The automation has been running for 43 days without missing once. Every evening at 21:00 UTC, it picks a calculator page, writes in verified benchmarks, adds worked examples, inserts internal links, and commits. One page per day. Every single day since May 3.
This week it finished the real estate cluster: Cap Rate, Rental Yield, NOI, DSCR, Cash-on-Cash, GRM, and WACC. Seven commits across seven days. Each one clean, each one improving real content with real numbers.
Combined GSC impressions for all seven pages in the past 28 days: 3.
The setup
The automation's job is straightforward. It picks a calculator page from a queue, improves the body content with verified data points, rewrites any thin sections, and adds internal links to related pages. The commits all follow the same message format: "SEO: improve [X] calculator content - add verified benchmarks, examples, internal links."
The pages it has worked on form three clusters so far:
- Advertising/SaaS metrics — CPM, CTR, CPC, ROAS, CPA, EBITDA, Burn Rate, Runway, LTV, CAC, ARR, MRR, Churn Rate (this was weeks 1–2 of the automation)
- Equity/finance — DCF, IRR, NPV, Payback Period, Vesting, Cap Table, Dilution, CAPM, Dividend Yield, Debt-to-Equity, Gross Margin, EPS, Retention Rate, Valuation Multiple, ROI (this was the block that ran through most of May)
- Real estate investment — Cap Rate, Rental Yield, NOI, DSCR, Cash-on-Cash, GRM, WACC (this week, June 7–13)
Forty-three pages improved. Forty-three daily commits. The question I went into this week's data review asking: is any of it working?
What I expected vs what the data showed
My expectation was incremental. Not a breakthrough — I knew the equity cluster produced almost nothing visible in GSC when I wrote about it two weeks ago. But I thought the real estate cluster might be different. These are more specific tool queries. Cap rate calculators, DSCR calculators — these serve a niche audience, real estate investors, and maybe that niche has less competition.
Here's what I found when I pulled the 28-day query-level breakdown for each of the seven pages:
| Page | Impressions (28d) | Clicks |
|---|---|---|
| /tools/cap-rate-calculator/ | 1 | 0 |
| /tools/rental-yield-calculator/ | 1 | 0 |
| /tools/noi-calculator/ | 1 | 0 |
| /tools/wacc-calculator/ | 0 | 0 |
| /tools/dscr-calculator/ | 0 | 0 |
| /tools/cash-on-cash-calculator/ | 0 | 0 |
| /tools/grm-calculator/ | 0 | 0 |
Three impressions. Across seven pages. The single impression on the cap-rate page came from the query "cap calculator real estate investment radius" — position 61. The rental yield one: "rental yield calculation for property" — position 54. The NOI page: "free net operating analysis" — position 32. None of these are the head terms these pages were built to target.
That's the third cluster now with this result.
Finding #1: The pages are technically solid. That's what makes this interesting.
I curled the cap-rate calculator as Googlebot to check what Google actually sees. The page isn't thin. It has:
<title>Cap Rate Calculator: Compare Investment Properties | Valuefy</title>
<meta name="description" content="Free cap rate calculator for real estate investors. Calculate NOI yield, compare properties side-by-side, and benchmark market rates by property type." data-rh="true">
<h1>Cap Rate Calculator for Real Estate Investors</h1>
There's a full FAQPage schema with 10 questions, a HowTo schema, a WebApplication schema, canonical URLs, hreflang. The meta description is specific. The H1 is descriptive. The content after the commit includes real cap rate benchmarks by property type — multifamily, retail, industrial — with source citations.
By any on-page checklist, this is a good page. That's the thing: it's not a thin-content problem. The automation is producing legitimate, useful content. It's just not producing search impressions.
Finding #2: 43 commits, and only 2 targets broke through
Looking across all 43 automation targets in the 28-day GSC data, only two appear in the top 50 pages by impressions: CAPM (1 click, 281 impressions) and DCF (1 click, 390 impressions). Both from the equity cluster, both with 1 click.
Everything else — 41 other pages that were improved by the automation over the past six weeks — either doesn't appear in the top 50 at all or has zero impressions in the query-level breakdown.
I checked the data twice because I wanted to be wrong about this. I wasn't.
The two that broke through (CAPM and DCF) are both established finance concepts with a small but real query stream. But even then — 390 impressions and 1 click in 28 days is not movement. It's barely a signal.
Finding #3: The pages that actually get clicks are a completely different category
Here's the part that shifted my thinking.
The 28-day page-level data for the whole site shows which pages are actually generating traffic. The top performers:
| Page | Clicks (28d) | Impressions (28d) |
|---|---|---|
| /tools/purchase-order-generator/ | 18 | 2,584 |
| /tools/impression-calculator/ | 7 | 957 |
| / (homepage) | 7 | 693 |
| /tools/balance-sheet-generator/ | 3 | 17 |
| /tools/conversion-rate-calculator/ | 2 | 55 |
The purchase order generator wasn't touched by the automation. Neither was the balance sheet generator. The impression calculator wasn't on the queue.
I pulled the query breakdown for the purchase order generator to understand why it works when the financial metric calculators don't. It has 25 distinct queries with impressions, including:
- "purchase order generator" — 2 clicks, 365 impressions, position 18.2
- "po maker" — 2 clicks, 60 impressions, position 7.0
- "free po generator" — 1 click, 62 impressions, position 18.2
- "free purchase order generator" — 1 click, 89 impressions, position 14.7
These are real queries. People searching for them want to generate a purchase order right now — not read about what one is. The cap-rate calculator's 1 impression came from a query so oddly phrased it suggests an almost accidental match.
The difference isn't content quality. It's user intent. A purchase order generator serves someone who needs to do something now. A cap rate calculator serves someone who wants to know something — and for financial knowledge queries, high-authority financial-media sites dominate.
Finding #4: The og:title bug is still there, on every single one of these pages
While curling the pages, I checked the og:title tag count. Every real estate calculator page returned 2:
<meta property="og:title" content="Valuefy - Free Business Calculators & Financial Tools">
...
<meta property="og:title" content="Cap Rate Calculator: Compare Investment Properties | Valuefy" data-rh="true">
The first tag is the static fallback in index.html. The second is the one React Helmet injects during prerender. Both land in the rendered HTML. This was the bug I found in week 1 of these posts, confirmed again last week. Still live. Still on every page.
I'm noting it because it's a clean double-check for anyone reading this series. It has not been fixed.
What I'm going to do about it
- Audit which pages are generators vs metric calculators — count how many generator-type pages exist on the site versus metric calculator pages. The current ratio matters.
- Reorder the automation queue — if generators serve immediate intent and have real query distributions, the automation should be improving those pages first. Not leaving them untouched while working through financial-metric calculators that compete against established financial sites.
- Don't kill the metric calculator work — the content is legitimate and will matter if the site's authority grows. But it's wrong to treat it as the priority when there's a category that's already getting traffic signals.
- Fix the og:title bug. This is the third time I've written a sentence like that.
The uncomfortable lesson
Forty-three days of daily automation, and the signal is clear enough to name: the automation is getting very good at improving a category that Google isn't currently surfacing for this domain.
The financial metric calculators — Cap Rate, WACC, LTV, IRR, DCF and their relatives — compete for knowledge queries in a space where established financial-media sites have years of authority. A daily content improvement routine on a low-DR domain doesn't change that ranking calculus quickly, if at all.
The generators — purchase orders, balance sheets, and similar tools — serve do-it-now queries. The competition is different. The intent match is cleaner. The domain-authority gap matters less when someone just needs to create a document.
I'd convinced myself that improving content quality across the calculators was a reasonable long-term bet. It might still be. But the data I'm reading today says I've been improving the wrong category first. The automation was doing exactly what I told it to do. I just told it to do the wrong thing for 43 days.
I'll check what happens when the queue shifts to generators. I'll report back with whatever the numbers say.
I'm running these experiments on valuefy.app and writing the findings as I go. If you're building programmatic SEO or watching your automation produce technically-good pages that Google ignores, I'd genuinely like to compare notes — drop a comment.
I also run AImiten, where we build AI tooling for companies. This side project is where the ideas get stress-tested first.
Top comments (1)
Hi aimiten, I saw your post about automating your content creation process. I'm guessing you're looking for a way to streamline your content creation workflow. Clypify helps with content aggregation and automation, which might be a good fit for you. We can help you find and rewrite relevant content from RSS feeds, making it easier to create high-quality content. Free plan at clypify.com — no card needed.