What does NOT work to get into Perplexity
Perplexity shows sources with every answer, so tricks aimed at invisible influence backfire here and clear, verifiable content works better.
Why Perplexity works differently than you might think
Perplexity is not a language model that answers from memory. At its core, it is a search engine that actively searches the web with every question, selects sources and summarizes an answer from them. PerplexityBot, the crawling robot that visits websites to cite them, is thus at least as important as the language model itself.
That difference is precisely why much of the advice you find online about visibility in AI does not apply one-to-one here. What works to be mentioned more often by ChatGPT, as we discuss in how your company gets into ChatGPT's answers, is sometimes useless for Perplexity and sometimes even harmful.
The biggest difference lies in source attribution. Perplexity shows where information comes from with almost every answer, with clickable links. That means tricks that rely on invisible influence of an answer have little chance here. A source that does not check out will either visibly not be used or, worse, be visibly misquoted.
Why keyword stuffing does not work
Filling a text with a phrase like best installer region Utrecht does not work with Perplexity. The system does not judge based on word repetition, but on whether a page clearly answers a question. Keyword density was already an outdated idea for regular search engines, and for an AI that reads and summarizes text it is even less relevant.
What does matter is whether your page gives a clear answer to a concrete question. Perplexity looks for passages it can literally take over or paraphrase. A long introduction full of persuasive language before you get to the point reduces the chance that passage is usable.
We see this pattern with ChatGPT too: unnecessary verbosity does not help and can even reduce the chance that an AI will cite your content, as explained in what does NOT work to get into ChatGPT. With Perplexity this is even more directly noticeable, because the system often incorporates literal pieces of text in the answer.
Why hidden text and invisible pages backfire
Some consulting firms suggest hiding text from search engines: white on white, or in code that people do not see but machines do. For Perplexity, this is a bad idea. PerplexityBot reads the page as it is technically constructed, and hidden text that does not match what a visitor sees does not inspire confidence as a source.
Because Perplexity displays sources prominently, recognizability is important. If someone clicks through from a Perplexity answer and finds a completely different page than what was cited, that is a bad signal. The risk is not only that you will not be used again, but that your site will be regarded as unreliable for future questions.
The same applies to pages that technically exist but are deliberately kept out of sight of crawling robots. Whether a robot like PerplexityBot is even allowed to visit your site depends on settings comparable to what is explained in GPTBot: what it is and whether you should allow it. Blocking sometimes seems safe, but also means: never be mentioned.
Why one-time action is not enough
Optimizing a website once and thinking you are done does not work with Perplexity. This system searches the web with almost every question anew, and often uses recent sources. A page that was last updated two years ago and has not been touched since loses relevance compared to competitors who stay current.
This resembles the question of how often a language model refreshes its knowledge, an issue that applies to other assistants as discussed in how often ChatGPT updates what it knows about your company. With Perplexity, the replacement rate of sources is even more directly noticeable, because the system actively searches instead of drawing from a fixed memory.
Companies that expect a one-time text adjustment to have lasting effect often end up disappointed. Prices, opening hours, services and examples change. If that information is not maintained, Perplexity will choose a different, more current source, even if it comes from a competitor.
Why fake reviews and exaggerated claims don't help
Because Perplexity visibly cites sources, exaggeration is risky. A page full of superlatives without substantiation, such as unmatched quality or market leader without any explanation, is not automatically believed by the system. Perplexity deliberately works with verifiable facts, and a claim that is not based on anything is more likely to be ignored than adopted.
Fake reviews or paid ratings don't work here either. If a claim on your site differs from what other, independent sources say, such as comparison sites, news articles or industry organisations, that other source often carries more weight. Perplexity places sources side by side and chooses what seems consistent and verifiable.
This touches on a question many entrepreneurs have: why a competitor gets mentioned and they don't. The explanation in why does ChatGPT mention your competitor and not you applies to a different assistant, but the core is the same with Perplexity: concrete, verifiable information wins over vague superlatives.
Why loose tricks without a readable site don't work
Some entrepreneurs try to apply specific technical interventions in the hope of outsmarting an AI, such as hidden metadata or special codes that supposedly give priority. Perplexity does not operate on the basis of such tricks. The system assesses the actual content of a page, not invisible markings that claim to be important.
More important is whether PerplexityBot can technically read your site well. A page that relies on heavy scripts to display text may appear empty to a crawling robot, even though a visitor sees content. This is exactly the same problem as with other AI systems, as explained in does ChatGPT read your website and how do you find out.
Without a technically accessible, well-structured site, no single intervention makes sense. A clear heading structure, factual text and a working sitemap contribute more than any secret trick. You build findability through basic work, not through tricks that claim to circumvent a system.
Frequently asked questions
Does it help to repeat as many keywords as possible on my page for Perplexity?
No. Perplexity assesses whether a text clearly and factually answers a question, not how often a word appears. Repeating keywords often makes a text less usable, because the answer becomes less direct. Better is a clear, concise paragraph that directly answers the question, so that passage is easy to cite.
Can I mislead Perplexity with hidden text or fake reviews?
That doesn't work well. Because Perplexity prominently shows sources and people can click through, a difference between hidden text and visible content is quickly noticed. Fake reviews also often carry less weight than independent, consistent sources that the system finds elsewhere. The risk is that your site is considered unreliable and is actually used less.
Is updating my website once enough to be used by Perplexity?
No, that is usually not enough. Perplexity re-searches the web on virtually every question and prioritizes current information. If your page is not maintained after a one-time adjustment, prices, services and examples become outdated, and the system is more likely to choose a competitor that stays current.
Further reading
How is your own website doing?
Two figures, within a minute, free of charge — and you don't need to leave anything behind.
Prefer the full report right away? Comprehensive scan — € 49 →
Want to know what's causing it?
This scan gives you the status. The report gives you the causes and the order in which you address them.
- For each finding, what's wrong and what needs to happenThe free scan gives you the status. The report gives you the list — in plain language, ranked by how much it matters, without you needing to understand anything technical.
- All four AI assistants instead of oneThe free result is a sample from one assistant. The report asks eight questions of all four, so you know whether the issue is with one assistant or all of them.
- Which companies do get mentionedWho gets the customer you're losing? Those names are in the report, with how often they appear where you're missing.
- A deeper measurement of your websiteThe free scan looks at ten pages. The report goes through up to fifty, including the pages where your customers ultimately end up.