Alternatives to the HubSpot AI Search Grader: 80 of its 100 points need the engine to know you already
HubSpot publishes the weight of all five dimensions it scores, and three of them read how a brand is described rather than whether it turns up at all.
Published by AI Knows Us · Updated 20 October 2026
HubSpot publishes the full scoring rubric for its AI Search Grader on its own page, which almost nobody in this category does, and reading it tells you exactly which question the tool is built to answer. Sentiment Results carries up to 40 points, Presence Quality up to 20 and Brand Recognition up to 20. That is 80 of the 100, and all three are readings of how an engine already describes your brand. Share of Voice and Market Competition carry 10 each, and those are the two that look outward at whether anybody else is in the answer with you. So the grader is a brand perception instrument, it is good at that job, and the business most likely to go looking for a free grade is the one the engines barely hold an opinion about. 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, both responses identical in length, bodies saved.
What the grader actually asks, in HubSpot's own words
The method is published in four steps and the first two decide everything. Step one: "Input your company name, location, industry, and product/service." Step two: the tool "sends your brand information to ChatGPT, Perplexity, and Gemini and asks each model how it characterizes your brand based on its training data."
Read that twice, because it is the whole design. The question carries your name into the engine, and what comes back is the model's account of a brand it has been handed. That is the right way to measure perception. It is a different reading from the one a buyer actually triggers, which starts with a category and no company in it at all, and ends with three names that are not necessarily yours.
HubSpot also names the model it reads rather than the brand. Its own FAQ says the Grader "currently analyzes GPT-5.4 mini (OpenAI), Perplexity, and Gemini (Google)". Naming the exact model variant is more disclosure than most vendors offer, and it matters, because a mini model is not the model your buyer is talking to.
The five dimensions and what each one needs before it can score
| Dimension, with HubSpot's published weight | What it reads | What it needs to exist first |
|---|---|---|
| Sentiment Results, up to 40 | General, contextual and source based tone across topics | An opinion the engine already holds about you |
| Presence Quality, up to 20 | Mention depth, source quality, data richness | Material about you in the training data to be deep or shallow |
| Brand Recognition, up to 20 | How widely the brand is recognised across training data | Recognition, which is the thing being scored |
| Share of Voice, up to 10 | Your rank against competitors at the time of analysis, plus a rank bonus | Nothing. This one works from zero |
| Market Competition, up to 10 | Category role as Leader, Challenger or Niche Player, and innovation perception as Innovator, Disruptor or Traditionalist | Nothing. This one works from zero |
HubSpot is straight about the consequence on the same page. It writes that a low Market Competition score "typically means answer engines have insufficient or inconsistent signals about where your brand stands competitively". That sentence is the honest description of a young company's whole result, and it applies to the 80 point block just as much as to the 10 point one.
What the Grader does better than we do, said plainly
We sell a product in this category, so here are the rows where HubSpot's free tool is simply ahead of ours.
- It scores sentiment in three layers. General tone, tone by topic, and the credibility of the sources shaping it. We publish no sentiment score at all.
- It will grade anybody. Its own page says the tool "accepts any brand; use it to analyze your own company or evaluate a competitor's AI positioning". Pointing a measurement at a rival is the most useful move a free tool allows and we do not offer it.
- It classifies your position in words. Leader, Challenger or Niche Player, and Innovator, Disruptor or Traditionalist, are categories a marketing director can take into a meeting. We hand over rates and answer text.
- It publishes the arithmetic. Five weights that sum to 100, and a described method of "deterministic scoring with schema validation and retry logic". A rubric you can read before you run it is a standard the rest of this category should be held to.
- It is finished in under two minutes and takes no account, although a short form stands between you and the full report.
Where to go when the score is low for the boring reason
If your five dimension breakdown comes back thin across the 80 point block, the finding is that the engines hold little about you. That is a real and useful finding, and the next instrument is a different shape. These are the alternatives worth a shortlist, grouped by the question you are actually asking.
| The question you now have | What to shortlist | Settle this before you commit |
|---|---|---|
| Am I in the answer when nobody types my name? | Any tracker that lets you write the questions yourself and leaves your brand out of them | Whether you can edit a question later without resetting the history behind it |
| Which pages did the engine read instead of mine? | A tool that records the cited URLs, not only a score | Whether the source list is exportable or only drawn as a chart |
| Is my number moving, or is it noise? | Anything that reports the question count beside the rate | How many readings sit behind a single percentage |
| How am I described, and is the tone fair? | The Grader. This is its actual job | Nothing. Run it, it is free |
| Is this the same on Claude and in AI Overviews? | A product that names five surfaces rather than three | Which surfaces sit in the tier you would be quoted |
A low grade also sends people looking for the wrong next purchase, so it is worth knowing how to read the result you already have before spending anything. How to read your first AI visibility report works through that, and it carried no durable inbound link from any body on this site when we counted on 20 October 2026: 3,456 internal links across 2,050 stored bodies against 1,960 live pages, with 911 pages, 46.5 per cent, carrying none. Linking it here is the sentence doing its job and the orphan getting one.
Questions buyers ask about the HubSpot AI Search Grader
Does the Grader tell me whether ChatGPT recommends my company?
It tells you how the engines characterise your company once your name is in front of them, which is the question it is designed around. The shortlist question starts from a category with no brand in it. Both readings are worth having and they are taken differently, so run the Grader for the first and write your own blind question set for the second.
Is it really free, and what does it take instead of money?
Its own page says there is no cost to run an analysis. It takes your company name, location, industry and product or service to run, and a short form to release the full report. The other cost is the shape of the result: one reading on one day, with nothing to compare it against until you run something else.
Which engines does it read?
GPT-5.4 mini from OpenAI, Perplexity and Gemini from Google, named on its own FAQ. Claude and Google AI Overviews are outside that set, so a buyer whose customers use either of those is reading three fifths of the picture.
Can I grade a competitor with it?
Yes, and HubSpot says so on the page. That is the most interesting thing you can do with it. Grade the two or three companies an engine already names in your category and read their Presence Quality and Brand Recognition against yours, because that gap is the work.
Why does my score look low when my website is good?
Because the 80 point block reads what the engines absorbed about you from the rest of the web rather than from your own site. Review listings, roundups, directory profiles and trade press are what feed Presence Quality and Brand Recognition. A strong website with nothing written about it elsewhere produces exactly the result you are looking at.
Is there a version of this that tracks over time?
HubSpot sells one, and the Grader page links to it in several places. The free tool is a snapshot and the paid product monitors the same three engines against prompts you specify. The move from one to the other is the real buying decision here.
The heaviest dimension is the one you control least directly
Sentiment carries 40 of the 100 points, more than any other dimension, and HubSpot says it is assembled in three layers including a source based one, meaning the credibility of whatever shaped the model's impression of you. Almost none of that material is on your own website. It is reviews, roundups, trade press, forum threads and directory profiles written by other people at other times. So the biggest single component of an AI brand grade is a reputation reading wearing an AI label, and the move it points to is not a site change at all: it is finding the four or five pages an engine in your category keeps reading, and getting your name, your specifics and your numbers onto them. That is slower than editing a page and it is the only work that moves the 40.