Perplexity automatically generates articles about your industry and picks a narrow set of sources to cite — ignoring everything else. If your site isn't part of that selection, then for anyone using Perplexity you simply don't exist, no matter how much work you've put into SEO. Your competitors who are in it get recommended every day to potential customers who will never see your name. Understanding the selection criteria is the key to getting in.
Perplexity Pages is a new type of AI-native content, where the platform generates articles by citing sources. If you’re not among those sources, you don’t exist.
This is the shift in perspective you need to take home today. Because continuing to think of Perplexity as “a search engine with a chat on top” is the most common mistake I see among entrepreneurs and marketing managers. Perplexity is a machine that produces answers and content starting from selected sources: your only game is to make it into that selection.
And the criterion by which Perplexity chooses sources is not Google’s. It doesn’t reward historical depth, it doesn’t reward the oldest domain, it doesn’t even reward the most complete content. It rewards three things together: domain authority, content freshness, and how precisely the source answers the semantic question.
What “source selection” means for a system like Perplexity
In the world of research on retrieval-augmented systems — the ones that first search for sources and then generate the answer — the documented mechanism is always the same: the model knows nothing on its own, it picks N sources from an index and processes the content starting from those sources. The quality of the answer depends entirely on the quality of the selection.
Perplexity hasn’t published technical papers on its own selection algorithm, so here we’re on an explicit deduction and not on a documented fact. From what can be observed in the product’s behavior and from the general principles of retrieval, the selection weighs three signals:
- Domain authority: how often that domain is cited as a reference in other authoritative documents in the field.
- Freshness: the date the content was last updated, not the date it was first published.
- Semantic matching: how well the page’s content answers that specific phrasing of the question.
From this follows something I repeat to my clients every week: your “historic” 4,000-word page published in 2022 loses to a page updated yesterday by a competitor who has half your content but fresh data.
Why freshness matters more here than elsewhere
On Google, a well-linked evergreen piece of content can hold its position for years. Not on Perplexity. The logic is different because the use case is different: whoever queries Perplexity wants the best synthesis available today, not the best historical archive.
If in earlier articles in this series I’ve already told you how E-E-A-T for AI builds the authority signal over time, here I’m adding the missing piece: authority alone isn’t enough if the content isn’t refreshed. An authoritative brand with content untouched for two years gets overtaken by a less authoritative brand that does update.
And if you’ve read what I wrote about backlinks as a citation proxy, the mechanism here overlaps: backlinks get you into the pool of candidate sources, but within that pool the winner is whoever has the most recent timestamps on content that answers the query.
Changing a comma to “rejuvenate” the date doesn’t work.
The test you can run right now in 20 minutes
Open Perplexity and run three queries about your industry, phrased the way one of your customers would. Look at the cited sources on the right. For each source, click through and look for the last-updated date.
Let me give you a concrete example with an industry I know well because I’ve worked on it for clients: Sardinian sheep farming and Pecorino Romano PDO. If on Perplexity you ask “difference between Pecorino Romano PDO and Pecorino Sardo PDO,” look at the sources it cites. In 90% of the cases I’ve checked over the past few months, the cited pages have an update within the last 12 months. The consortia that update the product page only when the production specification changes — that is, every 5-7 years — simply don’t make it in.
Then open Google Search Console on your site. Filter for your strategic pages. Check when you last updated them. If the answer is “more than 6 months ago” on the pages covering your business queries, you’ve found the reason Perplexity doesn’t cite you.
A simple, ternary decision threshold:
- Updated within the last 3 months → you’re competitive.
- Updated between 3 and 9 months → you’re in the gray zone, it depends on the industry.
- Updated more than 9 months ago → you’re dropping out of the selection.
Sign every page with a real, verifiable author.
What I’ve observed over these months
Over the last 8 months I’ve tracked Perplexity’s behavior on niche queries for a group of companies in the Sardinian dairy sector, between Nuoro and the Barbagia hinterland. Longitudinal observation, not an instrumented test: every 3-4 weeks I looked at which sources Perplexity was citing on about thirty recurring queries tied to Pecorino Romano PDO, extensive sheep farming, transhumance, and ewe’s milk traceability.
The pattern I saw repeats fairly cleanly: every time a producer or a consortium published a piece of news, a press release, an update to the production specification, or a refreshed product page, within 2-4 weeks that source started showing up in the citations. If the page was then no longer touched, within 4-5 months it slipped out of the citations even when the domain was historically authoritative.
An honest limitation of this observation: thirty monitored queries are not a scientific sample, and Perplexity’s behavior changes over time with product updates. The pattern, however, was clear enough to make me tell clients in the sector one specific thing: update something meaningful on your key pages at least once a month, even if it’s just a paragraph of context or a vintage-year figure.
This is an entry-level check. The real analysis, with systematic tracking of AI citations over time, requires professional monitoring tools that go beyond what you can do by hand.
The mistakes I see made most often
The cosmetic update. Changing a comma to “rejuvenate” the date doesn’t work. The systems that crawl and index for AI look at the content diff, not just the timestamp. A client who owns a dairy near Nuoro had asked me whether touching the date was enough: the answer is no, you need new substance — a figure, a vintage year, a paragraph of context.
The obsession with the 4,000-word article. In the Perplexity world, two 800-word updates over 6 months beat a 4,000-word monolith that’s been sitting still for a year. I write this because I still see marketing managers invest three weeks in a “pillar content” piece and then leave it there.
The orphaned press release. Many producers publish industry news on specialized portals but don’t republish it on their own site. Result: the portal gets cited, you don’t. Freshness has to be brought home to your own site.
Ignoring the authority of who writes. If I’ve already told you how much author entity recognition matters, I’ll repeat it here in Perplexity terms: pages signed by people with an established identity carry more weight than anonymous pages. All the more so in industries with a strong artisanal component like PDO products.
What you can do on your key pages
- Identify the 8-12 pages on your site that match real business queries (use Search Console to find them).
- Schedule a substantial update of each one at least every 30-45 days: a new figure, an updated paragraph of context, a Q&A section on customers’ recurring questions.
- Sign every page with a real, verifiable author.
- On Perplexity, query by query, compare which 3-5 competitors get cited in your industry: study their update frequency.
- Publish industry news on your own site first, then elsewhere.
Where this fits in the bigger picture of AI visibility
What matters, in the end, is always the same thing: being present in the answers that AIs generate for your customers. Perplexity makes it explicit because it shows you the sources on screen, but the same mechanism — fresh, authoritative sources beating old, deep ones — you’ll also find in a more hidden form on ChatGPT Search and Gemini.
In the upcoming articles in this series you’ll see how the other answer engines behave on similar criteria but weighted differently, how Perplexity Pages can be “cultivated” as an editorial format, and how to structure a monthly update flow that keeps your pages inside the selection radar without becoming a second job.
The game of visibility in AI answers isn’t played on quantity. It’s played on rhythm.
