EX-01. B2B SaaS growth tools. AI visibility audit.

HubSpot

How a category leader appears, and doesn’t appear, in the answers B2B SaaS buyers get from AI assistants.

HubSpot dominates category searches. It disappears in problem searches.

Across 270 responses, HubSpot appears in over half. When a buyer describes a symptom instead of naming a category, HubSpot appears in one in seven. And in 41 of 63 problem-first responses, no vendor gets named at all.

9/63

HubSpot in problem-first responses

HubSpot’s presence in symptom-shaped queries.

Overall category presence: 137 of 270 (50.7%). HubSpot is the highest problem-first brand mention. The runner-up (Google Analytics) is at 8, and 41 of 63 responses in this bucket name no vendor at all.

A B2B growth lead types this at 11pm. The response is 3,000 words with no vendor named.

Query. Bucket 8. Problem-first. Anthropic Claude with web search enabled.

how do I stop leads from ghosting after a demo

Great question — post-demo ghosting is one of the most common frustrations in sales. Here’s a practical playbook to reduce it significantly:

before the demo

1. Qualify harder upfront. Use a framework like MEDDIC or BANT to confirm budget authority, a real time-bound problem, and decision-makers involved.

2. Set a next-step expectation before the demo. "At the end of our call, assuming it’s a fit, we’ll agree on next steps together — sound good?"

during the demo

4. Tailor ruthlessly. Discovery questions first, then only show features that solve their specific pain.

6. Book the next meeting on the call. Don’t end with "I’ll send some times" — open the calendar live and get the next meeting booked before hanging up.

after the demo

9. Structured follow-up cadence. Day 1: recap. Day 3: value-add. Day 7: check-in. Day 12: different angle. Day 18: "break-up" email.

10. Send the "break-up" email. "Should I assume this isn’t a priority right now and close the file?" — often gets a response.

Response continues with 1,200 more words on multi-threading, video follow-up, and root-cause diagnosis. No named tools or vendors appear anywhere.

Vendors named in this response NONE

Appearance rank and pick-share rank tell different stories.

Every brand ranked two ways. Appearance = how often the brand is named at all. Pick share = how often the brand is the SOLE or PRIMARY pick, as a percentage of all 270 responses. HubSpot dominates both. The runner-ups do not.

#	Brand	Appear	App %	Pick %
01	HubSpot	137	50.7%	19.3%
02	Salesforce	77	28.5%	0.4%
03	Amplitude	66	24.4%	2.2%
04	Mixpanel	65	24.1%	1.1%
05	ActiveCampaign	58	21.5%	3.0%
06	Customer.io	50	18.5%	2.6%
07	Marketo	48	17.8%	0.0%
08	Google Analytics	47	17.4%	2.6%
09	PostHog	44	16.3%	2.2%
10	Segment	41	15.2%	3.3%
11	Stripe	28	10.4%	0.4%
18	Clay	21	7.8%	3.0%

HubSpot’s dominance is worse than the appearance number suggests. The 50.7% appearance share is a floor; the 19.3% pick share is 5x the nearest competitor (Segment and Clay tied at 3.3% and 3.0%). Salesforce appears in 28.5% of responses but is the pick in 0.4% — nine of its 77 appearances are UNFAVORABLE and 34 are NEUTRAL. Salesforce is being cited as an example, not chosen. Marketo appears in 48 responses with 15 UNFAVORABLE mentions (31%). That’s active pushback.

Clay’s pattern is unusual. 21 appearances, 90.5% recommendation-when-mentioned, 4 SOLE picks and 4 PRIMARY picks. Zero NEUTRAL, zero UNFAVORABLE. When AI engines name Clay they mean it as the answer, not the citation. Retrieval problem, not a recommendation problem.

Counts reflect responses where the brand actually appears, adjudicated by a two-stage LLM pipeline (matcher + recommendation classifier). Earlier versions of this exhibit used a single-pass matcher and reported inflated counts for Segment, Make, and Refine. See the audit protocol.

Top-mentioned brand in each type of buyer query.

Bucket	Top	n	Runner-up
Category discovery	HubSpot	36	Mixpanel 23
Comparison	Salesforce	17	HubSpot 15
Segment fit	PostHog	26	HubSpot 22
Evaluation	HubSpot	19	Salesforce 18
Pricing	HubSpot	18	ActiveCampaign 9
Trust	Clay, Amplitude	9	Make 8
Switching	HubSpot	18	ActiveCampaign 17
Problem-first	HubSpot	9	Google Analytics 8 (65% zero-vendor)

No incumbent owns the problem-first lane.

The highest-leverage queries: the buyer describes a symptom, not a category. HubSpot at 9 of 63 is the highest brand mention. Then a long tail of scattered vendors. Then 41 of 63 responses that name no vendor at all.

#	Brand	Count	% of bucket
01	HubSpot	9	14.3%
02	Google Analytics	8	12.7%
03	Google Ads	6	9.5%
04	Salesforce	5	7.9%
05	Mixpanel	4	6.3%
n/a	Zero-vendor responses	41	65.1%

Nothing owns problem-first. Across 63 problem-first responses, the highest brand mention (HubSpot at 9) covers 14% of the space. The verbatim exhibit above, 3,000 words of advice with zero vendors named, is the norm, not the exception. 41 of 63 responses in this bucket look exactly like it. This isn’t a competitive gap. It’s empty inventory.

Replicated finding. The category leader landing at 9 of 63 in problem-first responses replicates in EX-02 (Amplitude, product analytics). Two category leaders in unrelated verticals converged at 14 percent problem-first presence. Provisional read: category leaders may converge here as a category-of-buyer-behavior effect. The third audit will confirm or falsify.

The domains the assistants pulled from.

LinkedIn, Reddit, YouTube do most of the legitimate citation work. Roughly half the top-30 domains are lower-quality SEO listing sites.

Established, own content

linkedin.com72
reddit.com62
youtube.com55
amplitude.com own53
userpilot.com27
medium.com25
hubspot.com own24
gainsight.com24
zapier.com20
g2.com16

SEO listing sites

sequenzy.com66
saashero.net65
aigrowthagent.co35
default.com35
b2bsaastools.com28
hackceleration.com26
leadhaste.com22
hellyeahai.com21
miinfotech.com21
ustechautomations.com20

About half of the top-30 cited domains are lower-quality SEO listing sites. For a brand attempting to improve citation share: presence on LinkedIn, Reddit, YouTube, and G2, plus structured own-domain content, does more work than any single agency-owned publication.

What the data suggests.

The report shows what happened. This reads the data for what to do about it, without claiming to explain why a brand ranks where it does.

Observation

LinkedIn (72), Reddit (62), and YouTube (55) account for most legitimate citations. Own-domain content from Amplitude and HubSpot also gets picked up, when structured as factual data.

Move

Publish comparison content on LinkedIn, participate in Reddit threads, film YouTube deep-dives. Own-domain content should lead with data, not narrative.

Observation

Half of the top-30 cited domains are mid-tier SEO listing sites. These sites accept vendor submissions. Their prominence is a real signal, independent of editorial merit.

Move

Audit whether inclusion on these lists is a legitimate placement opportunity. A hygiene task most in-house teams skip.

Observation

HubSpot dominates queries where the buyer names the category. It nearly disappears when the buyer describes a symptom. The problem-first bucket has no incumbent, and 65% of responses in it name no vendor at all.

Move

Content targeting category vocabulary is winning half the game. Content that describes buyer problems in the buyer’s own language is the missed half.

Observation

Closing a 9-of-63 problem-first gap through content and citation authority is a 6 to 12 month project. Meanwhile, the empty inventory in problem-first responses is a paid arbitrage window.

Move

Ads put you in those conversations next week while the content strategy compounds behind them. The two aren’t alternatives. The paid pilot buys attention time so the organic build has room to accrue.

Methodology and caveats.

Run date. 2026-07-23 UTC. Every finding is a snapshot of that day.
Sample. 30 buyer questions across 8 buckets, 3 samples per engine. 270 successful responses.
Engines. Anthropic Claude (web_search tool), OpenAI GPT-4o Search Preview, Perplexity Sonar-Pro.
API is not the consumer app. The consumer ChatGPT, Claude, and Perplexity products use different retrieval and personalization. Directionally representative, not literal.
Answers vary. Every prompt sampled 3 times per engine. Frequencies, not binaries.
Sample, not census. 30 prompts, human-reviewed. Reasonable people would choose different ones.
Two-stage matcher (v0.4). Brand extraction is case-insensitive, word-boundary-anchored. Vendors whose canonical name overlaps with a common English word (Segment, Make, Refine, Default, Outreach, Apollo, Iterable, Heap) are flagged ambiguous in the config, and every candidate hit goes to a second-stage LLM adjudication (Claude Haiku 4.5) that reads +/-200 characters of surrounding context and answers whether the token is the company or generic usage. All decisions logged to the audit’s adjudication cache. Previous versions of this exhibit used a single-pass matcher and reported inflated counts for verb-collision brands.
Recommendation classification (v0.5). Every confirmed brand mention runs through a second LLM pass (Claude Haiku 4.5) that classifies the treatment of that brand in that response as one of six codes: SOLE (only vendor named as the answer), PRIMARY (named as the top pick), ALTERNATIVE (named in a short recommended set), CONDITIONAL (recommended only for a stated narrow case), NEUTRAL (mentioned without endorsement, or cited as a source only), or UNFAVORABLE (named with a negative or dismissive frame). Pick share of total = (SOLE + PRIMARY) ÷ total responses. All decisions logged to a cache.
No editorializing. The prescriptive section reads the data for moves without claiming causation.
Right of reply. Any named brand can request their own free audit. Findings publish alongside.

Want an audit like this for your brand?

Every 30-day pilot starts with one. Fixed fee. Ad spend billed to your account. Written verdict on day 30.

Apply for a pilot

Or read EX-02, Amplitude and product analytics.