Key Takeaways
- Plan for four requests, not one. A corporate buyer rarely asks an assistant once. They ask who does the work, narrow by size, compare what is left, and ask for a few names with reasons. Each request can cut you.
- Write down the facts each request looks for. The kind of work, the size of client, the timeline, a price range and one result that can be checked. A firm that leaves one out has a blank where its competitors have an answer.
- Give the buyer a reason they can repeat. The person asking has to explain the shortlist to someone above them. A sentence they can lift from your page does that job. A slogan does not.
- Remember the request is shortened before it reaches the web. Microsoft says Copilot searches with a few words drawn from the prompt. The plain terms on your page have to match those few words.
- Do not assume every buyer works this way. Nobody has published a count we would stand behind. Approved vendor lists and referrals still decide a great deal of corporate work.
A director at a mid-size company has a problem, a budget line and a boss who wants three names by the end of the month. She used to ask two colleagues and search for an afternoon. Now she opens the assistant in her work account and starts typing.
How do corporate buyers shortlist consultants with AI in 2026? Not with one question. They ask a run of them, and each one makes the list shorter. Your firm can be on the first answer and gone by the third.
Below is the short answer, the four requests that do the cutting, why each surviving name needs a reason attached, what the work account changes, and what nobody can honestly tell you yet.
How do corporate buyers shortlist consultants with AI?
Corporate buyers shortlist consultants with AI in a series of requests. The first asks who does this kind of work. The next ones narrow by company size or industry, compare what is left on approach, timeline and price, and ask for a few names with a reason for each. A firm stays on the list only while its own pages answer the request in front of it.
That is our description of the pattern. It comes from how these tools work and from how buying inside a company works, where one person gathers names and another approves them. It is not survey data, and we say more about that gap at the end.
The buyer here is the same person described in how executive coaches reach corporate buyers through AI: a sponsor with a budget and someone to answer to. That post covers the page to write for them. This one covers the steps their list goes through.
The four requests that turn ten names into three
The list is cut by follow-up questions, and each follow-up looks for a different fact. The figure follows one made-up search. An HR director at a 400-person software company in Denver needs outside help to redesign pay bands, and asks her assistant four things in a row.
-
Request 1
"Which consulting firms help mid-size software companies redesign pay bands?"
10 named.
Stays in: any firm whose pages say plainly that it does this work.
-
Request 2
"Which of these have worked with companies of 200 to 500 people?"
6 left.
Stays in: a client size written on the page.
-
Request 3
"Compare the six on approach, timeline and price range."
4 left.
Stays in: a timeline and a price range to put in the comparison.
-
Request 4
"Give me three to take to my VP, with one reason each."
3 sent up.
Stays in: a result someone outside the firm can check.
Nobody was judged on quality. Each firm left when a fact was missing.
The company, the director and the counts of 10, 6, 4 and 3 are invented to show the shape. Real searches are messier, and an assistant may add a name at step three that it did not mention at step one. The shape holds anyway: the list shrinks, and it shrinks on facts.
Here is the same sequence as a table, with what each request is looking for.
| The request | What the buyer is doing | What gets a firm cut | What keeps a firm in |
|---|---|---|---|
| "Which firms help companies like ours with this?" | Building a long list | Pages that describe the firm in general terms and never name the work | A plain sentence saying you do this work, for this kind of company |
| "Which of these have worked with companies our size?" | Narrowing by fit | No client size or industry stated anywhere | "We work with companies of 200 to 500 people", written as text |
| "Compare them on approach, timeline and price range." | Comparing what is left | Nothing to put in the comparison: "contact us for pricing" | A named approach, a typical length and a price range or a floor |
| "Give me three to take to my VP, with a reason each." | Preparing to defend the choice | Claims only the firm itself makes | A described result, a credential or a mention on a site that is not yours |
The third request is the one to worry about if you publish no price. A comparison needs something in every box, and "contact us" is not something. Why a missing price hurts, and the four ways to state one, are covered in whether consultants should list fees online.
Why every name on the list needs a reason
A corporate buyer has to explain the shortlist to someone else, so they ask the assistant for reasons along with names. An assistant can only give a reason that is written down somewhere it can read. That makes the last request the strictest one.
Think about what the director does with the answer. She pastes three names into a message to her VP. Under each name she needs a line: why this one. "They have done pay band redesigns for software companies our size, it takes about ten weeks, and they publish a range that fits our budget" is a line she can send. "They are a trusted partner for growing organizations" is not.
So write the line for her. One or two sentences near the top of the relevant page, stating the work, the kind of client, the length and one result. If she can lift it without editing, so can her assistant.
The reason holds up better when part of it comes from outside your own site. A credential with the name of the body that grants it, an article you wrote for a trade publication, a client who describes the project in their own words. The buyer is asking "says who?", and "says the firm" is the weakest answer available.
This work has a calendar. Shortlists for next year's projects are mostly written while budgets are being drafted, which is why the timing in how consultants win budget season with AI visibility matters as much as the pages themselves.
What the work account does to the request
A work assistant shortens the request before it searches, and in some companies it does not search the web at all. Both are documented by the companies that make these tools, and both change what your pages need to say.
Microsoft describes the first in its page on web search in Microsoft Copilot, last updated in August 2026. When web search is on, Copilot writes its own search query and sends that to Bing. The query, in Microsoft's words, "consists of a few words informed by the user's prompt". A careful four-sentence request about pay bands becomes a handful of terms. If those terms are not on your page as plain text, the match is weak.
The same page covers the second. An IT admin can turn web search on or off for the whole company with one setting, and Microsoft says its guidance also applies to Researcher, the Copilot tool built for longer research tasks like this one. Where web search is off, the shortlist is drawn from what the company already has on file and what the model already knew. More on that in whether Copilot answers differently at work.
Buyers who use Google meet a related step. Google's page on AI features and your website says AI Overviews and AI Mode may use a "query fan-out" technique, issuing "multiple related searches across subtopics and data sources". One request about consultants can become separate searches about price, about company size and about results. A page that covers those subtopics in text has more ways to be found. The same Google page says there are no extra technical requirements to appear, and that important content should be available in textual form.
Neither company publishes how names are chosen once the pages are found. What they do publish points the same way: short search terms, several searches per request, and text over images.
What nobody knows yet, and where this does not apply
Nobody has published a trustworthy count of how many corporate buyers shortlist this way, and a lot of corporate work never passes through an assistant at all. Know both before you rewrite anything.
We looked for a figure on the share of corporate buyers who use an assistant to choose consultants and did not find one we would repeat. Vendors publish surveys, and the numbers vary with who was asked and what counted as "using AI". We will not quote one here.
Three cases where the four requests matter less:
- An approved vendor list decides. If the company can only hire firms already on its list, the assistant's answer is background reading. Getting on the list is a procurement process, not a writing job.
- A formal request for proposals is running. The buyer already has names. An assistant may still be used to check each one, which is a question about your name, not a search for one.
- A referral arrived first. A colleague's recommendation skips the long list. The buyer may still ask an assistant about you by name, so the facts need to be right even then.
In all three, the same written facts still help, because a buyer who already has your name uses them to confirm it. They are just not what gets you considered.
One more limit, on our side. You cannot see into a buyer's work account, and neither can we. What can be measured is the open web answer, the one anybody gets outside a company account. The free check asks ChatGPT 25 real buyer questions from your niche at $0 and emails a plain-English readout: your score, who got named in your place, and the one fix to start with. It runs as soon as you ask for it, and the report lands within the hour between 7am and 5pm Central, or by 10am the next morning outside those hours.
Questions we hear the most
How do corporate buyers shortlist consultants with AI?
In a series of requests. They ask an assistant who does a kind of work, narrow the names by company size or industry, ask for a comparison on approach, timeline and price, then ask for a few names with a reason for each. A firm stays in while its pages answer each request.
Do corporate buyers really use AI to choose consultants?
Some do, and nobody has a reliable count. We have not found a published figure on how many corporate buyers use an assistant to shortlist consultants that we would repeat. Treat it as one route among several, next to referrals and approved vendor lists.
What gets a consulting firm cut from an AI shortlist?
A missing fact, most of the time. When a buyer asks an assistant to narrow by client size or compare on price and timeline, a firm that has not written those things down has nothing to put in the answer, so it drops out.
Does a work assistant search the web when a buyer asks about consultants?
Often, but not always. Microsoft says Copilot can search the web through Bing when web search is on, and that an IT admin can turn it off for a whole company. When it is off, your website is not part of the answer.
Can I see the shortlist a corporate buyer got from their assistant?
No. The conversation sits inside their account and you have no view of it. What you can do is run the same four requests yourself on the open web, using the words your buyers use, and see where your firm falls out.
What does the AnswerHalo free check show a consultant who sells to companies?
It shows what ChatGPT says on the open web. The free check asks ChatGPT 25 real buyer questions from your niche at $0 and emails your score, who got named in your place and the one fix to start with. It cannot see inside a company work account.
What does the $500 audit add to the free check?
The audit runs 125 checks across ChatGPT, Claude, Gemini, Perplexity and Google AI Overviews, keeps every answer so you can read it, and lands by this time tomorrow. The free check covers ChatGPT only, with 25 questions.
Your buyers are already asking. Find out what AI tells them.
The free check asks ChatGPT 25 real buyer questions about your niche and sends you the report within the hour, at $0. If it shows you are already getting named everywhere, we will say so in plain words.
Get my free check →