How can a private equity firm tell whether ChatGPT recommends its portfolio companies?

The value creation plan for a portfolio company usually assumes more inbound demand: a stronger brand, a better website, a sales team with more to work with. Meanwhile the company's buyers have started asking ChatGPT, Copilot and Google's AI answers which suppliers to shortlist. If the assistant names three competitors and not your company, or names a brand you acquired and retired two years ago as a rival, part of that plan is failing where nobody on the board can see it.

This matters because the shortlist now forms early. In 6sense's 2025 Buyer Experience Report, buying groups had already ranked their preferred vendors before first contact in 94% of cases, and bought from that first choice 77% of the time. In G2's April 2026 survey of 1,076 software buyers, 51% said they now start research in an AI chatbot. And the businesses private capital backs are mostly the kind of firm this affects: UK Private Capital says the industry backs 13,000 UK businesses, nine in ten of them small or medium-sized.

This guide is for operating partners and portfolio chief executives who want a plain way to check whether AI assistants recommend their companies, to report it to the board, and to fix what is missing. A shorter section at the end covers the fund's own visibility when founders ask assistants for investors.

#Why should an operating partner care what ChatGPT says?

Because it shapes who gets the meeting. When a buyer asks an assistant which firms to consider, the answer is a shortlist, and the 6sense figures above suggest most deals go to a firm already on it. A portfolio company that is not named loses deals it never knew about, and the cost shows up only as slower pipeline growth than the plan assumed.

There is a second reason. Assistants now describe companies to anyone who asks: prospective customers, candidates and, at exit, anyone researching the business. A company described wrongly, or described but never recommended, carries that into every conversation. We saw this with ID-Pal, an identity verification firm: "AI could describe ID-Pal. It never recommended it."

#Why do good portfolio companies go missing from AI answers?

Usually for reasons that come with ownership changes, not with the quality of the business. These are the patterns we see most in PE-backed and founder-led B2B firms:

The buy-and-build pattern is the one we would check first. When ID-Pal bought NorthRow and folded it into id-pal.com, "The redirect from northrow.com was technically correct, but nothing on either site told Google or the AI engines that NorthRow and ID-Pal were the same company." Tracking showed both the cost and the recovery once it was fixed: "AI answers still recommending NorthRow as a rival: six in May, one in September 2026." TCC Group, a compliance group with three brands, had the group version of the same problem: "Three brands, one group. AI could only see the pieces."

#How do you audit a portfolio's AI visibility?

Use a fixed method, so the numbers mean the same thing each quarter. Here is the template we would give an operating partner.

#Step 1: Write the buyer questions for each company

For each portfolio company, write 20 to 50 questions its buyers would type, in their words. Cover four kinds: category questions ("best payroll outsourcing firms for UK retailers"), problem questions ("how do we cut debtor days"), comparison questions ("X or Y for mid-sized manufacturers") and checking questions ("is [company] any good?"). Include the region and sector where buyers would. Agree the list with the company's chief executive and head of sales, then fix it for at least a year.

#Step 2: Name the competitors to measure against

List the three to five competitors the sales team meets most often, plus any acquired or retired brands. You are measuring share of answers against these names, and whether an old brand still appears as a rival.

#Step 3: Run each question in a clean session, several times a month

Start with ChatGPT and Google's AI Overviews and AI Mode, then Microsoft Copilot and Gemini, then Perplexity and Claude. Copilot matters in B2B: Microsoft 365 Copilot passed 30 million paid business seats in July 2026, and many buyers use it at work. Answers vary from run to run, so run each question several times and count a win only when it repeats.

#Step 4: Record the same fields every time

#Step 5: Put it on one page for the board

One row per company: share of answers now and last quarter, whether the description is correct, the main cause, the next fix, who owns it and when it will be done. Keep losses on the page as well as wins. ID-Pal's measurement followed the same rule: "The same buyer questions run every month in ChatGPT, Gemini and Claude, in a fresh chat. A win only counts when it repeats, and losses are reported too."

#Which fixes move the numbers, and how fast?

Start with the cheapest fixes to the largest causes. Assistants that search the web live, such as ChatGPT search, Copilot and Google's AI features, can reflect a new page or a corrected listing within weeks. Answers drawn from a model's own memory change only when it is retrained, which takes months.

The ID-Pal fix shows how small the first step can be: "About four hours of structured data work declaring NorthRow part of ID-Pal on both domains. Live in week one, with no copy changes and no risk to the redirect." The dated result: "UK buyer questions where AI names ID-Pal: none in May, two in August, four in September 2026." And the pattern that tells you what to build next: "Where a page has been rebuilt, ID-Pal appears. Where it has not yet, it does not, which tells us exactly what to build next."

#How can portfolio companies get recommended to their buyers?

By giving assistants specific, checkable evidence that matches the questions buyers ask. In practice that means a website organised around those questions rather than internal teams, named experts, consistent facts across every listing, reviews on the sites buyers read, and coverage in the trade press the buyers trust. Our sector guides go deeper for B2B technology companies and consultancies, and how to get ChatGPT to recommend your firm covers the general method.

Being named is one channel. Researched outbound, where a person writes and approves every message, reaches buyers before any assistant is asked, and works alongside it. When a portfolio board asks for the business case, our guide to getting board approval for AI search budget sets out the argument.

#The fund's own visibility: when founders ask for investors

The same test applies to the fund. Founders and advisers ask assistants questions such as "which private equity firms back founder-led B2B services businesses with £1m to £3m EBITDA in the UK?" and "which PE firms are known for being founder-friendly in the lower mid-market?". An assistant can only answer with firms whose criteria and record it can read.

• Criteria in plain text. Sectors, EBITDA range, equity cheque size, deal types, regions and what you do not do, on one page, not in a graphic.
• A portfolio with a paragraph per company. What it does, when you invested, the deal type and what has happened since, agreed with each company.
• Founders' own accounts. Founder-friendly is a claim assistants check against what founders say in the press, on podcasts and on their own sites.
• Independent records that agree. Claim and correct your profiles on Crunchbase and Dealroom, check PitchBook and Beauhurst, and make sure deal coverage in Real Deals, Sifted and regional business press uses your correct name. Awards such as the Real Deals Private Equity Awards, the Growth Investor Awards and the UK Private Capital Awards add independent evidence.
• The trade body's new name. The BVCA formally became UK Private Capital on 27 January 2026, so update any page or profile that still describes you as a BVCA member.

Keep performance and fundraising material behind a professional-investor gate, reviewed by compliance. The FCA's rules on promoting non-mass market investments (COBS 4.12B) restrict promotions to retail clients, with limited exemptions. Regulatory status also varies: some smaller managers, including managers of qualifying venture capital funds, can register with the FCA as small registered UK AIFMs rather than become authorised, and in July 2026 the FCA consulted on a new three-tier AIFM regime, with implementation envisaged for 2028. Describe your status accurately, and take legal advice before publishing anything that could be a promotion.

#What to do this month

1. Pick the three portfolio companies where inbound growth matters most to the plan, and write 20 to 50 buyer questions for each with their chief executives.
2. Run those questions in ChatGPT, Google, Copilot and Gemini, several times each in a clean session, and record the fields in the table above.
3. For every buy-and-build platform, check whether any acquired brand is still named as a rival, and fix the structured data first.
4. Agree one description for each group and brand, and correct the company's website, directory profiles and LinkedIn pages to match.
5. Add a one-page AI visibility summary to the next board pack, with share of answers, the main cause and the next fix per company.

#Frequently asked questions

#Who should own AI visibility: the operating partner or the portfolio chief executive?

The chief executive owns the fixes, because they sit in the company's website, sales and marketing. The operating partner owns the method, so every company is measured the same way and the board sees comparable numbers. A single fixed question list per company, agreed by both, keeps the discussion on results rather than anecdotes.

#How often should we run the audit?

Run the questions several times a month, because single answers vary. Report quarterly to the board, since most fixes take weeks to show in live search and months to reach answers drawn from a model's memory. Keep the question list fixed so quarter-on-quarter changes mean something.

#Can anyone guarantee a portfolio company will be named?

No. Nobody can guarantee what an AI assistant will say, and anyone promising a place in ChatGPT's answers should be treated with caution. A provider can guarantee effort, reporting and outcomes it controls, such as meetings held. ChatGPT has shown labelled ads to UK users on its Free and Go plans since 11 August 2026, and OpenAI says the ads do not change the answer itself.

#Does this matter if the company sells through referrals and outbound?

Yes. A buyer who hears about a company through a referral or an outbound message can check it with an assistant before replying, so a wrong or missing description can cost meetings in those channels too. Researched outbound and being named work best together.

#Which PE firms are known for being founder-friendly in the UK lower mid-market?

An assistant answers this from what founders and the press say, not from fund websites. Firms that publish how they work after completion, and whose founders describe the experience in their own words in public, give it something to repeat. A fund that wants to be named here needs those independent accounts more than a new tagline.

#What should a founder ask a private equity firm before accepting a term sheet?

In our view, start with what happens after completion: who sits on the board, how the value creation plan is set, what happens if the business misses the plan, and what role the founder keeps. Ask to speak to founders the firm has backed, including one where things went badly. Ask how decisions are made and how quickly. A fund that answers these plainly on its own website is also the fund an assistant can describe accurately.

For a portfolio company that needs a plan its own team can deliver, our Paid Discovery sets out every finding, ranked, with fixes written so the company's team can ship them. To start with a quick read on one company, book a free Growth Review.

#Sources