Hey Builders 👋

Last week I had a tiny AI model read every single Google search that Careerkit shows up for. All 55,643 of them.

And let me tell you… it did not hold back.

The short version:

  • 21 of my 26 resume-example pages hadn't been shown to a single person on Google in 90 days.

  • One of my money pages wasn't in Google at all.

  • And my biggest "miss"? It turned out not to be mine to win. (That one's my favorite. It's below.)

Total cost of finding all that out: $3.55. 3x less than a coffee in Zurich.

Quickly, the rest of my week:

  • Careerkit: 51 resume-example pages and 25 cover-letter pages went live, all built from what this found.

  • ReplyBoss: it now hunts for buyers on X, not conversations. Across 10 products over 7 days, a typical one had 0.6 real buyers a day. (Yes, really. That's its own issue soon.)

  • HelloBuilder: 103 readers, same as last week. Hi 🫡

If You Get Some Google Traffic and Only Ever Read the Top 20 Searches… This Is for You

I want to talk directly to the founder whose site gets some Google traffic. You open Search Console once a month, scroll the top of the list, nod, and close the tab.

You're not bad at SEO. You're reading a list Google sorted for you, and the stuff that matters is hiding further down.

Quick disclaimer: I'm not an SEO agency. I'm one person building three products with AI. These are my real numbers, not a listicle someone asked ChatGPT to write.

Now, the receipts. In the 90 days to Oct 14, Careerkit got 31,467 clicks from Google out of 1.6 million impressions. Not bad! But I had no idea what most of those searchers actually wanted.

Here's everything below 👇

  • 🔍 What I actually did (in normal, human words)

  • 😬 My biggest miss, and why I didn't fix it

  • 👻 The 21 pages nobody saw

  • 🚪 The money page Google couldn't see

  • 🙅 What I'm NOT fixing, on purpose

  • 🛠 How to run your own this week

What I Actually Did (In Normal, Human Words)

The Search Console screen shows you a sorted list. The API gives you every search Google is willing to name. For me, that was 55,643 searches.

(Fun fact: even "every search" is only half. Those named searches carried 50.1% of my clicks. Google hides the rest.)

Then I handed every search to Jev, a tiny decision model, and asked it four questions:

  • What does this person want? A tool, a template, an example, a how-to, a job, or a brand?

  • What language?

  • Which document?

  • Is there a job title in it?

Here's the part most people skip: I tested it before I trusted it. I checked it against 307 searches with known answers, 90 of them held back. The bar was 90% precision while answering at least 70%. "What do they want" passed at 91%. Two questions failed, so I threw them out.

Quick math break:

  • $0.042 per million input tokens

  • × 55,643 searches

  • = $3.55 total, test included

Here's what came back, for English:

What they want

Impressions

Clicks

Click rate

A tool

256,129

11,132

4.35%

A how-to

192,269

268

0.14%

An example

52,197

94

0.18%

A competitor

41,688

114

0.27%

Free tools carry the whole site: 71% of named clicks. Templates, examples and how-tos? All under a 1% click rate. Ouch.

My Biggest Miss Wasn't Mine to Win (My Favorite Lesson)

The #1 thing I was "losing" was… cover-letter fonts. Yes, really lol.

  • "best font for cover letter": 6,791 impressions at position 9.4. 2 clicks.

  • "cover letter font": 4,539 impressions at position 5.0.

  • "cover letter font size": 2,012 impressions at position 2.6.

That's about 15,200 impressions and 12 clicks, all on one blog post that's already on page one.

So obviously, rewrite the title, right? That's what the first read said too.

Then I checked. The page works. The title is already right. Google answers "what font" right there on the results page, so nobody needs to click.

So I closed it. No fix.

Steal this: before you rewrite a title, Google the query yourself. If the answer is already sitting on the results page… move on.

The Ouch: 21 Pages Nobody Saw

I had 26 resume-example pages. In 90 days they got 7 clicks at an average position of 30. And 21 of the 26 got zero impressions. Not low. Zero.

So I opened the pages that actually rank for an accountant resume:

  • BeamJobs: 30 examples

  • Resume Worded: 34

  • Resume Genius: 14

  • Careerkit: …1 🫠

My title literally said "Examples." Plural. Then it gave you one.

The searcher wants a CHOICE. So I rebuilt all 26, added 25 more roles, and gave each of the 51 pages 8 examples. Every new role got a cover-letter page too. All live as of Oct 18 and 19.

Steal this: open the page that ranks for your best query and count what it gives the searcher. Then count yours. (It's humbling. Do it anyway.)

The Money Page Google Couldn't See

Next I reviewed the 10 pages that bring in the most paying users, against 5,452 searches: mine, plus everything the top 3 competitor pages rank for.

Three things jumped out.

1. The words were missing. My free resume builder page used "resume maker"… 0 times. It said "designs" 12 times and "templates" 3 times, even though people search for templates. The headline was "Build your resume from scratch, free." Now it's "The free resume builder with no sign-up."

2. One money page wasn't in Google AT ALL. Search Console called /fr/cv-gratuit a "Duplicate without user-selected canonical". The page Google picked instead? "Crawled, currently not indexed". So neither French page was in Google 🙃 I requested indexing on Oct 17 and fixed the footer that was sending French readers to English pages.

3. Some pages promised things the product doesn't do. One blog post said "50,000+ job seekers". The real number is 18,336 users. The translator FAQ offered "paste the text"… the tool only takes PDF or DOCX.

13 changed pages went live on Oct 18.

And let me keep it real: words don't fix authority. My free resume page still sits at 33 to 36 for "free resume builder". That part is links. So far I've sent 12 outreach emails and got 0 links. So… that's next month lol.

Steal this: run URL Inspection in Search Console on the 5 pages that make you money. Five minutes. It could save you months.

What I'm NOT Fixing (On Purpose)

Not every missed search is yours to chase:

  • German templates miss 90% of their searches, at an average position of 29.7. One cluster alone, "Erzieher Lebenslauf", has about 2,800 impressions. But German programmatic pages are parked, so it waits.

  • Competitor names got 41,688 impressions and 114 clicks. People searching for a brand mostly want… that brand. One exception got a fix: "jobright ai pricing" had 1,940 impressions at position 3.4 and 0 clicks. The title now answers the price.

  • Not every miss is a bug. "how to create resume" kept landing on my French page. The searchers were in Morocco, Algeria and Ivory Coast, where Google serves the French page on purpose.

So… How Do You Run Your Own?

The lazy version, for your site, this week:

  1. Pull every search from the Search Console API, not the screen.

  2. Tag each one with the cheapest model that passes a test you write FIRST.

  3. Sort by searches where you're already on page one and still not getting clicks.

  4. Google those yourself. If the answer is on the results page, skip it.

  5. Run URL Inspection on your 5 money pages.

That's it. One afternoon, a few dollars.

If you take one thing away: your Search Console already knows what people want from you. Read ALL of it once. Then make sure your money pages are even in Google before you write a single new one.

🧠 What I Learned on X

📰 AI News

🛠 Tools to Try

⚡ Skills to Try

📈 Growth & Money

💬 Builder Discussions

Okay, that's it from me. Reply with your question, I answer one every week. And if you've ever read every search in your Search Console, tell me what surprised you 👀

Keep shipping,

Nishant