From our data · AI search visibility
Which B2B software do AI assistants recommend?
When a software buyer asks ChatGPT for "the best alternatives to Vanta" or "the best CRM for a 100-person company", they get a short list. If your company is not on it, you lose a deal you never knew existed. There is no click to measure, no visit to retarget and no lost opportunity in your CRM. We wanted to know how often that happens to the companies who should worry most: B2B software challengers, the ones competing against a much bigger market leader.
By Prasad Hassan, Founder & CEO · 7 November 2026 · 9 minute read
AI assistants named B2B software challengers in about 70% of buyer questions, usually below the market leader. Smaller challengers fell to 38% to 50%. Comparison and "alternatives" pages were the most cited sources, about a third of citations (32%). Review sites were about 6%. We tested 24 US B2B SaaS challengers across ChatGPT, Claude, Gemini, Perplexity and Google AI Overviews, and reviewed more than 600 answers. Challengers can influence most of what decides the answer: the pages that compare them, the roundups that list them and the facts they publish.
About this research
By Prasad Hassan, Founder and CEO, BrandVelocity. Research by Prasad Hassan and Aryaan Jeppu, with technical support from Priyadharshini Krishnan and Prasann Garg. First published on LinkedIn on 28 October 2026.
How did we test it?
Over three weeks of research, our team ran a structured test of how AI assistants answer buyers' questions.
- 24 US B2B SaaS challengers across compliance, sales, HR, developer tools, customer success and marketing software, chosen from G2 software categories and each paired with the leader buyers most often compare it with. One well-known company, Buffer, served as a control to check the method.
- Four questions each, written the way buyers ask them: "best alternatives to [leader]", "[company] vs [leader]", a specific question such as "best SOC 2 software for a 100-person SaaS company", and a pricing question.
- Five AI engines: ChatGPT, Claude, Gemini, Perplexity and Google's AI Overviews, all with web search switched on.
- More than 600 AI answers and over 5,000 cited sources, including the test we ran on ourselves (described below). The answers were collected between 19 and 21 October 2026. For every answer we recorded whether the company was named, where it appeared in the list and which sources the engine cited. Google AI Overview answers were captured and tagged by an AI model.
Two of the four questions name the company, so it is almost always mentioned. The other two do not, and those are the ones that matter: they show whether an AI assistant brings you up when a buyer has not heard of you yet.
AI search and B2B challengers: the numbers

| Figure | What it measures |
|---|---|
| 70% | of buyer questions named the challenger, usually below the market leader |
| 38 to 50% | for smaller challengers, the ones with most to gain |
| 57% | on Google AI Overviews, the hardest engine to appear in |
| 32% | of citations came from comparison and alternatives pages |
| 6% | of citations came from review sites such as G2 |
| 0 of 44 | unprompted answers named BrandVelocity, a new name |
Source
BrandVelocity.ai research, 19 to 21 October 2026. Each question was run once per engine, so read the figures as directional.
Challengers are named about 70% of the time, and usually below the leader
On the two questions that do not name them, the 23 challengers (excluding the control) appeared in about 70% of answers. When they did appear, they were typically third in the list. Buyers shortlist the first two or three, so being named third is often not enough.
And 70% flatters the market: most companies we tested are established, often well-funded challengers that buyers already know. The smaller ones in our sample fell to 38% to 50%, and BrandVelocity, a new name, to zero. For the many early-stage challengers we did not test, the gap is likely wider still.
The more specific the question, the more challengers drop out
Broad "alternatives" questions named challengers more often than specific ones about a use case or company size. Those specific questions are exactly what buyers ask when they are close to a decision.
One example stood out. We asked for the best SOC 2 compliance platform for a 100-person SaaS company. The engines consistently named Vanta, Drata and Secureframe. A well-reviewed challenger with a strong G2 profile appeared once, in sixth place.
Smaller challengers are often invisible
Well-known challengers were named in most answers. Smaller ones, the companies with the most to gain from being found, appeared in only 38% to 50% of answers.
Google's AI Overviews were the hardest engine to appear in
They named challengers in 57% of answers, the lowest of the five. Claude named them most often, in 89%. Each engine draws on different sources, so a single score across all of them hides where the real gap is.

| Category | ChatGPT | Claude | Gemini | Perplexity | Google AI Overviews |
|---|---|---|---|---|---|
| Compliance | 50% | 100% | 62% | 62% | 62% |
| Sales software | 50% | 90% | 90% | 30% | 40% |
| HR software | 62% | 88% | 100% | 88% | 38% |
| Marketing software | 83% | 100% | 83% | 67% | 67% |
| Developer tools | 80% | 100% | 100% | 90% | 90% |
How often AI engines name challengers, by category
Share of answers naming the challenger when the buyer did not ask about it by name. Two unprompted questions per challenger; categories with three or more challengers shown. Source: BrandVelocity.ai research, 19 to 21 October 2026.
Comparison pages shape the answers more than reviews do
Across all citations, the pages that compare vendors side by side did most of the work.

| Source type | Share of citations |
|---|---|
| Comparison and "alternatives" pages | 32% |
| The company's own website | 14% |
| Company blogs and resources | 13% |
| Editorial roundups ("best X tools") | 13% |
| Competitors' websites | 8% |
| Other companies' websites | 6% |
| Review sites such as G2 | 6% |
| Other: directories, communities, docs | 8% |
Where AI answers about B2B software come from
Share of 4,031 sources cited by ChatGPT, Claude, Gemini, Perplexity and Google AI Overviews. Source: BrandVelocity.ai research, 19 to 21 October 2026.
Reviews matter, but the pages that compare vendors side by side do most of the work. Some of them are written by competitors, including the market leader.
We ran the same test on ourselves
It seemed only fair. We asked the same kinds of questions about AI visibility tools, comparing BrandVelocity with eight competitors.
On questions that did not name us, BrandVelocity was mentioned in none of 44 answers. The engines named the better-known tools instead. On questions that did name us, we were mentioned in 95% of answers and ranked first in more than half, and our own website was the most cited source.
We have done no work on our own AI visibility yet. That gap, between buyers who already know us and buyers who don't, is exactly the problem we set out to solve for others. We will publish our own before-and-after as we fix it.
What should B2B brands do?
- Measure the questions buyers actually ask. Questions that include your name will flatter you. Track the alternatives, use-case and company-size questions where a buyer has not heard of you yet.
- Publish honest comparison and alternatives pages. They are the most cited source type. If you do not write them, your competitors' versions become the answer.
- Get into the roundups the engines already cite. Find the "best X tools" articles that appear in answers for your category and make sure you are in them, with accurate details.
- Make your facts easy to lift. Clear pricing, use cases, an FAQ and consistent descriptions help an engine describe you correctly.
- Use one name and one description everywhere. Engines match your website, review profiles, LinkedIn and directories. Inconsistent names split your identity.
- Treat each engine separately, especially Google's AI Overviews, and re-measure regularly. Answers change as models and sources change.
If you are the category leader, the same work protects your position. Challengers who fix their visibility will start appearing alongside you, so keep your comparison pages and facts accurate.
What are the limits of this test?
We want to be clear about the limits.
- Each question was run once per engine, so treat individual results as directional. Answers vary from run to run, and our next round will repeat each question several times.
- We used the engines' official interfaces with web search on. What a person sees in the consumer apps can differ.
- We have not yet scored accuracy: whether the engines described each company correctly. That is our next step.
- 24 challengers is a sample of the market. Our list leaned toward well-known companies, so the gap for smaller challengers may be larger than it looks here.
The takeaway
AI assistants are already building shortlists for software buyers. For challengers, the biggest risk is simply being left off the list. The good news is that most of what decides the answer can be influenced: the pages that compare you, the roundups that list you and the facts you publish about yourself.
We would be glad to hear how others are measuring this. If you want to know how AI assistants describe your company, get in touch and we will share what we see.
If you are an agency
The same test makes a clear first deliverable for a client. Run the alternatives, use-case and company-size questions across the five engines, record who is named and which sources are cited, and show the client where the gap sits. Repeat it on the same questions after the fixes, and the before-and-after is the report.
How we do this ourselves
Signal AI Visibility Optimization (AIVO) runs this kind of test for your own category every week. Signal AIVO runs the questions your buyers ask across ChatGPT, Claude, Perplexity, Gemini and Google AI Overviews, in the markets you choose, and reads the sources behind each answer, so you can see why a competitor was named. Brand DNA holds your voice and positioning, so the fix is drafted in it. Our team works with you on the changes, and the next run shows whether the answer moved. Signal AIVO is ISO 27001 certified, and you choose the supporting modules you need. Every engagement is shaped to your company, so contact us for pricing and we will walk you through it on your own questions.