Found by AI is a three-part Velo series on how B2B buyers now use AI to choose suppliers. Each sector is found in a different way. Manufacturers have to unlock what they already know. Technology firms have to prove what they claim. Professional services firms have to earn the right to be cited.
With most professional services companies using thought leadership, how do you make sure yours is cited in LLMs?
Ask an AI tool what a new employment regulation means for mid-sized firms and it writes a competent summary in seconds. It doesn’t need to cite anyone to do it. For years that kind of explainer has filled professional services websites, and it’s losing its value fast.
That makes professional services the hardest sector in which to be found by AI. It’s also the sector with the most to gain from getting it right.
Buyers use AI to research and humans to decide
Buyers of advisory services use AI like everyone else. Gartner found that 45% of B2B buyers used GenAI in a recent purchase, mainly to gather information on vendors. They don’t stop there. They use an average of seven information sources, and 51% expect GenAI to be more likely than other sources to mislead them. So they check what it tells them against peers, product experts and analysts.
For a professional services firm, buyers turn to check an AI answer with people. The same buyers who ask AI to “recommend transfer pricing advisers for a UK group expanding into the US” then look for evidence that those advisers are recognised authorities. That evidence includes:
- A quote in the financial press
- A ranking in a legal or accountancy directory
- A conference keynote
- A report co-authored with a trade body
When AI builds its answer, it draws on the same public record.
Relationships still matter enormously. AI doesn’t replace the partner lunch; it decides who gets invited.
What survives when AI writes the basics
If AI can produce the explainer, only three kinds of content still earn a citation:
- An original point of view. A clear, defensible position on a live issue that a buyer, journalist or AI tool would attribute to you.
- Proprietary data. Research only your firm could produce, from your client base, your deals or your surveys. Data gets cited; comment on other people’s data rarely does.
- Named expertise. Partners whose names appear in enough credible places that AI treats them as authorities in their own right.
The pressure goes beyond marketing. UK searches for “value-based pricing” are up 17% and “outcome-based pricing” is rising. When AI shortens the work, clients question the hours, and recognised expertise becomes what justifies the fee.
Searches for “professional services marketing” are up 52%, with keyword difficulty at just 6 out of 100, so plenty of firms know something needs to change. Few have acted on it.
The moments that count
Adviser relationships tend to stay put, and around 95% of buyers are out of market at any one time. Firms usually get their chance through one of five triggers:
- A current adviser fails, or a key partner leaves.
- A new need arises, such as a deal, a dispute or an expansion.
- Procurement reviews the panel. Procurement are decision-makers in 53% of buying cycles.
- A regulation changes.
- An innovation reshapes the service.
When the moment arrives, the firms already in the buyer’s mind usually win. 68% of buyers have a front-runner from the start, and it wins 80% of the time.
Steps to influence your appearance in LLMs: an editorial operating model
For advisory firms, AI visibility is less a website project and more a new way of producing and sharing expertise. The same three levers apply as in any sector, but here off-site signals carry the most weight.
Your people as recognisable authorities (architecture)
- Build every partner page as a full profile AI can recognise: consistent names and titles, credentials, publications, speaking history and links to LinkedIn. Add structured data for people.
- Remove duplicate content across office and practice pages, and make sure every practice area can be crawled.
A point of view worth citing (on-site content)
- Give every article a named author with a biography.
- Take one clear position per piece, not a balanced survey.
- Publish at least one piece of original research a year.
- Build an FAQ from the questions clients ask in first meetings, answered with your firm’s view.
- Describe service lines by the client’s situation (“preparing for sale”, “expanding into the US”).
- Show proof through anonymised outcomes and ranking commentary when confidentiality rules out named case studies.
Presence where buyers check (off-site signals)
- Run a digital PR programme that gets partners quoted quickly when news breaks.
- Invest in submissions to the rankings and directories your market relies on.
- Pursue speaking slots, podcast appearances and research partnerships with trade bodies or universities.
- Support partners in publishing under their own names on LinkedIn.
- Keep office listings and Google Business Profile reviews accurate.
Your first 90 days
Days 1 to 30:
- Days 1 to 30: Get access to Google and Bing Search Console, then review the errors it reports. Run a prompt analysis: ask ChatGPT, Perplexity and Google’s AI Mode the questions clients ask before appointing an adviser, and note which firms and people get named. Fix titles, headings, redirects and duplicate content. Complete partner profiles and publish credentials and insight you already have. Expect errors to fall.
- Days 31 to 60: Agree a small number of firm positions on live issues and publish problem-led pieces around them. Build anonymised proof and tidy up directory listings. Expect rankings and traffic on key terms to rise.
- Days 61 to 90: Launch the digital PR and citation plan, line up speaking and research partnerships, and set up a fast-track sign-off for press comment. Rerun the prompt analysis and compare. Expect your AI appearance to grow.
The obstacles here are human, not technical. Partners have little time, sign-off is cautious and firms tend to say similar things. Content built from interviews saves partner hours. Positions agreed in advance, together with a two-hour sign-off route for press comment, keep good opportunities from dying in review.
Earn the right to be cited
No firm can claim authority; it has to be earned in public, one credible opinion at a time. The firms AI names tomorrow will be the ones other sources are quoting today.
Also in this series: In Part 1, manufacturers learn how to unlock specification data AI can’t currently read. In Part 2, technology firms learn why AI rewards proof over promise.
Want to know whether AI recommends you? Velo can run the prompt analysis on your market and show you who gets named, and why.