How to Pitch AI Search to a Skeptical Client
The pitch that fails goes like this: AI is changing search, people are asking chatbots instead of Googling, you need to get ahead of it.
Every word of that may be true. It still fails, for three reasons. The client has heard it from four other people this quarter. They can't verify any of it. And it asks them to spend money on a belief about the future rather than a problem they have now.
The pitch that works doesn't make an argument at all. It shows them something about their own brand that they didn't know and can check in thirty seconds.
Open with their data, not the category
Before the meeting, ask the engines the questions their customers would ask. Not "tell me about [client]" — nobody types that. The questions a buyer types:
Best accounting software for small businesses in Australia. Who are the top commercial roofing contractors in Dallas. Compare project management tools for creative agencies.
Then bring the answers.
If the client is named, you have a baseline and a conversation about holding it. If they're not, you have the only slide that matters: four engines were asked a question their customer asks, and four times a competitor was recommended instead.
No prediction required. It already happened.
The objections, and what they actually mean
The volumes are tiny
Usually true today, and the wrong frame. The question isn't how many people arrive from AI engines this month — it's what happens when a buyer asks for a recommendation and gets a competitor's name. That's a lost deal, not a lost session.
We rank number one on Google
Often true and largely unrelated. Google rank is one input among many to an AI answer, and some engines lean on what they learned months ago rather than today's results. Show them a query where they rank first and still aren't named.
Isn't this just SEO
Partly. The technical groundwork overlaps. What's different is that there is no ranking to hold, answers change between runs, and being mentioned is not the same as being cited. Say that plainly rather than overselling the novelty.
The third one deserves honesty. Some of what you'll do is SEO you were already doing. Claiming otherwise gets found out, and a client who thinks they were sold a rebrand of existing work is a client who cancels.
Three moves that land
Show the competitor by name. A client who hears "you're not very visible in AI" shrugs. A client who sees a named rival recommended in answer to their own customer's question does not shrug. Use their real competitive set, not generic examples.
Run it live. Ask the question in front of them. It takes forty seconds, they watch it happen, and it removes any suspicion that you cherry-picked. It also demonstrates the thing people find hardest to believe — that the answer differs depending on which engine you ask.
Show the disagreement. Named by one engine, absent from another, same question, same day. This is the moment most clients stop arguing about whether the category matters, because it's visibly not a single thing they can ignore.
Be straight about what you can't promise
This matters more than any part of the pitch, and it's where most of this category oversells.
You cannot promise a position. There is no rank to hold in an AI answer.
You cannot promise consistency. The same question produces different answers on different runs, which means one good result isn't proof and one bad result isn't failure.
You cannot promise revenue from it. Referral volumes are small compared to organic search today. They arrive pre-qualified and they're growing, which is a real argument — it just isn't a revenue forecast.
What you can promise is visibility into something they currently cannot see at all, measured the same way every month, with the evidence kept.
Clients have been sold certainty in this category already and some of them have been burned. Being the person who says what isn't knowable is a better position than being the fourth person promising a number.
The ask
Don't ask for budget in the first conversation. Ask for one thing:
Let me run this properly for one month and show you the four numbers.
It's small, it's concrete, and it moves the discussion from whether the category matters to what their actual position is. Most clients say yes to that, and the budget conversation after a month of real data is a different conversation entirely.
Why this works
Because it isn't a pitch. It's a finding.
Agencies lose this conversation when they arrive as an advocate for a trend. They win it when they arrive with something specific and uncomfortable about the client's own brand, and no particular agenda beyond showing it.
The clients who are going to ask about AI search this year will ask somebody. The agency that already has their four numbers is the one that gets to answer.