Why Your Website Isn't Getting Cited by ChatGPT (Even Though It's Indexed)

I checked our analytics again last week. Stackjot has 22 posts live, all indexed by Google, all in the Bing index, with ClaudeBot, OAI-SearchBot, and PerplexityBot all hitting the site weekly. We've added FAQ schema, allowed every AI bot, set up canonical tags, the whole AEO checklist.

Total ChatGPT citations in the past month: zero that I could find. Total Claude citations: one, maybe. Total Perplexity citations: a handful, all for the same two posts.

If you've done the same kind of work and seen the same kind of nothing, this post is the diagnostic I wish I'd had a month earlier.

The short version: getting indexed is not the same as being citable. Indexing is the gate. Citation is a different game played by different rules, and most of the AEO advice online stops at the gate.

What "citation" actually requires

When someone asks ChatGPT a question that triggers web search, here's what happens behind the scenes:

1. ChatGPT writes a search query for Bing
2. Bing returns its top 10 or so results
3. ChatGPT fetches a few of those pages
4. ChatGPT picks the chunks it'll quote and attaches citations

Notice steps 1 and 2. If you're not ranking in Bing's top 10 for the specific query ChatGPT generates, you're not in the pool. It doesn't matter how good your content is. You're not eligible.

This was the first thing I got wrong. I assumed that if my pages were "indexed" in Bing, they were candidates. Indexing means Bing has the page in its database. It does not mean the page ranks. A site can have 100 indexed pages and rank for none of them.

Open Bing Webmaster Tools. Look at Search Performance, not Index Coverage. If you have lots of impressions but zero clicks, you're indexed but not ranking. That's the position most new sites are in. That's the actual problem.

Reason 1: You're indexed but never the top result

When I tested this on stackjot, I searched Bing for the literal question "how do I rank in chatgpt" — a query our rank-in-chatgpt-and-claude post is specifically about. We were at position 38.

ChatGPT's search picks sources from positions 1 to maybe 10 most of the time. Sometimes a bit further if no early result is great. Position 38 is invisible to it.

This was depressing to find. But it explained everything. We weren't being filtered out for a content quality reason. We weren't being filtered out at all. We were just buried under 37 other results that Bing decided were more authoritative or more relevant.

The fix is unglamorous: build backlinks to the specific pages you want ChatGPT to cite, until those pages reach the first page of Bing for your target queries. Bing rewards backlinks more directly than Google does, in my experience. Two or three quality links to a specific URL can move it from page 4 to page 2 in a couple weeks.

Reason 2: Your content discusses the topic instead of answering the question

Even if you rank, you can still get skipped at step 4.

I tested this with two of our posts. Both rank on page 1 of Bing for similar queries. One gets cited occasionally in Perplexity. The other never does. The difference:

The cited post opens like this:

AEO stands for Answer Engine Optimization. It is the practice of structuring your content so that AI answer engines extract from it and cite it in their generated answers.

The uncited post opens like this:

For 18 months now, the marketing world has been split into two camps about AI search. One camp says SEO is dead...

Both are decent paragraphs. The first one is a literal answer to the literal question. The second one is an essay opener. When ChatGPT scans a page looking for something to quote, the first one is a layup. The second one requires the model to do work — pull pieces from later in the post, stitch them together, decide what to attribute to you.

ChatGPT does the work sometimes. Most of the time it goes with the layup.

The lesson: every page meant for AI citation needs the literal answer to its title within the first 100 words. Not the setup for the answer. Not the framing. The answer. Then you can write the rest however you want.

I went back and edited our AEO vs SEO post to front-load a tighter definition. It's too early to know if that changes citation rates, but the principle is solid.

Reason 3: Your FAQ questions aren't questions people ask

This one took me a while to see.

FAQ schema is great. Every AEO guide tells you to add it. We did. But the questions you put in your FAQ should be the questions real people type, not the questions that sound smart for the topic.

When I looked at our FAQ on the AEO post, the first question was "What is the main difference between AEO and SEO?" That's a reasonable question. It's not the question people actually type. The question they actually type is "is AEO the same as SEO" — which has the same intent phrased the way people talk.

ChatGPT seems to prefer FAQs where the question matches the user's query closely. Reasonable. So if you write your FAQs to match marketing-style headlines instead of real search phrases, you're handing your competitors the citation.

I started rewriting our FAQs to match queries pulled from Google Search Console and from typing variations into ChatGPT itself ("what do people ask about X"). The questions get less elegant. That's the point.

Reason 4: Your site is too new to have trust signals

This is the hardest one to fix because there's no shortcut.

ChatGPT, like every search system, has signals it uses to pick between candidates when several pages look similarly relevant. Some of those signals reward established sites:

• How long the domain has existed
• How many other reputable sites link to it
• Whether the domain has been mentioned in major publications
• Whether the brand name is recognized

For a six-month-old site like ours, all of these are bad. There's no amount of FAQ schema that compensates for a domain Google calls "discovered" and ChatGPT has barely heard of.

The only thing that builds these signals over time is the boring SEO work nobody wants to do: backlinks from real sites, guest posts on existing publications, occasional brand mentions in newsletters and podcasts, basically anything that builds the kind of reputational footprint that takes a year to look real.

I keep wanting this part to be a hack. It is not a hack. It is a year of small consistent work.

Reason 5: You're competing with sources ChatGPT already trusts

For any query in a popular niche, ChatGPT has a small set of sources it tends to default to. For tech and software topics, it's often Stack Overflow, GitHub docs, Wikipedia, and the major publications (TechCrunch, The Verge, etc.). For productivity and AI tools specifically, it's often Zapier's blog, Notion's docs, and a handful of established creator sites.

If your query overlaps with what one of those sources covers well, beating them is hard. They have years of authority signals and ChatGPT has been "rewarded" for citing them many times. It takes a meaningful quality gap to dislodge a default citation.

This isn't impossible. It is a longer game than the AEO blogs imply. Sometimes the right call is to pick queries the default sources don't cover well — newer topics, narrower niches, more specific phrasings — and own those first. Then expand from there.

A 4-step diagnostic you can run today

If you're in the same spot we were, here's what I'd actually do, in order:

Step 1: Check Bing rankings, not Bing indexing. Go to Bing Webmaster Tools, Search Performance. Find the queries where you have impressions. Look at the position column. If everything is past position 20, your problem is ranking, not eligibility. Focus on backlinks.

Step 2: Test your top 3 target queries directly in ChatGPT. Ask ChatGPT each query five times with different phrasings. Write down which sources it cites. If your site never appears, you're not in the candidate pool for those queries. If it appears sometimes, you're marginal. The pattern tells you what's wrong.

Step 3: Audit your opening paragraphs. For your top 5 posts, check whether each one's first 100 words contains the literal answer to the title. If not, rewrite the opening. This is the cheapest fix that actually moves the needle.

Step 4: Rewrite your FAQ questions to match real queries. Pull queries from Google Search Console. Phrase them the way users typed them, not the way a marketer would write them. Use those as your FAQ questions, with tight 2-4 sentence answers.

Each of these takes an hour. None of them require backlinks or waiting for time to pass. Doing all four on a 20-post site is a weekend of work.

What I'm still uncertain about

I don't know how heavily ChatGPT weights freshness for evergreen queries. I have a hypothesis it matters more than I assumed, but my data is too small to be sure.

I don't know whether updating an old post's dateModified to today actually re-enters it into ChatGPT's candidate pool, or if the underlying Bing crawl has to happen first. The order of operations is unclear from outside.

I don't know how much ClaudeBot's crawl frequency correlates with Claude citation rate. We see ClaudeBot hit certain pages 10x more than others. The high-crawl pages aren't the high-citation ones. I have no clean theory for why.

I'm noting these so you don't take this post as the complete picture. The AEO field is too new for anyone to have all the answers. Anyone who claims they do is probably selling something.

What I'd skip

A few things I tried that didn't seem to help:

• Adding more pages to the FAQ schema (we capped out at 5-6, more didn't move citations)
• Using HowTo schema on procedural posts (no observable effect)
• Adding dateModified updates to every post weekly (didn't change citation rate)
• Submitting URLs to IndexNow more than once (one submission is enough, repeats don't accelerate)

These might work for other sites. They didn't move our citation rate measurably. Save the time for backlinks and content rewrites.

The thing nobody says

AEO is mostly SEO with extra steps that matter more than people realize and a few new steps that matter less than the gurus claim. If your site can rank in Bing and your content opens with a direct answer to its title, you've done 80% of what's necessary. The other 20% is mostly time.

If you're new and ranking badly and not being cited, the order of operations is: fix ranking first, fix opening paragraphs second, fix FAQs third. Everything else is a rounding error until those three are in place.

We're still in the part of the curve where citations are sporadic. I'll update this post in a few months when we have more data. If you're working on the same problem, I'd love to know what you're finding. Sometimes the only way to figure these things out is to compare notes.

---

If you want the broader strategic picture, the AEO vs SEO breakdown covers how the disciplines fit together. If you want the tactical layer for getting cited at all, the ranking in ChatGPT and Claude guide is the prerequisite reading for this post.