FixAEO MCP: the AI-search question no other tool can answer

I built the FixAEO dashboard. I still barely open it anymore.

Here's the ritual I kept catching myself in. I'd be deep in a conversation with Claude, planning a piece of content or prepping for a call, and I'd need one number: how are we doing in AI search this week? So I'd stop, open FixAEO in another tab, find the number, copy it, and paste it back into the chat. Every time, the same tiny detour. Leave the conversation, go fetch my own data, come back.

At some point the obvious question landed: why am I leaving the room to get something that's already mine? The interface to your data is quietly becoming the assistant you're already talking to. You shouldn't have to open a dashboard to ask "how's my AI visibility?" — you should just ask.

So we shipped an MCP server. And it can answer one question that nothing else on the market can.

So what is an MCP server, in plain English

MCP is the Model Context Protocol — an open standard that lets your AI assistant connect to a live data source and read from it while you work. That's the whole idea. Instead of you copying data into the chat, the assistant reaches out and pulls it itself, on demand, in normal language.

You don't have to be technical. If you use Claude, Claude Code, Cursor, ChatGPT, or any MCP-compatible client, you connect once with an API key and then just ask questions. No SQL, no exports, no second tab.

That's the bridge. Here's the point.

The wedge: where you rank on Google but AI never cites you

There's a gap most brands can't see, and it's the one that's costing them right now.

You rank on page one of Google for a query. Good. But when someone asks ChatGPT or Perplexity that same question, your brand isn't in the answer. Google sends you traffic; the AI engines quietly recommend someone else. You're winning the old game and losing the new one, and nothing in a normal analytics stack tells you where.

That's the rank-but-not-cited gap: the queries where you have real organic authority but zero AI presence. It's the highest-leverage list in AEO, because you've already earned the ranking — you just aren't being cited for it yet.

Now you can ask your assistant, in plain words: "Where do I rank on Google but AI never cites me?" And it hands you the list, pulled live from your own data.

I'll say it flatly: no other tool's MCP can answer that question. Not because they haven't gotten around to it — because of how the data is built.

Why FixAEO can answer it, and nobody else can

Two things have to be true at once, and most tools only have one of them.

First, breadth. FixAEO tracks eight AI engines, not one or two: ChatGPT, Claude, Gemini, Grok, DeepSeek, Perplexity, Copilot, and Google AI Overviews. "Are you cited in AI" is a meaningless question if you only look at a single model. You need the whole surface. You can see how we think about that spread in our rundown of the AI search engines.

Second, the fusion. We join Google Search Console rank data with that AI-citation data. Rank on its own is half a picture. AI citations on their own are the other half. Put them side by side and the gap falls out on its own — the queries where the rank is high and the citations are missing.

That fusion is the moat. A tool that only watches AI answers can tell you where you're not cited. A tool that only watches Google can tell you where you rank. Only a tool that holds both can tell you where those two facts disagree — and that disagreement is the whole opportunity. It's why the MCP can answer a question a single-source tool structurally can't. (New to the category? Start with what AEO actually is.)

What you can ask it today

The rank-but-not-cited gap is the headline, but it's one of fourteen read-only tools. A few of the questions I reach for most, in the assistant, in plain language:

• "Where do I rank on Google but AI never cites me?" — the gap, ranked.
• "Which sources cite my competitors but not me?" — the exact third-party pages feeding their citations, which doubles as an outreach list.
• "How did my AI visibility move this week across all eight engines?" — the weekly read, without the tab-switch.
• "Which AI bots crawled my site this week?" — GPTBot, ClaudeBot, PerplexityBot and the rest, so you can see the supply side of getting cited.
• "Write a 150-word AI-search update for my CMO." — your numbers, translated into plain business language, in the assistant's own words.

The nice part is that last one. The answer comes back in your format, in the voice of whatever you're already writing in — not as a dashboard export you then have to reword.

It's read-only, and it only sees your data

Worth being clear, because you're handing this to an AI: the MCP server is read-only. It reads your data. It doesn't write anything, it can't trigger scans, it can't spend anything on your behalf.

You connect it with an API key you create and can revoke at any time, and it only ever sees your own account. Nothing else. It's safe to hand to your assistant, which is exactly the bar it needs to clear.

How to connect

Connecting takes a couple of minutes. The endpoint is https://api.fixaeo.com/api/mcp, you authenticate with a bearer API key, and it works with Claude, Claude Code, Claude Desktop, Cursor, ChatGPT, and any MCP client.

Rather than reproduce a full tutorial here, I keep the copy-paste config for each client on one page: set it up on the MCP page. Create a key, paste it into your client, and ask your first question.

What's free, and where the MCP sits

Let me be straight about pricing, because it's the part most launch posts fudge.

FixAEO's free scan stays free. The free AEO tools stay free. What I'm not going to do is tell you the MCP is free on every plan — it isn't, and I'd rather say so than bury it.

The MCP is where the multi-engine data and the rank-but-not-cited gap become callable from your assistant, and that's a paid feature. The honest reason is cost: pulling live citation data across eight engines and fusing it with Search Console is the expensive part to run. So it lives on a paid plan. No enterprise-only gate, no fake countdown — just the real line.

If all you want is a first look at how the AI engines answer for your brand, run a free scan. If you want to live inside that data from your assistant, that's the paid layer.

Where this goes next

Today it reads. The natural next step is writing back — queuing an action or a tracked prompt from the conversation instead of just pulling numbers out of it. I'm not going to put a date on that, because I don't like promising roadmap I can't stand behind.

But the direction is the thing I actually believe. The dashboard isn't going away. More and more, though, the answer should come to you when you ask — in the tool you're already working in, in your own words. That's what this is a first step toward.

If you're on a paid plan, you already have everything you need. Connect it, and ask it where you rank but don't get cited. That list is where your next month of AEO work is hiding.

FAQ

What is an MCP server?

An MCP (Model Context Protocol) server exposes a data source to AI assistants through an open standard, so the assistant can read from it live while you work. The FixAEO MCP server exposes your AI-search visibility data — citations, competitor gaps, and the rank-but-not-cited gap — as tools your assistant can call in plain language.

Do I need a paid plan to use the FixAEO MCP?

Yes. The MCP server is a paid feature. Free scans and the free public tools stay free, but the MCP — where your multi-engine data and the rank-but-not-cited gap become callable from your assistant — is on a paid plan.

Which AI assistants work with it?

Claude, Claude Code, Claude Desktop, Cursor, and any MCP-compatible client work today using your API key as a bearer token. ChatGPT works through a custom connector on plans that support developer mode. Setup for each is on the MCP page.

Is my data safe?

Yes. The server is read-only — it can't start scans, edit anything, or spend credits. It authenticates with an API key you can revoke at any time, and it only ever returns your own account's data.

What makes the rank-but-not-cited gap unique to FixAEO?

It needs two data sets fused together: AI-citation data across eight engines, and Google Search Console rank data. Tools that watch only AI answers, or only Google, hold half the picture each. FixAEO holds both, so it can surface the queries where you rank well on Google but aren't cited in AI answers — a question a single-source tool can't answer.