HubSpot AI Search Grader vs HubSpot AEO: a free one time grade, and what the paid product is actually metered on
HubSpot answers its own comparison on its own page, and the honest reading is that the free tool and the paid one ask two different questions about the same three engines.
Published by AI Knows Us · Updated 20 October 2026
HubSpot settles this comparison itself, in a line on the Grader page worth quoting before anything else: "The Grader answers how does AI represent my brand right now? HubSpot AEO answers how is that changing, and what should I do about it?" So the free tool is a one time diagnostic with no second reading in it, and the paid product is the same three engines on a repeating cadence against prompts you specify. The useful question for a buyer is not which is better. It is whether a single dated snapshot is enough to act on, and for most businesses the answer turns on one thing: a score with nothing to compare it against cannot tell you which direction you are travelling in. Everything below was read from hubspot.com/ai-search-grader on 20 October 2026, HTTP 200 to a Chrome on Windows agent and to a Safari on macOS agent, body saved.
The two products, from the one page that describes both
| Read 20 Oct 2026 | AI Search Grader | HubSpot AEO |
|---|---|---|
| How often it runs | Once. Its own words are "free, one-time check" and "free one-time diagnostic" | Continuously. Its own words are "week over week" |
| Engines named | GPT-5.4 mini from OpenAI, Perplexity, Gemini | ChatGPT, Perplexity, Gemini |
| What you supply | Company name, location, industry, product or service | Specific prompts, per its own description |
| What comes back | A composite score out of 100, a five dimension breakdown, and a written interpretation | Visibility over time, competitor comparison, and prioritised recommendations |
| What stands between you and the result | No account, and a form. Its own instruction is "Complete the form to access your full AI search report" | A subscription, or two Marketing Hub tiers that include it |
| How it is sold | Free | One flat published monthly rate, buyable with no HubSpot subscription, or included in Marketing Hub Pro and Enterprise |
We record what a vendor publishes about price and do not restate the figure, which is a position two of our own pages already defend. HubSpot does print a monthly number on this page, it is a single flat rate rather than a ladder, and the two things worth noting about it are that it does not require you to buy the rest of HubSpot and that it is also folded into two Marketing Hub tiers. Go and read the figure on HubSpot's page, where it stays current.
What one reading can and cannot be used for
A single dated score is a strong instrument for three jobs and a weak one for a fourth.
- It establishes that a gap exists. A thin result across Presence Quality and Brand Recognition tells you the engines hold little about you, and that is a finding you can take to a budget conversation.
- It names the dimension to work on. Sentiment at 40 points, Presence Quality at 20, Brand Recognition at 20, Share of Voice at 10 and Market Competition at 10 are published weights, so the breakdown points at the specific weak block rather than at a vague total.
- It reads a rival. The page says the tool accepts any brand, so you can grade the two companies the engines already name in your category and compare their breakdown with yours on the same day, in the same instrument.
- It cannot tell you whether anything you do next worked. Improvement is a difference between two readings taken the same way, and one reading has no difference in it. That is arithmetic rather than a criticism of the tool, which calls itself a snapshot on its own page.
The fourth point is the one that quietly costs money, because a second grade run six weeks later looks like a before and after and is not necessarily one: the engines move, the model variant behind the reading can change, and a composite out of 100 does not carry the sample size that would let you tell a real move from a wobble. Whether a change in an AI visibility number is a result or a wobble is worked through in AI visibility improvement or noise, which had no durable inbound link from any body on this site when we counted on 20 October 2026, in a count of 3,456 internal links across 2,050 stored bodies against 1,960 live pages where 911 pages, 46.5 per cent, carry none.
The upgrade path is the design, and that is fine as long as you can see it
The Grader page points at the paid product in six separate places, including inside the method steps and inside four of its own FAQ answers. A free diagnostic that leads to a paid monitor is an ordinary and honest way to sell software, and HubSpot does it in the open: it describes the two products as answering different questions rather than pretending the free one is complete.
What a buyer gets from knowing this is a better sequence. Run the Grader, read the five dimension breakdown, and then decide whether your next spend goes on monitoring or on the thing the breakdown pointed at. A thin Presence Quality score is usually telling you that the pages an engine reads about your category do not mention you, and no amount of weekly monitoring changes that by itself. Monitoring earns its place once there is something moving to watch.
Where each of these beats us, and where we are ahead
We sell a product in this category, so the ordering here is a stated criterion rather than an opinion: what is published about the measurement, and which surfaces it reaches.
- HubSpot publishes its scoring weights and we do not publish a comparable rubric. Five dimensions summing to 100 is a document you can argue with before you run the tool.
- HubSpot names the exact model variant it reads. GPT-5.4 mini is more precise than almost anything published in this category, ours included.
- HubSpot AEO is bought without the suite. Its page states that no HubSpot subscription is needed, which is not true of every platform module in this market.
- We reach five surfaces where both HubSpot products reach three. Our pricing page, read 20 October 2026, lists ChatGPT, Gemini, Perplexity, Claude and AI Overviews on every plan that checks AI apps, and names the unit: one credit is one buyer question on one app once, with a Claude check costing three.
- Our question never carries your name. That is the difference in reading, not a difference in quality, and it is why the two numbers are not interchangeable.
- HubSpot publishes a certification and we publish none. Its Academy runs an Implementing AEO certification, linked from the Grader page.
- We publish one login per account and no SSO. Read on 20 October 2026, our own pricing page states one full access login on every plan and carries no SSO, no SAML and no white label line.
Questions about the free grade and the paid monitor
Should I run the Grader before buying anything?
Yes. It is free, it is finished in under two minutes on its own stated timing, and the five dimension breakdown gives you a vocabulary for the problem. Run it on yourself and on two rivals in the same sitting.
Does the free grade include a trend?
No, and HubSpot says so. It is a one time diagnostic and the trend is what the paid product is for. Two separate grades months apart are two snapshots rather than a tracked line, because nothing holds the question set still between them.
Do I need the rest of HubSpot to buy AEO?
Its page states that the monitoring product needs no HubSpot subscription, and also that it is included in Marketing Hub Pro and Enterprise. So there are two routes in and the right one depends on what else you already pay HubSpot for.
Is three engines enough?
It depends on where your buyers ask. Three covers a large share of consumer and B2B question volume, and it leaves out Claude and Google AI Overviews. Write down which apps your last ten enquiries actually mentioned, and buy against that list rather than against the longest list you are offered.
What should I ask HubSpot before subscribing?
Two things that no page answers. What happens to the history behind a prompt when you edit its wording, and whether the answer text and the cited sources come out of the product or only the score does. The source list is the part you will still be using in six months.
Can the Grader score be put in a board pack?
A single number out of 100 with a published rubric beside it is defensible as a starting position, and it should be dated and labelled as one reading on three named engines on one day. What it is not is a performance metric, because performance needs two readings taken the same way.
What to write beside the number before you file it
A composite out of 100 is a compact thing and it travels badly, because it arrives in a slide deck with none of the context that made it true. Three facts belong next to every Grader score you keep: the date you ran it, the three engines it read with the model variant named, and the fact that your company name was inside the question. Write those down and the figure stays usable for a year, because anybody can see what it is a reading of. Leave them off and the same number becomes a target somebody chases next quarter against a different model on a different day, and the second grade will move for reasons that have nothing to do with anything you did.