KNOWN33KNOWN33
Get Your AI Visibility Audit
Menu

Original Research

We Asked ChatGPT 12 B2B Buying Questions. The First Vendor Stayed the Same in All Six Categories.

KNOWN33 tested 12 ChatGPT buying prompts across cybersecurity, SaaS, AI support, and fintech. See which vendors came first and how the shortlists changed.

By Asher KnoxPublished 10 min read

Original research by KNOWN33 · November 20, 2026

Ask ChatGPT to recommend AI customer support software for a B2B SaaS company, and it names Intercom Fin first. Ask it which AI agents can help a SaaS team deflect tickets, and it starts with Intercom Fin again. Keep reading and the two answers begin to diverge. Plain and Pylon appear in the opening shortlist for the broad question. Ada, Forethought, and Decagon appear in the opening shortlist for the ticket-deflection question.

That was the most revealing example in a KNOWN33 pilot covering cloud security, endpoint detection, customer success, AI customer support, spend management, and subscription payments. We asked two unbranded buying questions in each category. One vendor led both answers in every pair. Below that first position, differences in wording opened up different routes onto the shortlist.

For B2B marketers, the distinction matters. A single generic prompt can tell you who leads a category query. It can miss the specific buying scenarios where your brand is visible or where a competitor enters the conversation.

The results at a glance

Industry Product category First listed in both answers Selected additional vendors recommended in both answers
Cybersecurity Cloud-native application protection platforms (CNAPP) Wiz Orca Security, CrowdStrike, Palo Alto Networks
Cybersecurity Endpoint detection and response (EDR) Microsoft Defender for Endpoint CrowdStrike, SentinelOne, Sophos
B2B SaaS Customer success platforms Gainsight ChurnZero, Planhat, Vitally, Custify
AI-enabled SaaS AI customer support Intercom Fin Zendesk AI, Salesforce Agentforce, Ada, Decagon, Forethought
Fintech Spend management Ramp Brex, BILL Spend & Expense, Airbase, Navan
Fintech Subscription payments Stripe Paddle, Braintree, Chargebee

Each answer is an observation of what ChatGPT presented to a hypothetical buyer. The first-listed position comes from the initial vendor shortlist or comparison. The right-hand column includes selected recommendations from anywhere in both answers. Those are two different measures of visibility. A brand can appear in both answers while making the opening shortlist in only one.

What the 12 prompts measured

There are three questions a marketing team can ask of an AI-generated shortlist.

  1. Who gets introduced first? The first name gives you a repeatable position to record, even when an answer uses a table or prose. List order can reflect how the response is organized. Its commercial effect would require a separate study of buyer behavior.
  2. Who appears across multiple buying scenarios? A vendor with repeat coverage is showing up across different ways buyers describe the problem. Here, two appearances are evidence about the two specific prompts.
  3. Who makes the opening shortlist for a particular need? A vendor may be named in a longer comparison for one prompt and surface near the top when the buyer names a problem that fits it. A binary mention metric would assign both appearances the same value.

We use the third distinction throughout the findings below. An opening-list change and a brand disappearing from an answer are different outcomes.

Cybersecurity

Cloud security | A familiar first choice, with different reasons to consider alternatives

We asked for a mid-market cloud security shortlist and then specified a SaaS company running on AWS. Both answers introduced Wiz first. Orca Security, CrowdStrike, and Palo Alto Networks also appeared in both. We grouped the Palo Alto references at the vendor level because the responses used different product naming.

The mid-market question put Microsoft Defender for Cloud in the opening shortlist and described it through an Azure or Microsoft-centric use case. The AWS SaaS question instead included Sysdig in its opening shortlist and described it in relation to Kubernetes and container runtime security. The cloud environment was one of several wording changes between the prompts, so the test cannot isolate which change produced that difference.

For a cloud security marketer, this suggests a specific measurement question. If you test your category visibility with a single CNAPP prompt, do you see the same competitors your sales team encounters in AWS-heavy accounts, Kubernetes-heavy accounts, and Microsoft-centric accounts? Each of those contexts can be tested with its own matched prompts.

Endpoint security | Team constraints changed the emphasis

Microsoft Defender for Endpoint was the first vendor in the initial lists for both a mid-sized business and a lean security team. CrowdStrike, SentinelOne, and Sophos appeared in both.

The lean-team answer made operational burden part of the comparison. It discussed the work of running the tool and brought Huntress into the initial vendor table, with an explanation tied to managed detection and limited in-house coverage. This example gives a marketing team something more specific to track than a mention. The answer supplied a reason the vendor might fit that buyer.

A defensible follow-up would pair a general EDR prompt with versions that change one constraint at a time, such as team size, existing Microsoft licensing, or the need for outsourced detection. That would help distinguish coverage across different environments from coverage tied to the buyer’s ability to operate the product.

B2B SaaS

Customer success | The opening name and the eventual fit can differ

Gainsight was named first when we asked for the best customer success platforms for B2B SaaS and when we asked which tools help reduce churn and manage renewals. ChurnZero, Planhat, Vitally, and Custify also received recommendations in both answers.

Both answers also grouped recommendations by company profile. Gainsight was associated with enterprise needs. For an earlier-stage or smaller team, the answers put forward options such as Custify. Vitally and ChurnZero were repeatedly associated with growing or mid-market teams.

This is a case where first position and use-case fit tell different parts of the story. A customer success vendor could appear below Gainsight in the general list and still be recommended for a particular team size. Tracking only the first name would miss that positioning. Tracking only presence would miss which buyer the assistant thought the vendor was for.

For a SaaS team, that makes the accompanying description worth recording alongside each mention. If your product is repeatedly associated with buyers below your target segment, an increase in mentions can coexist with a positioning gap.

AI-enabled SaaS

AI support | The difference between appearing in an answer and opening it

Intercom Fin was first for both AI support questions. Zendesk AI and Salesforce Agentforce were also in both opening shortlists. Yet the rest of the opening lists tell two different stories.

Broad B2B SaaS support question Ticket-deflection question
Intercom Fin Intercom Fin
Plain Zendesk AI Agents
Pylon Salesforce Agentforce
Zendesk AI Ada
Salesforce Agentforce Forethought
Decagon

The general B2B answer described Plain in relation to technical support and Pylon in relation to Slack-first support. The ticket-deflection answer focused more directly on resolving tickets end to end without a human agent. Ada, Forethought, and Decagon appeared later in the broad answer too. The meaningful difference was their move into the opening shortlist when the question explicitly asked about deflection.

This is where conventional visibility reporting can hide an important change. Counting mentions gives Ada credit in both answers. Recording opening-shortlist presence shows that the buyer’s description of the problem changed when Ada entered the conversation. A marketing team can use that distinction to decide which queries to test next and which value proposition to examine.

Fintech

Spend management | The same starting point across two ways of buying

Ramp appeared first for both spend management and corporate cards paired with expense management. Brex, BILL Spend & Expense, Airbase, and Navan were also recommended in both answers.

The opening lists moved at the margin. Rippling appeared in the opening shortlist for the spend management prompt. Navan appeared in the opening shortlist for the corporate card and expense management prompt and was also recommended later in the spend management answer. The responses linked products to scenarios such as travel, procurement, and existing HR or payroll workflows.

For a fintech team, this is a reason to test the language different stakeholders use. The product category might be the same for the vendor, while buyers ask about corporate cards, approval controls, procurement, or travel expenses. Each wording gives the model a different job to solve.

Subscription payments | Buyers can get several types of providers in one shortlist

Stripe led both answers. Paddle, Braintree, and Chargebee appeared in both opening shortlists. GoCardless appeared in the opening list for the recurring-revenue question. Adyen, Recurly, and Maxio appeared in the opening list for the subscription-software question.

The two answers grouped a payments processor, a merchant of record, and subscription billing platforms close together, even as they explained differences between these models. From a visibility perspective, the competitive set in an AI answer can include products that solve adjacent parts of the buyer’s problem.

If you sell payment infrastructure, a search for your precise category is one view of the market. A buyer asking how to handle recurring revenue can surface providers who also solve for billing, tax administration, or international sales. Measuring both categories of question shows which alternatives the buyer is actually being presented with.

What the recommendations had in common

Across the six categories, the answers often paired a vendor with a recognizable trade-off or buyer situation. Wiz was discussed in terms of cloud visibility. Microsoft Defender was linked to an existing Microsoft environment. Gainsight was framed around enterprise needs. Intercom Fin was described through AI support automation. Ramp was positioned as a starting point for startup spend management. Stripe was presented in terms of payments and billing for SaaS.

Those are observations about the language in the answers. This pilot did not test what caused each recommendation. Establishing a relationship between visibility and a vendor’s website, third-party reviews, directory listings, or editorial coverage would require collecting and coding the cited sources and comparing them with the vendors surfaced in each answer.

A marketing team can still use the results to build a more diagnostic benchmark. For each product category, test the category phrase, the job a buyer needs done, a constraint that changes product fit, and a switch or comparison question. Record the first-listed vendor, every brand in the opening shortlist, every additional recommendation, and the reason given for each. That will tell you whether the model puts your brand forward, when it does so, and how it describes you.

Methodology and limits

We used ChatGPT in fresh, signed-out chats on November 20, 2026. We chose six product categories relevant to KNOWN33’s audience and wrote two English-language, unbranded buyer prompts for each. Both prompts in a pair concerned the same broad buying decision. In several pairs, we changed the scenario as well as the wording.

Each prompt contributes one verified answer to the analysis. We excluded submissions that contained text from a previous prompt and reran the six affected prompts. We checked the displayed prompt against the intended wording and confirmed the answers had finished. We used the interface’s default behavior and did not independently control the underlying model version, location, or search settings.

We recorded the first vendor in the initial shortlist or comparison and manually identified additional recommendations in both answers. Repeated references to a vendor within one answer counted as one appearance. We normalized obvious product-name variants at the vendor level when comparing appearances. We did not code citation frequency, verify product claims in the answers, or measure buyer clicks or conversions.

The sample is small and purposive. We ran each prompt once, so the study measures agreement across these prompt pairs on the test date. It cannot estimate how often ChatGPT repeats an answer to the same prompt or how often the same vendor leads across users, dates, models, or an entire industry. A larger benchmark should repeat each prompt across multiple sessions, control one wording change at a time where possible, and record whether each answer visibly used web search.

Appendix | The 12 questions

Product category Prompt A Prompt B
Cloud security Which cloud security platforms should a mid-market company shortlist for CNAPP? What are the best CNAPP tools for a SaaS company running on AWS?
Endpoint security What are the best endpoint detection and response platforms for a mid-sized business? Which EDR vendors should a lean security team evaluate?
Customer success What are the best customer success platforms for a B2B SaaS company? Which customer success tools help a SaaS team reduce churn and manage renewals?
AI support What are the best AI customer support platforms for B2B SaaS? Which AI support agents should a SaaS company shortlist to deflect tickets?
Spend management What are the best spend management platforms for a US technology startup? Which corporate card and expense management platforms should a growing US startup evaluate?
Subscription payments What are the best payment processors for a US SaaS company with recurring revenue? Which payment providers should a US subscription software company shortlist?

Where does your brand enter the shortlist?

The question to ask about your own category is where your brand appears when buyers describe the problem they need solved. Start with the questions your prospects ask on sales calls. See which competitors the answers introduce, which reasons they give, and whether they describe your product in terms your best customers would recognize.

KNOWN33 helps B2B companies measure and improve visibility across search and AI answers. Get in touch to see how your brand appears for the questions your buyers ask.

MAKE YOUR NEXT MOVE AN INFORMED ONE

See how visible your brand is in AI search.

Get Your AI Visibility Audit