Most articles on "how to get cited by ChatGPT" are recycled SEO advice with the keyword swapped. They tell you to write "helpful, authoritative content," build backlinks, and implement schema — the same playbook that was written for Google in 2019.
That advice isn't wrong, exactly. It's just missing the point. AI citation doesn't work like ranking.
Over the last year we've tested what actually moves citations for B2B SaaS brands — entity work, content rewrites, different citation sources, schema variations, outreach patterns. A lot of things that feel important don't move the needle. A few things that feel unglamorous move it a lot. This is the honest version.
The one thing nobody tells you about ChatGPT citations
ChatGPT doesn't "rank" your page. It doesn't pick the best-written article. What it does is retrieve a set of source snippets based on its search and internal knowledge, then synthesize an answer that reflects the consensus across those snippets.
The brands cited aren't the ones with the best blog posts. They're the brands AI has built a strong internal representation of — because they appear consistently across many trusted sources, described in compatible ways.
This matters because it means publishing more content isn't the fix. If the underlying entity is weak, great content just sits there. Fix the entity first.
Start with the knowledge graph, not the blog
When we audit new clients, we almost always find the same thing: the company is invisible or barely visible in Google's Knowledge Graph. No panel. Sparse Wikidata. Inconsistent descriptions across directories. Google doesn't really know what they are.
That matters because ChatGPT, Perplexity, and Google AI all lean on structured entity data during retrieval and synthesis. If you're not a clean entity, you're a weak candidate for citation — no matter how many backlinks you have.
The fix isn't dramatic. It's mostly unsexy cleanup:
• Audit every place your brand appears (crunchbase, LinkedIn, G2, product databases, Wikidata) and make sure the descriptions are consistent and current. • Implement Organization schema with accurate sameAs links to every canonical profile. • File a Wikidata entry if you qualify (most Series A+ SaaS do). • Optimize your Google Business Profile even if you're SaaS — it feeds the knowledge panel.
None of this feels like marketing. But it's the highest-leverage work we do, by a wide margin.
Answer the specific questions, not the category
Here's the mistake most content teams make: they write for the category. "The Ultimate Guide to Data Observability." "Everything You Need to Know About CI/CD."
ChatGPT doesn't want your guide. It wants the answer to the exact question the user asked. And the user didn't ask "tell me about CI/CD." They asked something like "what's the best CI tool for a small team using a monorepo."
So the content that gets cited isn't the comprehensive overview. It's the specific-question page that answers a specific buyer question in the first paragraph, then gives the supporting reasoning.
Rewrite your top 20 buyer questions as standalone pages, each with the answer in the first 2–3 sentences. Yes, this feels weird. It's supposed to. That's the format AI engines reward.
Third-party citations do more work than you think
Here's something counterintuitive: a single mention of your brand in a well-regarded third-party publication often outperforms ten blog posts on your own site for AI citation purposes.
AI engines don't just read your content. They build their picture of your brand from everywhere your brand is discussed. If Dark Reading, TechCrunch, or InfoWorld describes your product in a specific way, that description weights heavily when the AI decides whether and how to cite you.
This is why earned media matters so much for AEO, and why most SaaS companies are underinvesting in it. A thoughtful expert-contribution program or a piece of original research that gets referenced externally is worth more than a huge internal content output.
Schema is table stakes, not a growth lever
People ask us about schema a lot. The honest answer: you need it, but it's not going to move citations by itself.
Organization schema, Product schema, FAQ schema — implement them properly and you're giving AI crawlers a cleaner signal. But if your entity is weak and your citations are thin, adding schema won't rescue you. It'll just make the weak signals slightly cleaner.
Treat schema like plumbing. Get it right, forget about it, move on to what actually moves the metric.
What to measure
You need two numbers, tracked weekly:
First, citation frequency across your top 30–50 buyer queries. Run them manually (or with tooling — there are options now) against ChatGPT, Perplexity, and Google AI. Track how often your brand appears.
Second, share-of-voice versus your top 3 competitors. This matters more than absolute count, because AEO is a relative game. You're trying to displace somebody.
Everything else — referral traffic from AI, pipeline attribution, downstream revenue — is useful but noisy. Citation frequency and share-of-voice are the leading indicators that give you something real to optimize against.
Timeline: be realistic
The best case we've seen is meaningful citation gains within 60 days, and that's when everything was already primed (good content, decent entity setup, just needed citation-building push). More typical is 3–4 months to see the needle move.
For companies starting from a weak entity baseline, expect 6+ months before the work really compounds. That's not pessimistic — it's realistic. AI engines update their internal representations slowly, and the work is cumulative.
If someone promises faster results, ask what they're actually measuring. "Appearing in Perplexity once" is not the same as "being consistently cited across major AI engines for your highest-intent queries."
AEO is new enough that a lot of what's being written about it is speculation. What we've laid out here is based on actual client work and actual citation tracking over the last year — which is not a long time, but it's what we have. The landscape will keep shifting. What works today may not work as well in 12 months.
If you're trying to figure out whether to invest in this for your brand, the first step isn't reading more articles. It's auditing where you stand right now, with real queries your buyers actually type. That's what our free audit does. We're happy to run one for you.