A prospect opens Perplexity and types "best B2B onboarding platforms for fintech." Three seconds later they have four vendors, a sentence on each, and a set of citation links. They click one, book a demo, and never run a Google search.
You already know the stakes — being absent from that answer is the new version of being on page two. What you probably don't have is a concrete list of what actually moves the needle. This is that list: how ChatGPT and Perplexity decide who to cite, and the specific things you can change on your site to become one of the citations.
How ChatGPT and Perplexity actually pick sources
The two engines don't work the same way, and the difference changes what you optimize for. Perplexity is a retrieval engine: for most queries it searches the live web, reads the top results, and synthesizes an answer with citation links back to the pages it used. ChatGPT answers from its training data by default, but when it browses (or via SearchGPT) it retrieves and cites live pages much like Perplexity does.
The practical implication: to show up in Perplexity you need to be in the live results it retrieves for a query and be clearly extractable once it lands on your page. To show up in ChatGPT you need both a strong presence across the sources it was trained on and clean, crawlable content for when it browses. Neither engine rewards keyword density or backlink volume the way classic SEO did — they reward clarity, structure, and being mentioned in the right places. We unpack the underlying shift in why your B2B website isn't showing up in AI search.
Make your pages extractable, not just readable
An answer engine cites the sentence it can lift cleanly and attribute with confidence. That means the first sentence under each heading has to work as a standalone answer — no throat-clearing, no rhetorical question, no "in today's fast-moving landscape." State the fact, then support it.
Structure the page so a model can navigate it: H2 headings phrased the way a buyer would ask the question, short paragraphs, and a clear hierarchy. A wall of marketing copy is hard to extract from; the same content reorganized with direct topic sentences becomes citable. This is the same discipline that separates the sites that get cited from the ones that don't — the difference between AEO and traditional SEO.
Get mentioned where the engines look
Your own site is only half the game. When Perplexity answers "best [category] for [market]," it frequently retrieves third-party pages — roundup listicles, review sites, comparison articles, Reddit threads — and synthesizes from those. If your company isn't named in that ecosystem, you can be absent from the answer no matter how good your homepage is.
So the work extends beyond your domain: earn placement in the "best [your category]" roundups, keep your review-site profiles current and specific, and publish comparison content that names competitors honestly. Every credible third-party mention is another source the engine can pull your name from. This is slower than editing your own pages, but it's often the missing piece when a competitor keeps appearing and you don't.
Fix the rendering problem before anything else
If your content only exists after JavaScript executes, most AI crawlers see nothing. I learned this on our own site: I rebuilt brandinglab.io on a modern SPA stack, and within weeks our AI-search traffic vanished — the crawlers were pulling an empty hydration shell and moving on. The fix was serving a fully rendered version of each page to bots.
Before you invest in content, confirm the engines can actually read you: view your page source (not the rendered DOM) and check the real text is present in the initial HTML. If it isn't, that's the first thing to fix — everything else is wasted until the content is visible. For React and other SPA stacks, the schema and rendering patterns we ship cover exactly how to do this.
Add the structured data that tells models what you are
Schema markup gives an answer engine a machine-readable statement of what your company is and what each page covers. At minimum: Organization schema site-wide, Product or Service schema on those pages, FAQPage schema where you answer questions, and Article schema on posts. It's the difference between a model inferring what you do and knowing it because you declared it in a format built for parsing.
Structured data won't rescue thin content, but on strong content it materially improves how confidently a model can cite and describe you. Pair it with topical depth — a cluster of related pages around each core buyer question — and you're signaling authority in the way these systems actually assess it. To gauge where you stand, run yourself through the AEO maturity model.
Measure whether it's working
There's no rank to track in AI search — you're either named in the answer or you're not. So the measurement is simple: list your ten most important buyer queries, run each one in ChatGPT and Perplexity monthly, and record whether you're cited, how you're described, and which competitors appear alongside you.
That log is your scoreboard. When a change to your site or a new third-party mention moves you into an answer you were absent from, you'll see it — and you'll know what to do more of. Continuous monitoring is also where an AEO monitoring service earns its keep, tracking presence across engines so you don't have to check by hand.
Where this fits in the bigger picture
Most of these moves — extractable content, schema at the component level, crawlable rendering, topical clusters — are hard to bolt onto a legacy site and straightforward to build into a new one. That's why AI-search visibility is one of the clearest signs you've outgrown your current website, and why we treat it as a design-time decision in every B2B website rebuild. We did exactly this for Opsi Systems — an enterprise transportation-tech rebuild engineered for credibility and AI-search visibility from the ground up.
Want to know where you stand in ChatGPT and Perplexity right now? Book a free AEO audit with BrandingLab. We'll run your buyer queries, show you who's getting cited instead of you, and map the specific changes that get you into the answer.