AI Consultant Discovery Call: What to Prepare So You Don't Waste 30 Minutes

Bottom line: bring your numbers, one named problem, a rough budget, and the actual decision-maker to an AI consultant discovery call, and you'll get a proper scope back instead of a generic slide deck. Turn up with "we want to use AI somehow" and you'll get a proposal that could apply to any business on earth. The prep takes about an hour. It's the highest-use hour in the whole process.

What this call is really for

A discovery call is not a sales pitch, though plenty of consultants run it like one. It's a diagnostic conversation where someone tries to work out three things fast: is there a real problem here, is there money to fix it, and can I fix it. Most last thirty to forty-five minutes. If yours runs to an hour and it's still vague at the end, that's a sign the consultant didn't know what to ask, or you didn't bring enough for them to work with.

I've sat on both sides of this table for years now, first as the client buying agency help, then as the person people book calls with. The pattern I see over and over is the same: businesses that come with numbers get useful, specific proposals within days. Businesses that come with feelings get a forty-slide deck about "AI transformation" that could have been written before the call even happened.

The five things to bring with you

1. Your actual numbers

Not turnover for the whole business, though have that ready too. I mean the numbers around the specific pain point. If your problem is slow lead response, know your current response time and how many leads you get a month. If it's content, know how many pieces you publish and how long each one takes. A consultant can't cost anything against "we're a bit slow at this." They can cost something against "we take four hours to reply to 60 enquiries a week and we lose about a fifth of them to a competitor who answers same day."

2. One named task, not a vision

"We want to use AI in the business" is not a brief, it's a mood. Pick the single most annoying, most repeated, most expensive task in your week and name it. Invoice chasing. Appointment booking. First-draft proposals. Social captions. If you can't narrow it to one thing, that itself is useful information, and a good consultant will help you narrow it on the call, but come with your best attempt at an answer rather than a blank page.

3. A rough budget, even a bad one

People hate giving this number because they think it will be used against them. It won't, or it shouldn't be, and if a consultant tries to squeeze every last pound out of a vague budget hint, that's a red flag worth noting for later. But not giving any range at all wastes the call, because the consultant ends up either underselling what's possible or pitching something three times your ceiling. Day rates for independent AI consultants in the UK sit roughly between £750 and £1,500, small audits run £1,500 to £5,000, and full implementation projects can run into five figures. Know roughly which zone you're in before the call. If you want a fuller sense of what different scopes cost, this breakdown of what an AI consultant charges is worth ten minutes before the call.

4. The person who says yes

This is the one people skip most and it costs them the most. If the owner, the ops manager, and the person who controls the marketing budget are three different people, get them all on the call, or at least tell the consultant who isn't there and why. I've had brilliant forty-minute calls with a marketing manager, sent a scoped proposal within the week, and then watched it sit unread for six weeks because the actual budget holder had never heard of the conversation. That's not the consultant's fault and it's not really the marketing manager's fault either. It's a structural gap that a five-second sentence at the start of the call would have closed.

5. What you've already tried

If you've dabbled with ChatGPT, tried a chatbot plugin, or had someone build you a workflow that broke after two weeks, say so. Consultants waste time re-pitching things you've already tried and abandoned, and it makes you look less prepared than you are, when you've done more homework than most. Bring the tool names, what worked, what didn't, and why it stopped. If part of what you tried involved auto-booking or calendar tools, mention specifics rather than "we tried some scheduling thing", because the fix might be as simple as setting up proper auto scheduling for appointments rather than a full system rebuild.

A call that went sideways, and what fixed it

A few months back I had a discovery call with the owner of a boutique estate agency in Kent, six staff, doing decent numbers but drowning in admin. The call started with "we want to be more AI-forward as a business" and honestly, for the first fifteen minutes, we were talking in circles. Nice enough conversation, no traction.

Then I asked what took the longest amount of time in an average week that felt pointless. She said writing property descriptions and chasing viewing confirmations by phone. Two sentences and the whole call changed shape. We could talk about volume (roughly 40 new listings a month), current time cost (about six hours a week on descriptions alone, done by someone earning £28,000 a year), and what a fix might realistically save. That's the difference one named task makes. If you're in property specifically, there's a longer look at what this kind of engagement covers in this piece on what an AI consultant for real estate agents does, worth reading before your own call if that's your world.

The proposal that followed was scoped, priced, and accepted within nine days. Compare that to calls that stay abstract, which in my experience take three to five times longer to close, if they close at all.

Questions to ask them back

A discovery call runs both ways. You're being assessed but you should be assessing too. Ask what they'll deliver by when, ask for one example of a similar business they've worked with (not a case study PDF, an actual named or anonymised example with a number attached), and ask what happens if the first approach doesn't work. If the answers are vague, mentally file that away. There's a fuller checklist on this in the guide to evaluating an AI consultancy before you sign anything, and it's worth reading before the call, not after you've already agreed to something.

Also ask directly whether the call is free and what it commits you to. Some consultants run discovery calls as free diagnostic conversations. Others run them as the top of a sales funnel where the "discovery" bit is thirty minutes of rapport-building before a pre-written pitch deck. Both models are fine, but you deserve to know which one you're in before you hand over your numbers.

Where discovery calls waste people's time

Here's the bit most people won't say out loud: a lot of discovery calls are a waste of time for the consultant too, and both sides quietly know it. If you show up unprepared, most consultants won't tell you the call was pointless. They'll smile, promise to "send something over," and send a generic proposal because that's all they had to work with. You'll read it, think it feels a bit off-target, and conclude the consultant wasn't very good. Often the consultant wasn't the problem. The brief was.

The flip side is uncomfortable too. Some consultants decide their fee and their pitch within the first five minutes based on how you talk, what your website looks like, and what industry you're in, regardless of what you say for the rest of the call. Preparation doesn't fully protect you from that, but a business owner who arrives with numbers, a named problem, and a clear budget range is much harder to pitch generic nonsense at. Specificity is a kind of self-defence in these calls, not just good manners.

What to have open on your screen during the call

• Last three months of the relevant numbers, whatever they are: enquiries, response times, hours spent, content output.
• A short list, no more than five items, of tools you currently use for the task in question.
• Your calendar, so you can book a follow-up on the spot rather than doing an email back-and-forth over the next fortnight.
• One competitor or comparable business you admire, if you have one, because it gives the consultant a useful reference point.
• Your rough budget range, written down so you say the same number you meant to say, not a smaller one out of nerves.

What happens after the call, if it went well

A good consultant follows up within three to five working days with something specific: a scope, a rough cost, a timeline, and usually a named first step. If what you get back is a long email about "the transformative power of AI" with no numbers in it, that's a sign the call didn't do its job, and it's worth a short follow-up email restating your numbers before you decide anything. If you want to see what proper groundwork looks like before this whole process starts, there's a good walkthrough in how to prepare your business for an AI consultant engagement, and it pairs well with everything above because it covers the operational side rather than just the call itself.

One more thing worth flagging before you commit to anything: a lot of what gets pitched as a full "AI transformation" is really scoped better as a smaller, cheaper audit first. If you're not sure which you need, this guide to what an AI audit covers and what it should cost will help you tell the difference before you agree to a bigger contract than you need.

The practical next step is on my hire an AI consultant page, and the engagement itself is described under how to find an AI consultant worth hiring.

For the closest example to your business, start with AI consultant by industry.

Frequently asked questions

How long should an AI consultant discovery call last?

Most run thirty to forty-five minutes. Anything under fifteen is too rushed to cover real numbers, and anything over an hour usually means the conversation stayed too general and needed a named problem earlier on.

Do I need to know exactly what AI tool I want before the call?

No, and honestly it's better if you don't arrive with a fixed tool in mind, because that can box the consultant into pitching around your choice rather than the right fix. Bring the problem and the numbers, and let them recommend the tool.

Should the discovery call be free?

Most are, but it's worth asking upfront rather than assuming. A free call is usually a diagnostic conversation with a proposal to follow; a paid discovery session, sometimes called a mini-audit, usually comes with a short written report attached, which can be worth it for a more complicated business.

What's the biggest mistake people make on these calls?

Turning up with a vision instead of a problem. "We want to be more AI-forward" gives a consultant nothing to price against. "We spend six hours a week writing property descriptions" gives them a scope, a cost, and a timeline within days.

Useful references

• Anthropic's own documentation
• the NIST AI Risk Management Framework