives among the Productivity Tools - set your range, set your count, tick No repeats, and publish your rules before you draw.

Which Online Tools Upload Your Files - and Which Never Do

Umar — Tue, 27 Oct 2026 16:34:11 +0000

Published: 22 October 2026 · by QuikTools Hub

Almost every free tool site works the same way: you choose a file, it uploads to a server somewhere, software on that server does the job, and you download the result. The site you are reading is built the other way round - every tool here runs in your browser, and your files never travel anywhere. Both designs can be legitimate, but they are not equally private, and the difference matters most for exactly the files people most often push through these tools: a scan of an ID card, a payslip, a signed contract, a personal photo that needs shrinking for a form. This article explains how each design works, and gives you two tests to run on any tool site - including this one - so you never have to take a privacy claim on trust.

How most online tools actually work

On a typical converter or compressor site, the round trip looks like this: your browser sends the file to the company's server; the server runs the conversion with programs like ImageMagick or FFmpeg; the converted file is stored, briefly or not so briefly; and your browser downloads it back. Storage windows vary - some services delete files within an hour or two and say so plainly, others are vaguer, and a file passing through a server can also end up in logs, backups and abuse-scanning systems along the way. None of this requires the operator to be dishonest or careless; it is simply what server-side processing means. Your document existed, complete and readable, on a computer you do not control.

There are fair reasons to build tools that way. Servers are powerful, they handle enormous files and exotic formats, and the processing software already exists there. If you are converting a two-hour video or a 400-page scanned archive, a server is probably the right tool. The trade-off only becomes a problem when the file is sensitive and the job is small - which, for everyday tools, is most of the time.

The other way: the tool runs in your browser

Modern browsers are quietly capable of the same work. JavaScript can read a file you select into the tab's memory, draw images onto a canvas and re-encode them at a lower quality, generate QR codes, hash text with the Web Crypto API, and produce a download - all without a single request carrying your data anywhere. Our Image Compressor works exactly like this: the image is drawn to a canvas on your device and re-encoded there, and the "download" is generated locally too. When a tool is built this way, there is no upload to intercept, retain or breach, because the file never leaves the machine it started on. It also works on a plane, which is a decent clue, as the next section shows.

Check for yourself: two tests

The network-tab test. Open the tool, press F12 (or right-click, Inspect) and go to the Network tab, then clear it. Now use the tool with a real file and watch what leaves your machine. A server-side tool shows a large request - usually a POST roughly the size of your file. A browser-based tool shows no such request: the file is never transmitted, though you will still see ordinary page and advertising requests.
The offline test. Load the tool page fully, then disconnect completely - airplane mode, Wi-Fi off, cable out. Now use the tool. If it still does the whole job and produces your download, the processing is happening on your device by definition; there was nowhere for the file to go.

Do not take our word for any of this. Run both tests on this site: load the Image Compressor, go offline, and compress a photo. It works, because there is no server in the loop to miss.

The honest part: free sites still run ads and analytics

Here is the part most "privacy-first" marketing leaves out, so we will not. A free site has to pay for itself somehow, and this one shows advertising; ad networks load their own scripts, and privacy-light analytics may count that a page was viewed. That is a real, if modest, flow of information about your visit - and it is completely separate from your tool input. The claim we make is narrow and specific: text you type, files you open, passwords you generate are processed in your tab and are not uploaded, stored, or placed in the page URL by these tools. You can read the detail on our privacy page, and the distinction is worth applying everywhere: "we may count page views" and "we receive your documents" are different claims, and only one of them should worry you about a payslip.

When a server tool is fine - and when to avoid one

Fine on a server: public or non-sensitive files, huge media your device would struggle with, obscure formats that genuinely need specialist server software.
Keep in the browser: ID and passport scans, payslips and bank documents, signed contracts, client or patient files under a duty of confidentiality, personal photos, and anything you would not email to a stranger.
Either way: a shared or public computer is its own risk - downloads and browser history stay on the machine after you leave, whatever the tool did.

The underlying point is simple: privacy claims are cheap, but architecture is verifiable. A tool either sends your file somewhere or it does not, and with a network tab and an airplane mode you can tell which in under a minute. Prefer tools that never ask for the upload in the first place - and check, the way you just learned to, rather than trusting. All 41 tools on this site are built that way; start from the homepage and run the offline test on any of them.

How Strong Is Your Password? Entropy in Plain English

Umar — Tue, 27 Oct 2026 16:32:10 +0000

Published: 22 October 2026 · by QuikTools Hub

Every password strength checker speaks the same language: bits of entropy. It sounds like physics, but entropy is really just a way of counting how many guesses an attacker would need. A password with 50 bits of entropy has about 2 to the power 50 possible combinations - roughly 1.1 quadrillion. Double the bits and you do not double the work; you square it. This guide explains where the bits come from, shows the numbers for real generated passwords, and explains why the boring advice - make it longer - beats every clever trick. If you want to check a password of your own as you read, our Password Strength Estimator gives an instant estimate, locally in your browser.

The only formula you need

Entropy (bits) = password length x log2(size of the character pool)

The pool is the set of characters each position is drawn from: 26 for lowercase letters only, more when you add capitals, digits and symbols. Log2 just converts the pool size into bits - a pool of 64 is exactly 6 bits per character, because 2 to the power 6 is 64. Every character you add multiplies the number of possible passwords by the pool size, which is why length is so powerful: with a pool of 73, each extra character makes the password 73 times harder to brute-force.

A concrete reference point: our Password Generator draws from a pool of 73 characters. That is a slightly unusual number, and it is deliberate - the generator leaves out look-alike characters (0 and O, 1 and l) so a password can be read aloud or typed from a screen without mistakes. Its pools are 23 lowercase, 24 uppercase, 8 digits and 18 symbols, so each character carries log2(73), about 6.19 bits. That gives:

8 characters - about 50 bits of entropy
12 characters - about 74 bits of entropy
16 characters - about 99 bits of entropy
20 characters - about 124 bits of entropy
For comparison: an 8-character password using lowercase letters only (pool of 26, about 4.7 bits each) has only about 38 bits - the bigger pool is worth roughly 4,000 times more combinations at the same length.

What those bits mean in practice

Take the worst realistic case: an attacker has stolen a password database with fast, weak hashing and can test 10 billion guesses per second on specialised hardware. Against a truly random 50-bit password they must try about 1.1 quadrillion combinations - a day or so of non-stop guessing, less on average. That is uncomfortably close, which is why 50 bits should be treated as a floor, not a target. At 74 bits the same attacker faces about 60,000 years of work. At 99 bits the answer runs to more than a hundred times the age of the universe. The jump from "a day" to "forever" is only a handful of extra characters.

Two honest caveats. First, these figures assume your password was chosen truly randomly, character by character - human-chosen passwords are far weaker than their length suggests, which the next section covers. Second, entropy only defends against guessing. It does nothing for a password that was reused and leaked in someone else's breach, or typed into a phishing page. Strength is necessary; it is not sufficient.

Why length beats complexity tricks

Most human password strategies add far less entropy than people think. Swapping a for @ and o for 0 is the first thing cracking software tries - those patterns are in every rule set. Appending the current year, a ! at the end, or a favourite team name adds almost nothing, because attackers try those shapes first. What actually works is raw length with randomness in it: a 16-character random string from the generator has about 99 bits no matter how it looks, while a 'clever' 9-character construction a human invented may test in the 20s. Strength estimators know this, which is why ours estimates from length and pool size and then penalises repeats, sequences and common words - paste a password into the Password Strength Estimator and a pattern-heavy one scores far below what its length promises.

Passphrases: the other way to buy length

If you must memorise a password rather than store it, a passphrase is the practical route. Five words chosen truly at random from a large word list (the classic Diceware list has 7,776 words, worth about 12.9 bits each) give you roughly 65 bits - comfortably past the floor - in something a human can actually remember and type, like a string of unrelated everyday words. The catch is the word "random": words you pick yourself drift toward predictable phrases, song lyrics and quotes, all of which are in cracking dictionaries. Roll dice or use a generator to choose the words, and the maths holds.

What to actually do

Generate, do not invent. Use the Password Generator at 16 characters or more for important accounts - that is about 99 bits, beyond any realistic brute-force attack.
Make every account unique. Entropy cannot save a password that leaks from one site and gets replayed on another.
Store them in a password manager so length and uniqueness cost you nothing to maintain.
Keep one strong memorised passphrase for the manager itself, built from randomly chosen words as described above.
Turn on two-factor authentication where it exists; it covers the cases - phishing and leaks - that entropy does not.

That is the whole subject, honestly: strength is counting, counting rewards length, and length is free when a machine does the remembering. Generate long, keep them unique, and the bits take care of themselves.

Wi-Fi QR Codes: The Format Explained

Umar — Tue, 27 Oct 2026 16:30:27 +0000

Published: 22 October 2026 · by QuikTools Hub

Point your phone camera at the little square on a cafe table and you are connected - no typing a password full of look-alike characters, no asking the staff to spell it. A Wi-Fi QR code feels like a trick, but there is no magic in it at all: the code simply contains one line of text written to a convention that phone makers agreed to recognise. Once you can read that line, you can check any Wi-Fi code before you print it, spot why a hand-made one fails to scan, and understand exactly what you are sharing. You can make one in about a minute with our QR Code Generator - it has a dedicated Wi-Fi mode - but first, the format itself.

The payload, field by field

Here is a complete, real example. Everything a phone needs to join the network "HomeNet" is in this single string:

WIFI:T:WPA;S:HomeNet;P:correcthorsebatterystaple;H:false;;

WIFI: - the opening label. It is how the scanner knows this QR code is Wi-Fi credentials and not a web link or plain text.
T:WPA - the authentication type. WPA covers WPA and WPA2 home networks. Older routers may use WEP; open networks with no password use the value nopass and leave the password field out entirely.
S:HomeNet - the SSID, which is simply your network name, exactly as it appears in a list of available networks, including capitals and spaces.
P:correcthorsebatterystaple - the password, stored as plain readable text. Hold that thought; the security section below explains why it matters.
H:false - whether the network is hidden (it does not broadcast its name). Almost every home and cafe network leaves this false.
The semicolons separate the fields, and the doubled semicolon at the end closes the string. Miss one out and strict scanners may refuse the code or quietly join the wrong thing.

That is the entire format. It was popularised by the ZXing barcode project, Apple added support to the iPhone camera in iOS 11, and Android followed - which is why virtually any phone from the last several years joins a network from a scan with no extra app. The same generator also has a vCard mode for contact cards, which works the same way: a small, standard block of text that phones know how to interpret.

Escaping: the fiddly part

Semicolons separate the fields, so what happens if your password itself contains a semicolon - or a comma, a quote, a colon, or a backslash? Those five characters have jobs inside the format, so each one must be escaped: written with a backslash in front of it. A password typed as p@ss;word has to appear in the payload as p@ss\;word. Get this wrong and the phone reads the password as ending early, then tries to join with the wrong credentials and fails without any useful error. This is the single most common reason hand-written Wi-Fi codes do not work, and it is why generating the code with a tool that escapes automatically - the QR Code Generator does - beats typing the string yourself.

An honest note on security

A Wi-Fi QR code is not a way of protecting your password. It is the opposite: it is your password, printed in a machine-readable font. Anyone who scans the code - or photographs it and scans the photo later - can read the actual password out of it in seconds. Treat a printed Wi-Fi code exactly like writing the password on a noticeboard, because that is what it is. For a guest network, a cafe, a holiday let or a workshop, that is usually fine and is the entire point. For your main home network, it is still reasonably safe while the code stays inside your house - but never post a photo of it publicly, and if one does escape, change the password and print a fresh code. For anything genuinely sensitive, the right answer is a separate guest network with its own password, which most modern routers offer.

Printing a code that actually scans

Size: 2 cm (about 0.8 inches) square is a workable minimum for a simple code on a table tent; go bigger - 3 to 4 cm - for a poster people scan from a step away, or when a long password makes the pattern dense.
Density: the more text in the payload, the more tiny squares the code needs. Long network names and long passwords both add density, so give dense codes more physical size.
Contrast and quiet zone: dark squares on a light background, with a clear margin of blank space around the code. Do not crop tight to the pattern, invert the colours, or print it over a photo.
Test before you commit: scan the printed copy with two different phones, at the distance and in the lighting where guests will actually use it. A code that only scans on your phone, up close, in bright light, will fail in the wild.

How to make one here

Open the QR Code Generator and choose the Wi-Fi tab (the other tabs make link/text codes and contact cards).
Type your network name exactly as it appears on your phone, and your password.
Pick the encryption your router uses - WPA/WPA2 for almost every modern network, WEP only for very old ones, or None for an open network - and tick Hidden network only if yours does not broadcast its name.
Click Generate QR code, then download the PNG (256, 512 or 1024 pixels - pick 1024 for print) and test-scan it before putting it out.

Everything happens in your browser: the network name and password you type are encoded on your own device and are never uploaded, which is rather the point when the thing you are encoding is a password. If a guest ever needs the details read back from an existing code, our QR Code Scanner decodes codes from an image or a camera, also locally. Browse the rest of the QR & Barcode Tools for barcodes and related utilities.

How to Calculate a Percentage Increase (With Examples)

Umar — Tue, 27 Oct 2026 16:28:12 +0000

Published: 22 October 2026 · by QuikTools Hub

A percentage increase tells you how much something has grown compared with where it started. It is the number behind pay-rise announcements, rent increases, price rises and exam-score improvements, and it is surprisingly easy to get wrong: use the wrong starting value and the same change can come out as 13% or 15% depending on how you divide. This guide walks through the formula once, slowly, then works three real examples so you can reuse the method on any pair of numbers. If you would rather skip the arithmetic, the Percentage Calculator has a percentage-change mode that does exactly this calculation and shows its working.

The formula

Percentage increase has two steps. First, find the increase itself: subtract the old value from the new value. Second, divide that increase by the old value - the starting point, never the new one - and multiply by 100 to turn it into a percentage.

Percentage increase = ((new value - old value) / old value) x 100

Dividing by the old value is the whole trick. You are asking: the growth is what fraction of what I started with? A rise of 12,000 sounds huge until you learn it started from 80,000, at which point it is a solid but unspectacular 15%. The same logic is why a shop can honestly advertise a large percentage increase on a tiny starting price, and why percentages quoted in news stories are always worth one question: increase relative to what?

Worked example 1: a pay rise

Your salary goes from 80,000 a year to 92,000 a year. Step one: the increase is 92,000 - 80,000 = 12,000. Step two: divide by the old salary, 12,000 / 80,000 = 0.15, and multiply by 100. Your pay rise is 15%. A quick sanity check: 10% of 80,000 is 8,000 and 20% would be 16,000, so an increase of 12,000 sitting between them at 15% makes sense.

Worked example 2: a price jump

A train pass costs 250 one month and 300 the next. The increase is 300 - 250 = 50. Divide by the old price: 50 / 250 = 0.20, so the price has gone up by 20%. Notice that the increase, 50, is 20% of the old price but only about 16.7% of the new one - which is exactly why dividing by the wrong end of the change gives a different, wrong, answer.

When the answer is negative: decreases

The formula does not change when a value falls; the answer simply comes out negative. Say a jacket drops from 500 to 425. The change is 425 - 500 = -75, and -75 / 500 = -0.15, so the result is -15%, which you report as a 15% decrease. Treat the minus sign as information rather than an error, and describe the result in words - "a 15% fall" - so nobody has to decode it. On the Percentage Calculator, the percentage-change mode shows the direction for you, which saves a step when you are checking a whole list of figures.

Three mistakes that give wrong answers

Dividing by the new value. 12,000 / 92,000 gives about 13%, not 15%. The old value is always the denominator for an increase or decrease.
Confusing percentage points with percent. If an interest rate moves from 4% to 5%, that is a rise of 1 percentage point, but a 25% increase in the rate itself (1 / 4 = 0.25). Both statements are true; they answer different questions.
Adding percentages up and down. A price that rises 10% and then falls 10% does not return to its start: 100 becomes 110, then falls by 11 (10% of 110) to 99. Percentages of different bases never simply cancel out.

All three mistakes have the same root cause: forgetting that a percentage is always a percentage of something, and that the "something" changes as the value changes. Keep the base straight and the rest of the calculation looks after itself. For chains of workings - adding tax, then a discount, then a fee - a plain Standard Calculator helps you keep each intermediate step visible.

Percentage increase vs percentage difference

These two get mixed up constantly. Percentage increase (and its sibling, percentage change) compares a value against its own earlier value, so order matters: going from 80 to 100 is +25%, and going back from 100 to 80 is -20%. Percentage difference compares two values that have no before-and-after relationship - two quotes for the same job, say - and divides by their average instead, so the answer is the same whichever way round you put them. Use change when time or sequence is involved, difference when you are comparing two things side by side. Both modes are built into the percentage calculator linked above, alongside our other Calculators.

Common questions

Can I calculate a percentage increase from zero? No. Division by zero is undefined, and any growth from nothing is not meaningfully a percentage - describe it in absolute terms instead ("sales began at 40 units"). In practice, if the old value is zero, there is no percentage answer, and any spreadsheet that shows one is hiding an error.

How do I turn a percentage increase back into the new value? Multiply the old value by 1 plus the percentage as a decimal. A 15% rise on 80,000 is 80,000 x 1.15 = 92,000. For a decrease, subtract instead: a 15% fall is 80,000 x 0.85 = 68,000.

Why does my spreadsheet show a slightly different answer? Usually rounding - the sheet is keeping more decimal places than it displays - or one of the mistakes above, most often dividing by the new value. Recompute by hand with the formula at the top of this page; if the two disagree, the hand calculation is the one to trust.