Flagship · Answer-engine visibility at scale
Cited 241,500 times. Clicked 2,500 times. Same platform.
Microsoft Copilot uses this site as a source roughly a hundred times for every organic click Bing sends it. Across 1,640 grounding queries, in five languages, with 52% of those citations on commercial-intent searches. One page alone has been cited 108,546 times. This is the study the rest of this portfolio was building toward.
Cited by Copilot
241,500
Times a page on this site was used as a source in a Microsoft Copilot answer.
6 months, ending 13 August 2026.
≥ 97 : 1
citations per
organic click
Clicked from Bing
2,496
Organic visits Bing sent to the site over the same period it was citing it.
24 months, ending 13 August 2026.
Why “≥” and not “=”. Both figures come from Bing Webmaster Tools for the same property, but the reports offer different windows: AI citations are only available over 6 months, organic clicks over 24. Dividing them yields 97, and because the citations were counted over a quarter of the time, the true same-window ratio is higher than that — not lower. I publish the floor rather than estimate the real number.
1,640
grounding queries Copilot cited this site for
197
of those queries above 50% citation share
70.6K
Google AI appearances, across 804 pages
51.7%
of citations on commercial or comparison intent
0.6%
Google click-through rate — and that is fine
Property
ZPlatform.ai · owned and operated
Type
Independent AI tools directory, hand-tested reviews
Method
Every tool used 3–10 hours before it is ranked
Window
16 months, ending 13 August 2026
Pages cited by AI
804 in Google, 294 in Copilot
Paid promotion
None
Why this is the flagship
An order of magnitude beyond everything else I own
The four case studies before this one each found one piece of the picture. Coimbatore Junction showed that a page can be cited far more than it is clicked. Velankanni Shrine showed that citation share is measurable and winnable. MatchMaker.lk showed that recommendation queries behave differently from fact queries.
ZPlatform.ai is all of it at once, at a scale that makes the pattern impossible to dismiss as a quirk of a small property.
Bing Copilot citations across the owned portfolio
All five figures are the “Total Citations” headline from the same Bing AI Performance dashboard, over 6-month windows. The chart is linear and deliberately not log-scaled — the gap is the finding.
Source: Bing Webmaster Tools AI Performance dashboards, exported 12–13 August 2026. SEOTamil.com reports per-page citations rather than a dashboard total, so its figure is the sum of its page export. Bing states these reports are a sample of overall activity, so every figure is a floor.
Evidence · AI search
Both platforms, both dashboards, same direction
Bing · 241.5K citations, 24 pages cited per day
Citation sources: Microsoft Copilots and Partners. Both the citation line and the cited-pages line climb steeply from June onward, with cited pages reaching 60–80 a day. The spike in early April is a single query going viral.
Google · 70.6K appearances across 804 pages
Ending on a spike to 3.8K impressions in a single day. The per-page export runs to 804 rows — the widest AI footprint in the portfolio by a wide margin, and the opposite shape to MatchMaker's 87%-on-one-URL concentration.
Copilot citations per page
Top 10 of 294 cited pages. The image-generator guide alone carries 46% of the site's entire citation volume — and earns 193 Google clicks.
236,676
citations across all 294 cited pages, of which the top 10 are charted
Source: Bing Webmaster Tools AI page citation report, exported 13 August 2026. This export sums to 236,676, which is 98% of the 241,500 dashboard headline — Bing attributes most but not all citations to a specific page.
Citations by query intent
Bing labels every grounding query by intent. Commercial and Comparison together are 51.7% of attributed citations — people choosing a tool, not reading about AI. This is the number that makes AI citation a commercial channel rather than a vanity metric.
157,058
citations attributed to the 1,640 grounding queries in the export
Source: Bing Webmaster Tools grounding-query export, grouped by Bing's own intent label. Bing notes these labels are AI-generated. Six minor intent categories with fewer than 500 citations each are omitted from the chart.
Biggest grounding queries by citation volume
| Grounding query | Intent | Citations | Citation share |
|---|---|---|---|
| how much water does ai use | Informational | 7,061 | 8.64% |
| top AI SEO agencies AI-driven services | Informational | 6,562 | 15.91% |
| best free ai image generator | Commercial | 5,974 | 22.46% |
| meta ai | Informational | 4,558 | 0.24% |
| best free AI image generators 2026 | Commercial | 4,195 | 10.35% |
| best free AI image generator 2026 | Commercial | 2,556 | 9.68% |
| top AI SEO agencies | Commercial | 2,476 | 10.59% |
| what jobs are safe from ai | Research | 2,084 | 36.10% |
Note “meta ai” — 4,558 citations at a 0.24% share. On a query that large, a fraction of a percent is still thousands of appearances. Share and volume are separate goals.
The finding nobody is exploiting
An English-only site, cited in five other languages
Nothing on this property is localised. There is no hreflang, no translated content, no regional subfolder. And yet:
| Grounding query | Language | Citations | Citation share |
|---|---|---|---|
| beste kostenlose KI Porträtgeneratoren | German | 171 | 40.05% |
| meilleurs générateurs de portraits IA gratuits | French | 136 | 68.34% |
| beste kostenlose KI Gesichtsgeneratoren 2026 | German | 134 | 55.37% |
| Spotify Black Friday promoções 2026 | Portuguese | 92 | 34.46% |
| en iyi ücretsiz yapay zeka portre oluşturma araçları | Turkish | 91 | 36.84% |
| mejores generadores de rostros IA gratuitos | Spanish | 70 | 72.92% |
57
non-English grounding queries in the export
1,832
citations earned from them
5
languages: German, French, Spanish, Portuguese, Turkish
The model is not translating the page. It is answering in the user's language from an English source — and crediting the source. A French search for the best free AI face generators returns this site at 81.82% citation share. A Spanish one at 72.92%.
It also reframes what a “market” means. Traditional international SEO says: to rank in Germany, publish in German. Answer-engine visibility says: to be cited in Germany, be the clearest source in any language the model can read.
Uncomfortably relevant
Copilot cites this property 22,810 times on “AI SEO agency” queries
Which is, precisely, the category I sell into. Worth stating plainly rather than leaving for you to notice.
| Grounding query | Citations | Citation share |
|---|---|---|
| top AI SEO agencies AI-driven services | 6,562 | 15.91% |
| top AI SEO agencies | 2,476 | 10.59% |
| expert SEO services for Google and AI search engines | 2,402 | 3.97% |
| top AI SEO agencies AI-driven SEO services | 1,307 | 13.09% |
| top AI SEO agencies AI-powered services | 1,229 | 15.44% |
| top AI SEO agencies AI-powered SEO services | 1,222 | 13.39% |
/guides/best-ai-seo-agencies/ has been cited
22,810 times — the fourth most-cited page on
the site. When somebody asks an assistant which AI SEO agencies are worth considering, this
is one of the sources the answer gets built from.
There is an obvious conflict of interest in an SEO agency owning a directory that gets cited on “best AI SEO agency” queries, and the honest thing is to name it rather than let you discover it. The relevant points: the guide is editorial and predates my agency's relaunch, the citation figures here are Bing's not ours, and the reason it is on this page is that it is the most direct possible demonstration that the method works — not an argument that you should trust the guide's rankings.
The constraints
A category that rewrites itself, saturated with generated content
The category refreshes itself every quarter
An AI tools directory is describing a market that reinvents itself constantly. A “best free image generator” list is wrong within months, and a wrong recommendation is worse than none — so the maintenance burden is the product, not an overhead on it.
Every competitor is doing the same thing with AI
The category is saturated with generated listicles. Being one of them earns nothing. The only durable difference is testing the tools — 3 to 10 hours each — and publishing the result of that testing rather than a summary of everyone else's.
The clicks were never going to come
These are queries an answer engine can satisfy in place: which free tool is best, how much water does AI use, what is a token. A 0.61% Google click rate was the predictable outcome. The choice was to accept it and optimise for the citation instead.
What was actually built
Six decisions that produced 241,500 citations
None of them is a trick. The first one is the whole moat.
Hand-test everything, publish the method
Every tool is used for 3 to 10 hours before it is ranked, and the scoring method is published. That is the one thing a generated competitor cannot copy, and it is the reason a model treats the page as a source rather than as another aggregation of the same claims.
One comprehensive page per commercial question
“Best free AI image generators” is one deep, maintained, comparison-table page rather than twelve thin ones. It now carries 108,546 Copilot citations — 45% of the site's total — because it is the most complete answer to a question people keep asking.
Answer the adjacent questions too
How much water does AI use. What is a token. What jobs are safe from AI. Those guides earn 34,000 citations between them and cost nothing in competitive terms — nobody is bidding for them, and they establish the site as a source across the whole subject, not just its shopping end.
Structure for extraction, not for reading time
Comparison tables, explicit criteria, verdict lines, real pricing. The same choices that made the Coimbatore pricing pages quotable, applied to a market where 52% of the citations are on commercial-intent queries.
Own the seasonal spikes properly
Individual Black Friday pages per tool, published early. This is where the organic clicks actually are: “grok black friday deal” converts at 21.17% from position 1.67, because a deal query needs a page and a date, not a summary.
Write plainly enough to survive translation
The site is English-only, and it is cited for queries in five other languages. That only happens when the sentences are simple enough for a model to render the meaning into German or Turkish without losing the fact.
Evidence · Search Console
1.42M impressions, 8,666 clicks, and why that is not a failure
A 0.61% click rate on 1.42 million impressions looks alarming until you read the queries. An answer engine satisfies almost all of them in place.
Search Console · 16-month performance
8.67K clicks, 1.42M impressions, 0.6% CTR, average position 16.2. All four metrics are on so the divergence is visible: impressions climbing hard through 2026 while clicks stay roughly flat.
Bing · 2.5K clicks, 167.3K impressions
Over 24 months at 1.49% CTR. This 2.5K is the denominator in the ratio at the top of this page — and Bing was citing the site 241,500 times over the last six months of it.
Where 1,000 the top 1,000 exportable queries actually sit
49 queries average position #1 and 451 sit beyond page one — a wide, shallow footprint, which is exactly what a directory covering a fast-moving category looks like. Search Console caps its export at 1,000 rows, so this is the distribution across the top 1,000 queries by clicks.
102
Top 3 · 10.2% of the queries measured
549
Page one · 54.9% of the queries measured
Source: Google Search Console, zplatform.ai, Search type: Web, 16 months ending 13 August 2026. Band thresholds are ours: #1 is an average below 1.5, since Search Console reports position as a decimal average rather than an integer rank.
Top queries · where the clicks actually are
| Query | Clicks | CTR | Pos |
|---|---|---|---|
| grok black friday deal | 239 | 21.17% | 1.67 |
| grok black friday | 194 | 24.62% | 1.71 |
| zplatform | 124 | 13.48% | 9.06 |
| suno black friday sale | 100 | 2.82% | 6.91 |
| zplatform ai | 80 | 31.75% | 1.68 |
| cvpr 2027 | 61 | 0.95% | 6.91 |
| cursor black friday | 60 | 7.29% | 4.68 |
| supergrok black friday | 41 | 17.60% | 1.71 |
Every high-CTR query here is a deal or a brand — “grok black friday deal” at 21.17% from position 1.67. A deal needs a page. A definition does not.
Top pages · Google, 16 months
| Page | Clicks | CTR | Pos |
|---|---|---|---|
| /black-friday-ai-tool/grok-ai/ | 1,416 | 0.85% | 7.5 |
| /black-friday-ai-tool/suno-ai/ | 554 | 1.41% | 7.3 |
| / | 475 | 2.49% | 10.86 |
| /black-friday-ai-tool/cursor-ai/ | 260 | 3.27% | 6.71 |
| /black-friday/ | 236 | 0.54% | 18.14 |
| /black-friday/lifetime-deals/ | 212 | 1.65% | 11.35 |
| /best-ai-tools/best-free-ai-image-generators/ | 193 | 0.71% | 10.13 |
The Black Friday pages take the clicks. The image-generator guide takes 193 of them — and 108,546 Copilot citations.
Google search appearances · structured data earning its keep
| Appearance | Clicks | Impressions | CTR | Avg position |
|---|---|---|---|---|
| Product snippets | 143 | 14,114 | 1.01% | 10.1 |
| Review snippet | 50 | 12,674 | 0.39% | 13.34 |
| Translated results | 40 | 1,573 | 2.54% | 9.66 |
Product and review snippets appear as distinct search appearances, which is the schema markup doing visible work — and the same structured claims a model extracts.
Who it reaches
Genuinely global, and overwhelmingly desktop
Google AI appearances by country
| Country | AI appearances | Share |
|---|---|---|
| United States | 23,430 | 33.2% |
| India | 6,961 | 9.9% |
| Japan | 4,555 | 6.5% |
| United Kingdom | 3,980 | 5.6% |
| Australia | 2,290 | 3.2% |
| Canada | 2,263 | 3.2% |
| Germany | 2,104 | 3.0% |
| Brazil | 1,152 | 1.6% |
The US is a third of AI appearances, but Japan, Germany and Brazil are all in the top eight — from an English-only site.
Google organic by country
| Country | Clicks | CTR | Pos |
|---|---|---|---|
| United States | 1,801 | 0.26% | 20.42 |
| India | 722 | 1.42% | 14.83 |
| Japan | 609 | 0.49% | 6.79 |
| Germany | 492 | 1.73% | 11.74 |
| United Kingdom | 393 | 0.45% | 14.4 |
| Canada | 292 | 0.78% | 12.94 |
| Pakistan | 287 | 3.14% | 20.51 |
Japan averages position 6.79 — far better than the US at 20.42 — on 123,608 impressions. A market hiding in plain sight in row three.
By device · Google organic
| Device | Clicks | Impressions | CTR |
|---|---|---|---|
| Desktop | 5,258 | 1,103,831 | 0.48% |
| Mobile | 3,282 | 305,463 | 1.07% |
| Tablet | 126 | 10,826 | 1.16% |
By device · Google AI appearances
| Device | AI appearances | Share |
|---|---|---|
| Desktop | 53,396 | 75.6% |
| Mobile | 16,347 | 23.1% |
| Tablet | 875 | 1.2% |
76% of AI appearances are desktop, matching the 78% desktop split on organic impressions. This is people researching tools at work.
Coverage, stated honestly
What is proven here, and what is not
| Surface | Status | Finding | Evidence |
|---|---|---|---|
| Bing Copilot | Documented | 241,500 citations in 6 months across 294 distinct pages and 1,640 grounding queries, at an average of 24 pages cited per day. 197 of those queries sit above 50% citation share. | AI Performance dashboard + 3 exports |
| Google AI features | Documented | 70,618 appearances spread across 804 pages — the widest AI footprint in the portfolio by an order of magnitude. | Search Console Generative AI export |
| Google organic | Documented | 8,666 clicks from 1.42M impressions at 0.61%, average position 16.2. The low click rate is the point, not a caveat. | Search Console exports + screenshot |
| Bing organic | Documented | 2,496 clicks from 167,326 impressions at 1.49%. This is the denominator in the citation-to-click ratio. | Bing Webmaster Tools exports |
| Multilingual AI answers | Documented | Cited for 57 non-English grounding queries across German, French, Spanish, Portuguese and Turkish — from a site published only in English. | Grounding-query export |
| Google rich results | Documented | Product snippets and review snippets both appear as distinct search appearances, at positions 10.1 and 13.34. | Search Console Search Appearance export |
| ChatGPT, Gemini, Perplexity | Not audited | No per-answer citation export exists for these surfaces. Given the Copilot and Google numbers it is reasonable to assume presence, but assumption is not evidence and this study does not claim it. | No export available |
What to do with this
The five things this property proves, in order of how much they should change your plan
One. Your AI visibility is probably already large and unmeasured. Two platforms publish it free — Search Console's Generative AI report and Bing's AI Performance report. Most teams have never opened either. Do that before anything else; you cannot manage a channel you have not looked at.
Two. Depth beats breadth for citation. One maintained, hand-tested, comparison-table page took 45% of this site's citations. The instinct to split it into twelve keyword-targeted variations would have destroyed it.
Three. Commercial intent is where the citations are. 52% here. An AI citation on “best tool for X” is a brand impression at the moment of decision, from a source the buyer trusts more than your ad. That is worth more than the click it replaced.
Five. Stop reading a site-wide CTR. 0.61% on this property is the sign of a successful source, and 9.27% on a booking page is the sign of a successful task page. One benchmark across both tells you nothing, and acting on it deletes the wrong pages — as the Coimbatore study shows.
For a SaaS company the translation is direct: your docs, glossary, comparison and integration pages are the citation assets. Test something properly, publish the method, structure it for extraction, and maintain it. That is the entire mechanism, and it is the same one that put 241,500 citations against an AI tools directory.
Who did this work
500+ tools bought, installed and tested
Alston Antony
Founder & reviewer
ZPlatform.ai is owned and operated by Alston Antony, who has personally bought, installed and tested 500+ software products since 2019. Every tool on the directory is used for three to ten hours before it is ranked, against a published methodology. That testing is the reason a model treats the pages as a source rather than as another aggregation of the same claims — and it is the one part of this that cannot be shortcut.
He has worked in search since 2010, holds an MSc in Computer Software Engineering (Distinction) from the University of Greenwich, and is a professional member of BCS, The Chartered Institute for IT. His day role is Senior Digital Marketing Manager at Brainstorm Force, whose software powers 7M+ sites.
How to verify every number on this page
- The headline ratio is a floor, not a measurement. Citations cover Bing's 6-month AI window; clicks cover its 24-month search window, because those are the windows the reports offer. The true same-window ratio is higher than 97, and I publish the floor rather than estimate.
- Three different Bing AI totals appear, and they should. The dashboard says 241,500; the per-page export sums to 236,676 across 294 pages; the grounding-query export sums to 157,058 across 1,640 queries. They are three different attributions of the same activity, and Bing states all of it is a sample.
- Google AI and Bing AI figures are never summed. One counts appearances inside generative features, the other counts citations in Copilot answers. Different metrics, different windows.
- Google totals come from Search Console, Search type: Web, the 16-month window ending 13 August 2026, using the Devices export for site totals. Query and page exports both hit the 1,000-row cap, so position bands are shares of that sample.
- Citation share is Bing's figure, not a calculation of ours. The chart is filtered to 40+ citations; the six queries at a literal 100% all have single-digit counts and are described in the copy instead of charted.
- Non-English queries were identified by scanning the grounding-query export for non-ASCII characters and known non-English terms. 57 matched. That is a detection method, not a certified language classification, and a handful of transliterated queries may be missed.
- Intent labels are Bing's, and AI-generated — its interface says so.
- The conflict of interest is declared in the AI SEO agencies section above rather than left for you to spot.
- The site is live. Open zplatform.ai and ask Copilot for the best free AI image generator.
Case study FAQ
The questions this raises
Is the 97:1 citation-to-click ratio a real measurement?
Why is a 0.61% click-through rate the headline of a success story?
How is an English-only site cited for German and Turkish queries?
Are these citations actually worth anything commercially?
What does one page with 108,546 citations tell you?
/best-ai-tools/best-free-ai-image-generators/ holds 45% of the entire site's Copilot citations and earns 193 Google clicks. It is one maintained, hand-tested, comparison-table page answering a question millions of people ask, rather than twelve thin variations of it. The same effort spread across twelve pages would almost certainly have earned a fraction of the citations.