AEO for HubSpot users

HubSpot users can track answer engine visibility in three places in their account. HubSpot's AEO tool measures how often ChatGPT, Gemini, and Perplexity mention the brand for a set of prompts. The AI Referrals traffic source labels visits from AI assistants on contact, company, and deal records. A custom form field can record what buyers say about how they found you. The AEO tool does not track Google's AI Overviews or AI Mode, and no tool can guarantee a citation.

This guide explains what each HubSpot feature does, according to HubSpot's knowledge base as checked on September 27, 2026, where independent evidence agrees or disagrees with HubSpot's advice, and how to set up measurement you can trust. HubSpot labels its AEO tool as beta, so details may change. AEO HQ wrote this guide without HubSpot's involvement, and HubSpot has not reviewed or endorsed it. It is part of AEO HQ's complete guide to answer engine optimization (AEO).

Scope and definitions

This guide is for marketing and revenue teams whose CRM or website runs on HubSpot. It covers HubSpot's AEO tool, its free AI Search Grader, the traffic source properties, custom properties and forms, and the robots.txt settings in HubSpot's website tools. Setting up the same tracking in Google Analytics is covered in how to track AI referral traffic in GA4, and the metrics themselves in AEO metrics and KPIs.

• Answer engine. An AI system that answers a question in its own words and names or links its sources. HubSpot's AEO tool supports three: ChatGPT, Gemini, and Perplexity (opens in a new tab).
• Prompt. HubSpot defines prompts as "questions that people might ask AI answer engines (such as ChatGPT) when searching for brands, products, or services related to your business" (opens in a new tab).
• Mention and citation. A mention is the brand's name in an answer. HubSpot records a citation when an answer includes a link to a webpage or a domain name embedded in its text, and counts citations separately from brand mentions (opens in a new tab).
• Brand visibility. HubSpot's term for how often your brand is mentioned among the prompts you track (opens in a new tab).
• Share of voice. HubSpot defines it as the proportion of total brand mentions that belong to your brand across tracked prompts (opens in a new tab).
• Original and Latest Traffic Source. HubSpot properties that record the first and most recent known web source through which a contact interacted with your business (opens in a new tab).
• Self-reported attribution. Asking buyers directly how they heard of you, and storing the answer.

What HubSpot offers for answer engines

HubSpot launched HubSpot AEO on April 14, 2026, as a standalone product at $50 a month with no HubSpot plan required, and built the same tool into Marketing Hub Professional and Enterprise (opens in a new tab). HubSpot's product pages and knowledge base describe these tools:

The Grader's five dimensions are sentiment (up to 40 points), presence quality (20), brand recognition (20), share of voice (10), and market competition (10) (opens in a new tab). Because it reads the models' training data, it describes what the models remember, not what they find when they search. Search matters for questions about specific vendors: Anthropic's documentation, for example, says Claude searches for information that is "current, changing, or outside its training data," including information about specific organizations and products (opens in a new tab). Our reading: treat the Grader as a snapshot of model memory, and use tracked prompts for anything current.

How HubSpot's AEO tool measures visibility

What the documentation states

• Engines and schedule. The tool covers ChatGPT, Gemini, and Perplexity, and HubSpot says it "runs tracked prompts daily. Because responses change over time, review performance across multiple days or weeks before evaluating trends" (opens in a new tab).
• Prompts. HubSpot generates suggested prompts from the brand, competitors, products, and ideal customer profiles you confirm, and in Marketing Hub from your business context. You can edit them, add your own, or generate 8 to 15 more with AI (opens in a new tab). Each prompt is assigned one buyer journey phase: awareness, consideration, evaluation, or decision (opens in a new tab).
• Brand visibility. The score is analyzed on prompt coverage (the percentage of prompts where the brand appears at least once), frequency across answer engines, and relative competitor presence (opens in a new tab).
• Citations. Reports show the domains and pages cited, and group citations by channel: owned domains, competitor sites, earned media, peer companies, user-generated content, affiliate sites, and social platforms (opens in a new tab).
• Changes. Changes to competitors only affect future data runs, and the tool tracks one owned domain per setup (opens in a new tab).

What the documentation does not state

The knowledge base articles we read do not say how many times each prompt runs on each engine per day, whether answers are collected through the engines' consumer apps or their APIs, or whether the scores carry an error range (opens in a new tab), and we found no published formula that combines the three parts of the brand visibility score. At the Professional limits, 25 prompts on 3 engines every day would produce about 2,250 answers in a 30-day month, close to the 2,500-answer limit. That is our arithmetic, not HubSpot's statement. If each prompt runs once a day per engine, a single day's result for one prompt on one engine rests on one answer.

These gaps matter because answers vary. When the same prompts were run 2,961 times, ChatGPT and Google's AI returned the same list of brands less than once in 100 runs (opens in a new tab) (industry study; a co-investigator works for a tracking vendor). The interface matters too: in an audit of product recommendations, the API and the consumer app of the same assistant shared only 12.0% (ChatGPT) and 14.8% (Gemini) of cited domains (opens in a new tab) (preprint). HubSpot's own advice to judge trends over days or weeks fits this evidence. Our recommendation: read the scores over rolling four-week windows, and ask HubSpot which interface its answers come from before comparing them with what buyers see.

Which engines it leaves out

In a March–April 2026 survey of 519 U.S. B2B professionals who use AI at work, product research ran on ChatGPT (71%), Gemini (61%), Microsoft Copilot (45%), Perplexity (18%), and Claude (14%) (opens in a new tab) (vendor survey). HubSpot's tool covers three of these. For the rest, use first-party reports where they exist:

• Google AI Overviews and AI Mode. Search Console's generative AI performance report counts impressions of your links in AI Overviews and AI Mode (opens in a new tab).
• Microsoft Copilot. Bing Webmaster Tools' AI Performance report shows citations, cited pages, and grounding queries across Copilot and Bing's AI summaries (opens in a new tab).
• Claude. Run your key prompts by hand, or with another tool, on a fixed schedule.

How to read HubSpot's recommendations

HubSpot says recommendations are generated from citation patterns across your tracked prompts: frequently cited domains, common content formats, recurring themes and keywords, and competitor presence (opens in a new tab). Each has a content type, a channel, and a priority, and a "citation lift" column shows "the projected increase in your citation rate if you act on the recommendation" (opens in a new tab). They fall into four groups: a site audit of content and technical issues (including structured data markup, canonical tags, page speed signals, and crawlability), owned content, social amplification on Reddit, LinkedIn, and YouTube, and outreach to third-party publishers (opens in a new tab).

Our reading of this design: the recommendations describe which sources engines cite today for your prompts. That is useful for finding gaps, but it is a correlation, and the citation lift is a projection, not a measured effect. We found no independent test of whether acting on them raises citations. Three groups line up with independent evidence:

• Outreach and earned coverage. Answers lean on other sites: in API tests with ranking-style prompts, ChatGPT drew 93.5% to 95.1% of its cited sources from earned media (opens in a new tab) (preprint). The guide to brand mentions covers how to earn coverage honestly.
• Social amplification. Google says seeking inauthentic "mentions" across the web "isn't as helpful as it might seem" (opens in a new tab). Take part in threads under your own name and within each community's rules; our answer engine optimization antipatterns page covers the risks of undisclosed posts.
• Generated blog posts. In Marketing Hub, generating a post from a recommendation uses 1,000 HubSpot Credits, and the drafts include answer-first sections and FAQ coverage (opens in a new tab). HubSpot itself says that when optimizing content with AI, "it is strongly recommended to manually review and edit your content" (opens in a new tab). Independent evidence supports caution: in a peer-reviewed laboratory pipeline, rewriting page text to optimize it for generative engines hurt retrieval (opens in a new tab), and Google says using generative AI to generate many pages without adding value for users may violate its scaled content abuse policy (opens in a new tab). Use a draft as an outline, add facts only you have, and give each question one page. How to write content for answer engines covers the method.

Where HubSpot's advice and the evidence differ

Most of HubSpot's advice matches independent evidence. The table lists where it does not, or where the evidence is thinner than the advice suggests.

Tracking AI referrals and pipeline in HubSpot

AI Referrals is one of the values of HubSpot's Original and Latest Traffic Source properties. For such visits, the first drill-down property holds the AI platform's domain (for example, chatgpt.com or claude.ai), and the second holds the campaign name from the utm_campaign parameter (opens in a new tab). HubSpot assigns a visit to AI Referrals when a UTM parameter contains a recognized AI platform's domain, or the referring domain is a recognized AI platform (opens in a new tab). It recognizes ChatGPT, Claude, Perplexity, Microsoft Copilot, Google Gemini, Meta AI, Mistral, Poe, and Grok (opens in a new tab).

Four limits follow from the documentation:

1. ChatGPT links are labeled even without a referrer. ChatGPT automatically adds utm_source=chatgpt.com to referral URLs (opens in a new tab), which meets HubSpot's rule for AI Referrals (our reading of the two documents).
2. Google's AI Overviews and AI Mode are not separated. HubSpot says regular Google search and Bing search remain Organic search, and only AI-specific domains such as gemini.google.com count as AI Referrals (opens in a new tab). Google reports AI Overview and AI Mode traffic within the overall "Web" search type in Search Console (opens in a new tab), and GA4's AI Assistant channel counts it as Organic Search (opens in a new tab).
3. Some assistant visits look direct. Traffic referred by Claude's native app carries no Referer header (opens in a new tab), and HubSpot files a visit with no referring domain or tracking URL as Direct traffic (opens in a new tab). HubSpot also notes that ad blockers may block its tracking cookies and affect traffic source values (opens in a new tab).
4. The original source is the first known visit. When HubSpot matches a new contact to earlier anonymous visits, Original Traffic Source updates to the source of the earliest visit (opens in a new tab). A buyer who first came from a search result and later from ChatGPT keeps the search value.

The gap can be large. One AEO agency compared its HubSpot records with what leads wrote in a free-text "how did you hear about us" field: HubSpot's first-touch Original Source credited 28 of 189 self-reported AI-sourced leads to AI Referrals; the others were recorded as Organic Search (95), Direct Traffic (59), or referral and other channels (7) (opens in a new tab) (single firm; the agency sells AEO services).

For pipeline, a deal's Original Traffic Source is the original traffic source of the associated contact with the earliest activity (opens in a new tab), and the traffic source values can be used in saved views, segments, workflows, and custom reports (opens in a new tab). Treat the AI Referrals count as a floor, not a total. The GA4 guide linked above sets up the same view in Google Analytics.

Recording how buyers heard of you

Referral data misses much of AI's influence, so ask buyers directly. In HubSpot, custom properties are created under Settings, Properties, Create property, where you choose a field type such as a dropdown (opens in a new tab), and form fields store visitors' answers in specific CRM properties (opens in a new tab).

Our recommendation:

1. Create a contact property named "How did you first hear about us?" with options for ChatGPT, Gemini, Google AI Overviews or AI Mode, Perplexity, Claude, Microsoft Copilot, Google search, Bing, LinkedIn, a colleague or referral, an event, and other, plus a free-text property for details.
2. Add it to demo, contact, and sign-up forms, and ask the same question on first sales calls.
3. Build a report that sets the self-reported answer beside Original Traffic Source for contacts and deals, and review it monthly.

Checking crawler access on a HubSpot site

An answer engine cannot cite a page its crawler cannot fetch. In HubSpot, robots.txt is edited under Settings, Content, Pages, on the SEO & Crawlers tab, either as a default for all connected domains or overridden for one domain (opens in a new tab). These are the search crawlers to allow for the main answer engines:

• Know which group a crawler follows. Under the Robots Exclusion Protocol, a crawler obeys the group that names it, and the * group applies only when no named group matches (opens in a new tab). If you add a named group, repeat in it any Disallow lines you still need.
• Publish on a connected domain. HubSpot says content on its system domains containing hs-sites is always set to noindex in robots.txt (opens in a new tab).
• Do not combine noindex with a robots.txt block. HubSpot notes that blocking a page in robots.txt stops search engines from seeing its noindex tag (opens in a new tab).
• Check that key facts are in the HTML. In Vercel's December 2024 network data, none of the major AI crawlers rendered JavaScript (opens in a new tab). Fetch your pricing and service pages without running JavaScript, and confirm the facts are there.

Steps

These steps are AEO HQ's recommendations for a HubSpot account. Steps 1 and 3 follow HubSpot's documentation; the rest apply the evidence above.

1. Check crawler access in HubSpot's robots.txt settings, as described above.
2. Write your prompt list before accepting HubSpot's suggestions. Start from real buyer questions in sales calls, support tickets, and form submissions, with several phrasings of each. HubSpot asks that prompts reflect how your target audience would naturally search (opens in a new tab) and be written in the language your ideal customers would use in their location (opens in a new tab).
3. Set up the AEO tool under Marketing, AEO: enter the brand, domain, and brand name variations, which are case-sensitive (opens in a new tab), then competitors, products, and ideal customer profiles. For prompt generation and recommendations, a Super Admin must turn on the required AI settings (opens in a new tab).
4. Record a baseline for at least four weeks before changing pages, and do not act on day-to-day moves.
5. Add the self-reported source field to your forms.
6. Build one report of contacts and deals by Original Traffic Source, with the AI Referrals drill-down, beside the self-reported answers.
7. Add the reports HubSpot's tool does not cover: Search Console for AI Overviews and AI Mode, and Bing Webmaster Tools for Copilot.
8. Treat each recommendation as an experiment. Change a set of pages, leave comparable pages alone, and compare the two groups over the same weeks.
9. Review every AI-generated draft line by line before it is published.
10. Check what answers say about you, not only whether they mention you. In March and April 2026, 11.0% of 98,020 atomic claims in Google AI Overviews were not supported by the pages they cited (opens in a new tab) (preprint). Correct the source pages when an answer is wrong.

What the evidence shows and does not show

Antipatterns

An antipattern is a practice that looks helpful but fails or backfires.

1. Reading the AI Search Grader as a visibility score. It is a one-time check of what models say from their training data (opens in a new tab). Instead, track prompts over weeks.
2. Reacting to daily changes. HubSpot itself advises reviewing multiple days or weeks (opens in a new tab). Instead, compare four-week windows.
3. Crediting every rise to your changes. Unchanged pages grew 3.5 times in one field study (opens in a new tab). Instead, keep a control group.
4. Publishing generated drafts without review, or at scale. Instead, review every draft (opens in a new tab), as HubSpot also advises, and give each question one page.
5. Treating AI Referrals as the whole picture. It misses assistant visits without a referrer (opens in a new tab) and keeps regular Google search, where AI Overviews appear, in organic search (opens in a new tab). Instead, add a self-reported source field.
6. Tracking only the engines the tool covers. 45% of surveyed B2B professionals used Microsoft Copilot for product research (opens in a new tab). Instead, add Bing's and Google's reports.
7. Posting in communities as a hidden promotion. Google says inauthentic mentions are not as helpful as they seem (opens in a new tab). Instead, take part openly, under your own name.
8. Blocking crawlers by accident. A named robots.txt group replaces the * rules for that crawler (opens in a new tab). Instead, read the file for each crawler you care about.

Checklist

FAQ

What is HubSpot AEO?

It is HubSpot's tool for tracking how answer engines describe a brand. It tracks prompts across ChatGPT, Gemini, and Perplexity and reports brand visibility, competitors, citations, and recommendations (opens in a new tab). HubSpot sells it on its own and includes it in Marketing Hub Professional and Enterprise.

Do I need Marketing Hub to use HubSpot AEO?

No. HubSpot says the standalone tool costs $50 a month with no additional HubSpot subscription required (opens in a new tab). The Marketing Hub version adds CRM-based prompt suggestions and blog post generation (opens in a new tab).

Does HubSpot AEO track Google AI Overviews?

No. Its supported engines are ChatGPT, Gemini, and Perplexity (opens in a new tab). Use Search Console's generative AI performance report (opens in a new tab) for AI Overviews and AI Mode.

Is the AI Search Grader the same as HubSpot AEO?

No. The Grader is a free, one-time check based on the models' training data (opens in a new tab). HubSpot AEO runs your prompts every day.

Where do visits from ChatGPT show up in HubSpot?

Under the AI Referrals traffic source, with chatgpt.com as the first drill-down value (opens in a new tab).

Can HubSpot show which prompt led to a deal?

Not from the traffic source properties. For AI Referrals they record the AI platform's domain and any utm_campaign value (opens in a new tab), not the question the buyer asked. A self-reported field can capture more.

Will HubSpot's AI-generated blog posts help AEO?

We found no independent test of it. The evidence on rewriting pages for AI engines is mostly negative, as described above, and HubSpot itself recommends reviewing AI-optimized content by hand.

Next steps

If you sell software, SEO, AEO, and GEO for B2B SaaS companies applies this to SaaS buying. AEO HQ's Technical Implementation add-on (from $5,000), which is included in the Blueprint + Implementation package ($8,995), lists a HubSpot attribution tie-in among its features. See what each AEO service includes.

Change log

• September 28, 2026: First published.