We Have Zero Google Clicks and Perplexity Cites Us on Page One
Zero Google clicks, then Perplexity cited us. We ran the same build for three clients: 419 articles scanned in Google AI Overviews, 162 cited.
machina sculpsit · homo probavit · MMXXVIIn 1747 a naval surgeon named James Lind ran a trial on twelve scurvy patients aboard HMS Salisbury. Six treatments, two men each. The pair given oranges and lemons were back on duty inside a week. The Royal Navy then took another forty-eight years to put citrus on its ships, which is a long time to sit on an answer that already worked.
We found ours on the wrong engine. Fifteen days after tryres.ai launched, Google had sent us 408 impressions and zero clicks, and Perplexity was citing us as a primary source. We had no backlinks, no domain age, and two articles.
So we ran the same build for paying clients, and this is what came back.
Fifteen Days In, 408 Impressions and Zero Clicks
The site was on page one for hundreds of queries. Not one of them produced a click.
Google Search Console reported 408 impressions, 0 clicks, and an average position of 9.9 for tryres.ai across the 15 days after launch (tryres.ai GSC, May 2, 2026). The site was on page one for hundreds of queries. None of them produced a click.
The most-impressed query was “scrupp.com capterra reviews rating 2024 2025”: 66 impressions, zero clicks. Someone wanted Scrupp’s reviews on Capterra. Google returned two essays about citation patterns that each mention Scrupp in passing, ours at #2.
Nobody clicks a research paper when they asked for a star rating.
This was not a crawling problem. We were indexed, we were visible, and we were sitting on page one for queries Google’s ranking system considered reasonable. The standard prescription from here is to keep publishing, build links, and wait two years for trust signals to pile up.
We had a different signal on the screen.
Perplexity Cited the Article Google Would Not Click
On the Perplexity query “domain authority in AI citations,” our Authority Is Not the Moat in AI Search came back as the #1 cited source, above PRLog, DigitalStrategyForce, DigitalApplied, and Chudi (Perplexity, May 2, 2026). Perplexity quoted our 93-of-100 finding verbatim and attributed it to Res by name.
None of the domains we sat above are small. PRLog has distributed press releases since 2007. The others run content libraries with thousands of indexed pages. On the thing Perplexity was scoring that day, none of them beat a fifteen-day-old domain carrying one complete article on the exact question.
The same two weeks, the same file, two engines, two verdicts. Google: position 9.9, zero clicks. Perplexity: the document the answer was built from.
Only the reader changed.
Both Articles Were Built the Same Way
Both cited pieces shared one build: a methodology block with a stated sample size, a multi-row comparison table naming real domains, an FAQ answering practitioner questions, and a third-party citation under every claim (Res competitor page audit, May 2026).
We did not publish thirty articles before launch. We published two that cleared the structural floor. Our 852-article study had already found six features present in 94% of top-cited B2B pages and 0% of the bottom fifty, so we wrote to our own findings and watched an engine confirm them on a two-week-old domain.
That is a sample of one, on one engine, in one market. A finding that small is a hypothesis with good manners.
So we spent the next four months testing it on other people’s money.
We Ran the Same Build for Three Client Programs
51% of B2B software buyers now begin research with an AI chatbot more often than with Google (G2, 2026), so we rebuilt three client content programs to the same specification and scanned every published article. The three markets share nothing: an AI security company selling to enterprise buyers, a Fortune 500 automotive group selling cars in one metro, and a regional contractor running more than 100 crews.
Every article title is run as a query, and an article counts as cited when a page on the client’s own domain appears in the answer’s citation slots. Errors and empty reads are dropped rather than recorded as misses.
| Market | Articles scanned | Cited | Rate | AIO citations | Avg position |
|---|---|---|---|---|---|
| AI security | 201 | 76 | 38% | 198 | 3.1 |
| Automotive group | 132 | 45 | 34% | 98 | 2.0 |
| Regional contractor | 86 | 41 | 48% | 58 | 2.0 |
| Combined | 419 | 162 | 39% | 354 | 2.4 |
Every Program Started From Nothing
No cited pages, no authority argument, no link program. A standing start, three times over.
Not one of these three brands had a page cited in an AI answer when we started. No cited pages, no authority argument, no link building program, and in two of the three cases no content library worth the name. A standing start, three times over, in three markets with nothing in common.
Nothing meant something different in each case. The security company sells into a category where the incumbents are the largest names in enterprise software. The contractor competes in metros where rivals have run town-specific sites for years, the kind with the town in the domain name.
The automotive case is the sharpest of the three, because its record was not blank. It was wrong. The store had traded under one name for decades, was bought by a national group, and renamed, and the equity stayed behind with the retired brand. AI would frequently cite the old identity, answering best dealer near me with the retired domain sitting in the citation slots. Buyers were asking those same engines what had happened to the former name, whether it was still open, and who owned it now.
The dealership had not lost a ranking. It had lost its name, and the machines were still pointing buyers at the one it no longer used.
So that job was to take an identity back. The scans above are what taking it back looks like.
That is the same position our own domain was in at fifteen days old, and it is the position most companies assume disqualifies them. The received wisdom says you earn citations the way you earned rankings: years of domain authority, a backlink profile, an aged domain. None of these three had any of it.
What they had was published articles built to a known shape. So the question worth asking is not whether a brand is old enough.
It is how many articles are on the board.
More Articles Published, More Citations Earned
The automotive program went from 45 published articles to 132 across five weekly scans, and its Google AI Overview citations climbed from 67 to 98 over the same stretch (Res first-party scans, August to September 2026). The line moves with the count.
This program earns more than one citation per article, because a single article can hold a slot in several different overviews at once. At 45 articles it was carrying 67 citations.
The ratio moves in a way that looks like decline and is not. At 45 articles the program earned 1.49 citations for every article scanned; by September it earned 0.74 across 132. The count of citations climbed the whole time. Newer articles had not finished queuing yet, and every batch you add dilutes the average on its way in.
Falling ratio. Rising results.
The Contractor Program Ran the Same Line
The contractor went from 52 published articles to 86, and its citations climbed from 13 to 58 (Res first-party scans, August to September 2026). Different market, different buyer, same relationship between what is on the board and what comes back.
The contractor climbed more slowly and from a lower base, which is what a young library in a crowded local market looks like. Its citation rate is now the highest of the three at 48%.
The youngest library, the best rate.
The AI security program sits at the other end of the same line: 201 published articles returning 198 citations in a single scan. Nothing about that company is older or better linked than it was in May. There is simply more of it on the board.
Only the count changed.
Google Seats the Cited Client at Position Two
Position inside the overview decides whether a reader sees the name at all. Both current programs average position 2.0 in the overviews that cite them, and the security program averaged 3.1 at its final scan (Res first-party scans, 2026).
Share of the available citation slots lands between 8.1% and 10.7% across the three. An overview cites several sources, so holding a tenth of every slot in your market while sitting at position two is a strong result.
The Queries That Convert Are Not the Ones You Buy
The prompts that earn these citations are not head terms. Across all three programs the winning queries cluster into a few kinds, and none of them is the keyword a media plan would pay for.
| Query type | What it looks like | Where it wins |
|---|---|---|
| Branded entity | A company’s own name plus a question | Position 1, repeatedly |
| Local best-of | Best provider in a named town | Position 1 to 3 |
| Process and consideration | What to ask before you sign | Position 1 |
| Comparison | One provider against a named rival | Position 1 |
The contractor’s strongest performers are consideration-stage questions a buyer asks before requesting a quote, and town-level best-of queries across the metro. The automotive program wins branded service and process questions. The security company wins branded product questions and head-to-head comparisons against named rivals.
That is the part traditional keyword work misses.
These are long, specific, low-volume questions, the ones a volume-first program deletes for having no search volume, and they are exactly where a buyer with a wallet ends up.
Where This Sits Against the Other Approaches
Every program in this market picks one of four ways to spend the money, and they differ less in cost than in what they actually put on the internet. The dimensions that matter are what each one publishes, how long it takes to earn a citation, and what it cannot see.
| Approach | What it produces | Time to first citation | The blind spot |
|---|---|---|---|
| Visibility dashboard | A chart of your absence | Never, it publishes nothing | Watching is not competing |
| Traditional SEO | Rankings and links | Quarters to years | Optimizes for clicks that stopped coming |
| Volume publishing | Pages, quickly | Unpredictable | Volume alone does not earn citations |
| Structure-first publishing, how Res works | Articles built to the cited shape | Weeks | Needs a human who knows the subject |
Then It Showed Up in the Blue Links
The AI security company published 76 articles in six weeks from a standing start. Scanned against both surfaces afterwards, 43% were the #1 AI answer and 34% were on page one of Google organic, with 28% at the #1 result (Res scan of the six-week cohort, July 2026).
That last pair is the part nobody expects. The same articles built for the citation went and took the blue links too, above Microsoft, Gartner, IBM, Anthropic, and OpenAI on the questions that pay, from a domain with no link program and six weeks of history.
The reason is not mysterious. Both surfaces read the same page and ask a version of the same question: does this document answer the thing that was typed? A page that answers one specific question cleanly, with its evidence attached and its sections legible, satisfies the overview and the ranking system in the same motion. The trade-off everyone braces for does not appear, because it was never two jobs.
Google’s ranking system and Google’s AI Overview are reading the same pages. Write for the one that answers questions and you are not giving up the one that lists them.
How Res AI Gets 39% of a Client Library Cited
There is no auction here and no budget to outspend. A contractor with 100 crews and a Fortune 500 group get the same hearing from the same engine, and the better, more honest answer takes the slot. You cannot buy the position.
The only variable left is the work.
Two things reliably fail. Flooding a site with generated pages fails, because the engines were built to see through volume. Watching a dashboard fails too, because it reports impressions while the pipeline stays empty.
A chart of your absence has never once ended it.
If you are looking for a vendor who can get you cited, Res runs content programs for companies of every size and category: enterprise software, a Fortune 500 automotive group, a regional contractor with more than 100 crews. Different markets, different buyers, the same program.
Across three of those programs we have scanned 419 published articles inside Google AI Overviews, and 162 of them earned a citation at an average position of 2.4. One client published 76 articles in six weeks from a standing start; 43% became the #1 answer in AI search and 28% took the #1 result on Google, above Microsoft, Gartner, IBM, Anthropic, and OpenAI on the questions that pay.
We do not sell a dashboard. Content goes in, citations come out, and a person signs every page.
Questions
Do I need domain authority to get cited?
No. Our own fifteen-day-old domain was cited as Perplexity’s #1 source above four established content libraries, and our 1,000-query study found non-giant domains hold the stable #1 citation on 93 of 100 B2B queries. None of the three client programs here began with an authority position.
How many articles does it take?
More than two and fewer than you fear. One program was carrying 67 citations at 45 published articles; another reached 58 citations at 86. Lind settled scurvy with twelve patients, so the bar for a useful sample is lower than the industry likes to pretend.
Why did the citations-per-article ratio fall while citations rose?
Because new articles dilute the average before they mature. The automotive program went from 1.49 citations per article at 45 articles to 0.74 at 132, while the raw count climbed from 67 to 98.
What counts as cited in this measurement?
A page on the client’s own domain appearing in the citation slots of the Google AI Overview returned for that article’s title. It is domain-level on purpose: Google frequently answers an article’s query using a different page on the same site.
Does winning an AI answer cost you the organic ranking?
The opposite, on this evidence. Of the 76-article cohort, 43% were the #1 AI answer and 28% held the #1 organic result, from the same pages.
What kinds of queries should a program target first?
Consideration-stage questions, local best-of queries, and named comparisons. They look unattractive in a keyword tool because the volume is low, and they are where buyers with budget arrive.
We have no content library at all. Where does that leave us?
At the same starting line as all three programs here, none of which held a citation position when they began. A scan measures what you have published, so the only position you cannot recover from is not publishing.
How often should citations be re-checked?
Weekly. Between 40% and 60% of domains cited in one month are absent the next on the same prompts (Profound, 2026), so a single screenshot is a lucky afternoon rather than a result.
Res AI is the answer to a record that is blank or pointing at the wrong name. Ten articles free, published into the CMS you already run.