If I told you the businesses ChatGPT and Perplexity are citing most right now have, on average, fewer backlinks than those getting ignored, you’d probably think I got that backward. I didn’t. A new analysis of over 7,000 AI citations just confirmed it: total backlinks, total keywords, total traffic — all three showed either no relationship or a negative one with whether a page gets cited.
The metric that mattered most wasn’t a link. It was whether the content actually said enough to be worth quoting.
Hi, I’m Jeff Payne. You’re listening to The Jeff Payne Show, Episode #38: The Wrong Kind of Authority
For about 15 years, “build authority” has meant one thing: get more links, rank for more keywords, drive more traffic, and hope Google notices. Every agency pitch, every SEO audit, every dashboard you’ve ever looked at was built around that scoreboard. It made sense — it’s what the algorithm rewarded.
But AI platforms don’t work the way search engines used to. ChatGPT, Perplexity, and Google’s AI Overviews aren’t ranking ten blue links and letting you click through. They’re reading the actual content on a page and deciding, in real time, whether it’s worth citing as a source. And it turns out the scoreboard you’ve been playing to isn’t the one they’re using.
Here’s what actually separated the most-cited content from everything else: word count and sentence count — depth, essentially — showed the strongest positive relationship with citations across every platform tested. Readability mattered too. Meanwhile, backlinks, keyword rankings, and total traffic were flat or slightly negative.
There’s a wrinkle worth knowing if you’re paying attention to more than one AI platform: ChatGPT still gives some weight to domain rating — your site’s traditional authority score. Perplexity and Google’s AI Overviews barely do. They’re almost entirely reading the content itself and asking one question: Does this actually explain the thing thoroughly and clearly, or does it just gesture at it?
That’s a real shift in what “authority” means. It used to be a reputation you accumulated. Now, on the platforms increasingly deciding what your prospects see before they ever land on your site, it’s closer to a test you pass or fail on a page-by-page basis.
Here’s how stark the gap gets. One article on this exact topic — long, detailed, and comprehensive — ran north of 10,000 words and had a moderate readability score. It earned 187 total citations across AI platforms, 72 of them from ChatGPT alone. A different page, covering the same ground in under 4,000 words with a lower readability score, earned 3 citations. Total.
Same topic. Same basic intent. A citation gap that size isn’t explained by links or domain age. It’s explained by one page doing the work thoroughly, while the other one doesn’t.
This isn’t a call to pad your content or hit a word count for its own sake — that’s the wrong lesson, and it’s an easy one to draw if you stop listening halfway through this segment. Length is a byproduct, not the goal. The actual shift is in the question you’re asking before you publish anything. Stop asking “how many links does this need?” Start asking, “Does this actually say enough, clearly enough, to be the thing someone would want to quote?”
That’s a harder standard than a backlink count. It’s also the one that’s starting to matter more.
The businesses that win the next few years of search aren’t going to be the ones with the biggest link profile. They’re going to be the ones whose content actually holds up when an AI reads the whole thing and has to decide, on the merits, whether it’s worth repeating. Make sure yours does.
Why the metrics you’ve spent 15 years chasing barely register with the AI platforms deciding what your prospects see
For about 15 years, “build authority” has meant one thing: get more links, rank for more keywords, drive more traffic, and hope Google notices. Every agency pitch, every SEO audit, every dashboard you’ve ever looked at was built around that scoreboard. It made sense — it’s what the algorithm rewarded.
But AI platforms don’t work the way search engines used to. ChatGPT, Perplexity, and Google’s AI Overviews aren’t ranking ten blue links and letting you click through. They’re reading the actual content on a page and deciding, in real time, whether it’s worth citing as a source.
And a new analysis of over 7,000 AI citations across 1,600 URLs just confirmed something uncomfortable: the scoreboard you’ve been playing to isn’t the one they’re using.
The metric that mattered most wasn’t a link. It was whether the content actually said enough to be worth quoting.
What Actually Predicts A Citation?
Total backlinks, total keywords, and total traffic — the three pillars of traditional SEO authority — showed either no relationship or a slightly negative one with whether a page gets cited by AI. Word and sentence counts, on the other hand, showed the strongest positive relationship across all platforms tested. Readability mattered too, particularly for ChatGPT and Perplexity.
There’s a wrinkle worth knowing if you’re paying attention to more than one AI platform: ChatGPT still gives meaningfully more weight to domain rating than the other two.
Perplexity and Google’s AI Overviews lean almost entirely on content depth and clarity instead. Different engines, slightly different filters — but none of them reward what SEO trained us to chase.
It used to be a reputation you accumulated. Now it’s closer to a test you pass or fail on a page-by-page basis.”
How Wide The Gap Gets
One article on this exact topic — long, detailed, and comprehensive — ran north of 10,000 words and had a moderate readability score. It earned 187 total citations across AI platforms, 72 of them from ChatGPT alone. A different page covering the same ground in under 4,000 words, with a lower readability score, earned 3 citations. Total.
Same topic. Same basic intent. A citation gap that size isn’t explained by links or domain age. It’s explained by one page doing the work thoroughly, while the other one doesn’t.
The Lesson isn’t “Write More”
This isn’t a call to pad your content or hit a word count for its own sake — that’s the easiest wrong lesson to draw here. Length is a byproduct, not the goal. The actual shift is in the question you ask before you publish anything.
Stop asking, “How many links does this need?” Start asking, “Does this actually say enough, clearly enough, to be the thing someone would want to quote?”
Stop asking how many links this needs. Start asking whether it says enough to be worth quoting.”
That’s a harder standard than a backlink count. It’s also the one that’s starting to matter more. The businesses that win the next few years of search aren’t going to be the ones with the biggest link profile. They’re going to be the ones whose content actually holds up when an AI reads the whole thing and has to decide, on the merits, whether it’s worth repeating.
Data and analysis referenced in this episode draw on Kevin Indig’s research at Growth Memo, “What Content Works Well in LLMs?”, examining over 7,000 AI citations across 1,600 URLs, with supporting data from Profound.
COMPLETE THE FORM TO
BOOK A STRATEGY CALL
"*" indicates required fields
CHAPTERS:
COMPLETE THE FORM TO
BOOK A STRATEGY CALL
"*" indicates required fields
Subscribe and Share – WE APPRECIATE YOUR SUPPORT
