HubSpot AEO Grader vs AEO Checker: Two Tools, Two Different Questions
Same URL, two different scores
If you've run a site through HubSpot's free AEO Grader and then found your way here, you've probably noticed the results don't line up cleanly with what we do. That's not a bug in either tool. HubSpot's Grader and AEO Checker are answering two different questions that look like one from the outside.
HubSpot's Grader asks: how does AI perceive my brand? It queries AI models — ChatGPT, Perplexity, Gemini — about your business and scores what comes back across five dimensions: Sentiment, Presence Quality, Brand Recognition, Share of Voice, and Market Position. It's asking the models what they already think, based on their training data.
AEO Checker asks a different question: is my site built so AI engines can find it, understand it, and cite it — and what exactly do I change? We crawl your actual site (up to 100 pages, depending on tier), check it against six technical readiness modules, and hand you paste-ready fixes.
Neither question is more legitimate than the other. But if you're a consultant trying to answer a client's "are we in ChatGPT?" question with something you can act on this week, it matters which one you're holding.
What each tool actually does
HubSpot's Grader doesn't crawl your website. By its own documentation, it queries large language models and scores the perception it gets back. There's no fetch of your homepage, no read of your robots.txt, no check of whether you have schema markup. It's measuring a symptom — what AI currently says about you — not the underlying cause of why AI does or doesn't cite you.
AEO Checker crawls your site directly. We pull your homepage, robots.txt, llms.txt, and sitemap.xml, then run six technical checks: structured data, content structure, entity and trust signals, AI crawler access, technical health, and llms.txt presence. Every check produces a PASS, WARN, or FAIL with a specific finding — not a perception score, a diagnosis.
The output is different too. HubSpot's report is a set of perception scores. Ours is a set of scores plus three ready-to-use assets: a filled-in llms.txt file, LocalBusiness/FAQPage JSON-LD schema markup, and step-by-step fix instructions for your three lowest-scoring modules. You can hand our PDF to a developer and they'll know exactly what to paste where.
The comparison table
Accurate as of July 2026. Sourced from HubSpot's own public product documentation. We re-verify this table quarterly — if any row changes, we update it or take it down.
| HubSpot AEO Grader | AEO Checker | |
|---|---|---|
| Crawls your site | No — queries LLM training data | Yes — up to 100 pages, depending on tier |
| Copy-paste fixes | No | Yes — llms.txt, LocalBusiness + FAQ JSON-LD |
| Six technical module scores | No | Yes |
| Email/account for full result | Email required | Email for free grade; PDF in your inbox within 5 minutes of purchase, no account |
| Next step after your score | $50/mo monitoring subscription | A one-time report you keep — no subscription |
| Best for | Watching brand perception over time | Fixing a site so it gets cited |
When HubSpot's tool is the right call
If what you actually need is continuous, multi-model tracking of how AI describes your brand over time — sentiment shifts, share-of-voice changes across ChatGPT and Perplexity and Gemini, month over month — that's a real, legitimate need, and it's not something AEO Checker does. We run a single scan. If your client's brand perception is actively shifting week to week and someone needs to be watching that dial, a subscription monitoring tool earns its keep there.
What we don't think holds up is treating a perception score as a diagnosis. A dropping Sentiment score tells you something is wrong. It doesn't tell you your site is missing FAQPage schema, or that GPTBot is blocked in your robots.txt, or that your llms.txt file doesn't exist. Perception and technical readiness are related — a site that's citable to begin with is more likely to be perceived well — but one is a symptom you re-measure and the other is a cause you fix.
Why this is worth billing for
If you're a consultant with a client asking "why don't we show up when someone asks ChatGPT about us," a monthly score that moves up and down isn't something you can put in an invoice. You can't bill a client for "your Sentiment score is 6 points lower than last month." You can bill for a report that says here's what's broken on your site and here's the paste-ready fix — llms.txt, schema, the works — delivered once, for $19.95 to $69.95 depending on how much of the site needs covering. No subscription, 30-day money-back guarantee, and it lands in an inbox in under five minutes.
Run the grade before you decide
You don't have to take our word for any of this. Run the free Grade Check on the site you're actually thinking about — the client's, or your own.
It scans one page and produces the same 0–100 score and letter grade as the paid reports, using the same scoring engine. Costs nothing, takes about a minute, and we ask for an email so we can send you the result. No card, no account, no subscription at the end of it.
Then decide with a number in front of you. If it comes back an A or a B, the technical side is largely handled — your problem is probably perception, and a monitoring tool is the honest next purchase. If it comes back a C or lower, something on the site is blocking AI engines from citing you, and a full report will tell you exactly what's wrong and hand you the fix. Either way you'll know which of the two questions you're actually trying to answer, which is more than you had before you ran it.