Reading time: 11 minutes
The HubSpot AI Search Grader, formerly the HubSpot AEO Grader, is a free check of how AI describes a brand. It asks ChatGPT, Gemini and Perplexity how they describe the company. Then it gives a score from 0 to 100. However, the score doesn’t show whether your site becomes a source in the answers.
In short
- The AEO Grader became the AI Search Grader, but the old address leads to the new page. The analysis is free, needs no account and accepts any brand, even a competitor’s.
- The score adds up five dimensions, and sentiment is worth up to 40 of the 100 points. According to the HubSpot AEO page, ChatGPT and Gemini answer from training, while Perplexity searches the web.
- The five official pages we read don’t say how many questions the tool asks, from which country or in which language. They also cite three different OpenAI models.
- Use the score as a list of hypotheses and check each one. To track a brand month by month, however, the AEO Grader isn’t enough.
What is the HubSpot AI Search Grader?
The HubSpot AI Search Grader is a free, one-time check of how three AIs describe a brand. The official page now uses the name AI Search Grader. Also, the old address, hubspot.com/aeo-grader, redirects to the new one. So people who search for the HubSpot AEO Grader reach the same tool.
To run the analysis, you enter the company name, location, industry and product or service. According to HubSpot, the result is ready in under two minutes, with no account. The tool also accepts any brand, including competitors. The full report, with the score, the dimensions and the written interpretation, however, requires a form. In it, name, email, phone, website and number of employees are all required.
Still, the help center calls the tool the HubSpot AEO Grader. On the tool page, HubSpot even presents it as a free AEO checker. The new name isn’t unique, either, because Mangools and Gushwork have free checkers called AI Search Grader. So check the address before you compare scores.
AEO stands for answer engine optimization, and our guide covers the difference between GEO, SEO and AEO. Finally, HubSpot sells paid monitoring, HubSpot AEO, and presents it as the next step after the grader.
Which engines and criteria does the HubSpot AI Search Grader use?
According to the tool page, the HubSpot AI Search Grader currently analyzes GPT-5.4 mini (OpenAI), Perplexity and Gemini (Google). The 0 to 100 score adds up five dimensions, and sentiment weighs the most. On the other hand, AI Overviews, AI Mode, Copilot and Claude are not on the list.
| Dimension | Weight | What it assesses |
|---|---|---|
| Sentiment Results | Up to 40 points | Overall tone, by topic and by source |
| Presence Quality | Up to 20 points | Depth of mentions, sources and data |
| Brand Recognition | Up to 20 points | How much the model can say about the brand |
| AI Share of Voice | Up to 10 points | Estimated share of the category, plus a rank bonus |
| Market Competition | Up to 10 points | Role in the category and image as an innovator |
The HubSpot AEO page details the collection: the tool runs pre-set queries. ChatGPT and Gemini answer with what they learned in training, while Perplexity uses live search. In other words, in two of the three engines, the answer doesn’t go through today’s web. The difference between answering from training and searching the web explains why this matters for a brand.
From which country and in which language does the AI Search Grader ask?
The page doesn’t say. The tool asks for the company’s location, but no official page says where the questions come from. It also doesn’t say which language they run in. There is a Portuguese version of the page, but it cites another model, GPT-4o. So, for a brand in Brazil, the score doesn’t show what a customer sees when asking in Portuguese.
How does HubSpot calculate the AEO Grader score?
HubSpot publishes the weights and a short description of the math. According to the page, it uses deterministic scoring with schema validation and retry logic. It also cross-validates the three engines. However, it doesn’t show the questions, the number of runs or how each dimension turns into points.
Each dimension combines different signals. Sentiment, for example, has three layers: overall tone, variation by topic and the credibility of the sources that shape the description. Market Competition, in turn, labels the brand as a Leader, Challenger or Niche Player. It also labels it as an Innovator, Disruptor or Traditionalist.
Deterministic scoring means the same answers lead to the same score. AI answers, however, change with every run. For example, in a SparkToro study, 600 volunteers ran 12 prompts 2,961 times in ChatGPT, Claude and Google’s AI. The chance that ChatGPT repeated the same list of brands in two answers was below 1 in 100. Gumshoe.ai, a startup that sells AI tracking, took part in the study.
So nothing ensures that two analyses of the same brand get the same score. Also, the page doesn’t say how HubSpot handles this variation.
How do you read the AEO Grader score?
Read the AEO Grader score as a snapshot of the brand’s image in the models, not as presence in answers with search. After all, HubSpot itself separates the two and says the grader measures brand perception. So look first at the five dimensions and the written interpretation, and only then at the total.
On a page that compares the grader with Otterly, HubSpot splits AI visibility into two parts. The first is brand perception, that is, how the models describe the brand based on training. The second is search visibility, or whether answers cite the brand’s content. According to the company, the tool measures only the first.
Each low dimension points to a different problem:
- Sentiment: the model describes the brand in a neutral tone or with caveats. So write down each negative claim in the interpretation.
- Recognition: the model knows little about the company. This is common for a new brand, because training has a cutoff date.
- Share of voice: competitors take up the category. Therefore, run the AI Search Grader on two or three of them, with the same location and industry.
- Market Competition: according to HubSpot, a low score means insufficient or inconsistent signals about where the brand stands in the category.
Compare the score only with itself, because other checkers use other 0 to 100 formulas. Also, run the analysis twice on the same day, with the same data. Then, if the score changes, the gap shows the size of the noise.
What are the limits of a free, one-time check?
The first limit is the single snapshot: the AEO Grader keeps no history and doesn’t run the questions you choose. Also, in two of the three engines, the answer comes from training, not from today’s web. As a result, the score doesn’t follow recent changes or show whether your site becomes a source.
- No history: each analysis is a snapshot, so write everything down and repeat it the same way.
- Fixed questions: HubSpot picks the questions and doesn’t say how many there are. So the score may miss your customer’s questions.
- Training, not search: your customer’s ChatGPT may search the web when the question needs current information. In that engine, however, the grader uses what the model learned.
- No site citations: the score assesses the sources that shape the brand’s image. But it doesn’t count how many answers cite your site.
- Country, language and margin of error: no official page says.
- Commercial interest: HubSpot sells the paid monitoring and asks for contact details for the full report.
Without the number of questions, by the way, no one can work out the score’s margin of error. For comparison, our measurement guide shows that the margin of a presence rate with 30 answers ranges from ± 18 to ± 31 points.
What should you do after the AEO Grader diagnosis?
Treat the AEO Grader report as a list of hypotheses. First, record the score and check each claim in the interpretation. Then test your customer’s questions and fix what is wrong on the site and in the sources.
- Record the analysis: total score, five dimensions, date and the data you entered, such as location and industry. Without this, you can’t compare the next run.
- Check the facts: price, products, audience and competitors that the interpretation cites. If something is wrong, fix the source and report the error to the AI.
- Test your customer’s questions: ask ChatGPT, Gemini and Google the category questions, without the brand name and without logging in. The step-by-step free test is in our guide on how to check AI visibility.
- Audit the site: check robots, indexing and cited sources with the free GEO audit.
- Repeat it the same way: after the fixes, run the AI Search Grader with the same data. Then compare the dimensions, not just the total.
When is the AEO Grader not enough?
The AEO Grader stops being enough when the question becomes whether the brand is growing in the answers. For that, you need your own questions, several runs, fixed competitors and history. After all, none of these fit in a single check.
HubSpot itself draws this line on the tool page:
“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?'”
HubSpot, AI Search Grader page (2026)
HubSpot AEO, for example, costs $50 per month, with 25 prompts in ChatGPT, Gemini and Perplexity. According to the product page, it updates the score daily and comes included in Marketing Hub Pro and Enterprise. In our GEO tools comparison, by contrast, entry plans range from $29 to $99 per month.
To measure trends, finally, the sample has to grow. The protocol in the measurement guide uses 50 questions, 5 runs and 4 engines per month, or 1,000 answers per brand. NotHuman, which publishes this blog, also works in this market.
What we found when we compared five official HubSpot pages
We read the five official HubSpot pages that describe the tool. They are the English page, the Portuguese version, the HubSpot AEO page, the comparison with Otterly and the help center. Then we counted what each one says about the name, model, collection and weights.
- Name: 2 use AI Search Grader as the main name, while 3 still say AEO Grader.
- OpenAI model: 3 versions appear. The English page cites GPT-5.4 mini, the Portuguese version cites GPT-4o, and the comparison cites GPT-5.2 and GPT-4o.
- Collection: 4 pages describe the collection in 3 ways. Two mention only training, while the Portuguese version mentions your customers’ queries. The HubSpot AEO page, by contrast, mixes training and live search.
- Weights: only 1 of the 5 publishes the points for each dimension.
- Country, language and number of questions: none of the 5 says.
In other words, the pages don’t agree on the OpenAI model, and only the English page cites GPT-5.4 mini. So save the date and the tool’s name with each result.
Frequently asked questions
The page has a Portuguese version, and the tool asks for the company’s location. However, no official page says in which language or from which country the HubSpot AEO Grader asks its questions. So check the written interpretation and test your customer’s questions by hand, in their language.
No. According to HubSpot, the AI Search Grader is free, needs no account or card and accepts any brand. The full report, however, requires a form with name, email, phone, website and number of employees. Ongoing tracking, on the other hand, is HubSpot AEO, which costs $50 per month.
Run the AEO Grader before and after each major change, such as a site fix, always with the same data. Since the tool keeps no history, write down the score and the five dimensions on each run. To see month-by-month trends, however, use fixed, repeated questions.
Want to know how your brand shows up in AI answers today?
NotHuman is opening spots for agencies and brands to track how they appear in AI answers. Join the waitlist and we’ll let you know when a spot opens up for you.
Main sources
- HubSpot, AI Search Grader: the tool page, with the engines and the score weights (2026).
- HubSpot help center, the AEO Grader as a starting point for a content audit (2026).
- HubSpot, HubSpot AEO: price, prompts and how the AEO Grader collects answers (2026).
- SparkToro, research on how inconsistent AI brand recommendations are (2026).
- HubSpot, AI Search Grader, in Portuguese (2026).
- HubSpot, AEO Grader vs. Otterly.ai, with the definition of brand perception (2026).
