AIToolsRecap: one of my pages has 395,000 impressions and 161 clicks

I posted here in May arguing that you should build comparison content before you launch, chase impressions, and that the floor matters more than the launch spike. three months and 4.89M impressions later I can say which part was right and which part I got badly wrong.

THE NUMBERS

March 18 launch. today:

4.89M Google impressions, 4,260 clicks
314 ChatGPT citations, 127 Copilot, 468 total across six AI engines
DR 30, 799 referring domains
$0 on ads, $0 on backlinks

sounds fine until you divide. 4,260 clicks out of 4.89M impressions is 0.087%. that is a terrible number and it took me a long time to understand why.

WHAT I FOUND WHEN I ACTUALLY LOOKED

I have two completely different kinds of page and I'd been counting them as one thing.

huge impressions, nobody clicking (6 months):

grok voice mode limits — 395,824 impressions, 161 clicks (0.04%)
i built 5 grok agents — 280,375, 123 clicks (0.04%)
best grok ai agents — 110,793, 244 clicks (0.22%)

small impressions, real clicks (28 days):

kimi moderato vs claude pro — 1,370 impressions, 88 clicks (6.4%)
kimi vs claude code vs codex — 4,923, 165 clicks (3.4%)
gemini 3.6 flash vs 3.5 flash vs luna — 2,410, 75 clicks (3.1%)

150× difference in click rate.

the big pages rank for queries where Google answers it in the SERP. someone searching "grok voice mode limits" wants a number, gets it from Google, and never arrives. someone searching "kimi moderato vs claude pro" is thirty seconds from a decision and has to click to make it.

to be fair to the data — pulling those three pages out only moves my site CTR from 0.087% to 0.091%. they're not the cause, they're just the clearest example of a pattern that runs through most of what I published.

WHAT I GOT WRONG IN THE MAY POST

I told people to build comparison content. still true. but I framed the goal as impressions and that was the mistake. a page ranking #4 for a 200,000-volume query can be worth less than a page ranking #4 for a 1,400-volume query.

I spent three months feeling good about a graph that was going up while the thing I actually needed was flat.

the encouraging part: once the comparison pages started taking share, clicks went up while impressions went down. last 28 days = 2,630 clicks. last 7 days = 1,260 of them. 48% of the month's clicks in a quarter of the time, on falling impressions. CTR went 0.2% to 0.5%.

THE THING THAT ALMOST STOPPED ME BUILDING THIS

Ahrefs currently estimates my organic traffic at 121 visits a month. Search Console says 2,630 clicks in 28 days. it is off by more than 20×.

if you are pre-launch and checking whether a niche is worth entering by looking up incumbents in Ahrefs or Similarweb — those tools badly undercount newer sites. I nearly talked myself out of this on a number that was wrong by an order of magnitude.

ON AI CITATIONS, SINCE EVERYONE ASKS

314 ChatGPT citations, but they come from 14 pages. Copilot, 127 across 14 pages. out of a couple hundred published. a very small number of pages do all the citation work.

spread: ChatGPT 314, Copilot 127, Gemini 10, Google AI Mode 7, AI Overviews 6, Perplexity 4. ChatGPT and Copilot are basically the whole game, the rest is rounding error.

what I'd like to know from anyone else running content: what's your worst impressions-to-clicks page? I want to know if 0.04% is unusual or if everyone has these and just doesn't look.

(and if you're building an AI tool and want a comparison page pointed at you rather than at a competitor, listing is free — https://aitoolsrecap.com/JoinUs.aspx. not the point of the post, just leaving it here.)