Bottom line: An AI consultant can only move as fast as your data, your decision-makers, and your honesty about what is broken will let them, so the prep work you do in the two weeks before day one decides whether you get a six-week win or a six-month stall. Most of that prep costs you nothing but time, and most businesses skip it anyway.
What happens when you don’t prepare
A few years back I was brought into a 14-person accountancy firm outside Reading. The brief sounded simple: automate client onboarding and cut the time it took to respond to routine queries. Good project. Clear win available. It took me eighteen working days before I wrote a single workflow.
Why? Because nobody could tell me where the client data lived. There were three versions of the client list, one in the practice management software, one in a shared spreadsheet that two partners kept privately updated because they didn’t trust the software, and one in a junior’s head because she’d been the one fielding calls for four years and just knew things. Nobody owned the decision about which one was correct. Nobody had checked in over a year whether the CRM fields matched what staff typed into emails.
None of that is unusual. It is, in fact, the normal state of most small and mid-sized businesses I walk into. But it meant the client paid for eighteen days of me sitting in meetings asking “who owns this file” instead of building anything. That is not a story about a bad consultant or a bad client. It is a story about a business that hired help before it had done its own homework.
If you are in Scotland, the same preparation applies to a local engagement: see what I cover as an AI consultant in Glasgow.
Get your data in order before anyone shows up
This is the boring bit and it is also the bit that determines everything else. AI tools, whatever the sales pitch tells you, are only as useful as what you feed them.
- Pick one source of truth for your customer or client data and say it out loud to your whole team, in writing, before the consultant arrives.
- Delete or archive anything duplicated. If you have three spreadsheets tracking the same leads, that is not redundancy, that is three chances to get the automation wrong.
- Check your data for basic hygiene: missing fields, inconsistent formatting, dead email addresses. A consultant will spend real hours doing this for you if you don’t, and you will pay for those hours at consultant rates, not admin rates.
- Know roughly how much data you have and where it sits: which CRM, which spreadsheets, which inboxes, which cloud folders. Write it on one page. It sounds simple. Almost nobody does it.
There’s a wider point here that gets lost in the excitement. The generative AI business statistics I pulled together for 2026 show that businesses using AI tools well report meaningful time savings, but the ones that see nothing back are almost always the ones that skipped this step. The tool was never the bottleneck. The mess underneath it was.
Decide what problem you’re solving
“We want to use AI more” is not a brief. It is a feeling. A consultant worth their fee will push back on that feeling in the first meeting, but you’ll save real money if you arrive with something sharper already worked out.
Ask yourself three questions before you sign anything:
- What specific task, repeated often, currently eats the most hours across your team?
- What would “better” look like in a number, not a feeling: fewer support tickets, faster quote turnaround, more leads followed up within an hour instead of a week?
- Who in your business does that task today, and have you asked them what’s wrong with it?
I’ve seen this go both ways. One client, a small property management firm, came to me wanting “AI for everything.” When we sat down, the real pain was that tenant maintenance requests sat unread for two to three days because they went into a shared inbox nobody owned. That’s not an AI transformation project. That’s a triage and routing fix that took nine days start to finish, and it saved staff roughly six hours a week almost immediately. Compare that to a business that wants a sweeping AI strategy but can’t name the task it’s trying to fix. That second business pays for a lot of workshop time before anything real gets built.
Get the right people in the room, not just the most senior ones
Here’s the uncomfortable bit that most guides on this topic gloss over: the person who signs the contract is rarely the person who knows what’s broken. Owners and directors tend to describe their business as they wish it worked. The receptionist, the ops manager, the person answering support tickets at 4pm on a Friday, they know where the actual friction is, and they are almost never in the kickoff meeting.
Bring them in. Not for the whole engagement, but for at least the discovery sessions. A consultant asking “walk me through your day” to the person who does the work will surface more useful, specific problems in twenty minutes than three hours of a leadership team talking about “strategic AI adoption.” I have sat in far too many kickoff calls where the loudest voice in the room hadn’t touched the software being discussed in over a year.
This matters for a second reason too: adoption. If the frontline staff feel a new tool was designed around them rather than done to them, they use it. If it lands on their desk with no warning, they quietly work around it, and six months later you’re paying for software nobody opens.
Sort your access and permissions before day one
This sounds like a small thing. It is not. A consultant cannot audit your customer service workflows if they’re waiting four days for admin access to your helpdesk software, and that wait comes out of the engagement clock you’re paying for.
Before the engagement starts:
- Prepare read or admin access (whichever is needed) to the tools involved: CRM, email platform, helpdesk, project management software, accounting package.
- Check your data protection obligations. If you’re UK based and any customer data will pass through a third-party AI tool, you need to know where that data is processed and whether your privacy policy already covers it. This isn’t optional paperwork, it’s the difference between a smooth rollout and a compliance headache six weeks in.
- Name one internal point of contact who can grant access, answer questions, and make small decisions without needing a full leadership sign off every time. Engagements stall badly when every question needs a Thursday meeting to answer.
Budget for the whole picture, not just the day rate
Most businesses budget for the consultant’s fee and forget the tool subscriptions, the staff time needed for training, and the cost of any process cleanup that gets uncovered along the way. If you want a realistic sense of what an engagement should cost and where the money goes, it’s worth reading through what an AI consultant typically costs before you get quotes, so you can spot a proposal that’s suspiciously cheap or padded with fluff.
Pricing works both ways here too. If you’re a small business owner thinking about eventually offering AI-adjacent services yourself, or you simply want to understand how consultants structure their fees so you’re not overpaying, the same logic that governs how consultants price AI services for small business clients tells you a lot about what a fair quote looks like for your own engagement. If a fee seems to have no relationship to hours, scope, or deliverables, ask why.
Get honest about what will change
Here’s the part nobody likes putting in writing: an AI consultant engagement will expose things about how your business runs, not how you think it runs. Response times that are worse than you believed. A star employee who is quietly the single point of failure for a process nobody else understands. A customer service backlog that’s been “fine” for years because nobody measured it.
This is not a flaw in the process. It is the point. If you go in expecting only good news, you’ll be tempted to argue with the findings rather than fix them, and that’s how engagements turn into six months of politely disagreeing with the person you’re paying to tell you the truth. The businesses that get the most out of this kind of work are the ones who arrive already braced to hear something uncomfortable, and who treat it as useful information rather than an insult.
If part of what’s being examined is customer service, it helps to have already looked at how AI is changing customer service for small businesses so you arrive with a sense of what’s realistic, rather than expecting a chatbot to fix a staffing problem.
A short checklist for the two weeks before day one
- Day 1 to 3: Agree one source of truth for your core data (customer list, product catalogue, whatever’s central to the problem).
- Day 3 to 6: Write a one-page brief naming the specific task, the current time or cost it takes, and the target improvement.
- Day 6 to 8: List every tool and platform involved, and prepare admin access ready to hand over.
- Day 8 to 10: Identify and brief the two or three staff members who’ll join discovery sessions, including at least one frontline person, not just managers.
- Day 10 to 12: Check data protection and privacy policy coverage for any third-party AI tool that will touch customer data.
- Day 12 to 14: Set one internal point of contact who can answer questions daily without needing full sign off each time.
Do this and most consultants will tell you, quietly, that you’ve saved yourself somewhere between one and three weeks of billable discovery time. That’s not a small thing when you’re paying a day rate.
What this looks like if you’re a one-person business
Solo operators often assume this kind of prep is only for companies with a team to organise. It isn’t. If you’re running things alone, your prep work is smaller but no less important: know exactly which tools hold your client data, know what task is eating your week, and be honest about what you’ll delegate to a system versus what you’ll keep doing yourself. Solopreneurs who’ve already spent time working out which AI tools fit a one-person operation, the kind covered in this honest guide to AI tools for solopreneurs, tend to get far more out of a consultant because they walk in already speaking the language.
Industry-specific prep matters more than generic advice
General checklists only get you so far. A real estate agency preparing for an engagement needs different things ready than a professional services firm: property data quality, CRM lead routing, compliance around client communications. If that’s your world, it’s worth reading what an AI consultant for real estate agents does before your first meeting, so you can ask sharper, more specific questions rather than generic ones about “using AI more.”
The same goes if marketing is the focus of the engagement rather than operations. Knowing what an AI marketing consultant does and when you should hire one will help you separate a marketing-shaped problem from an operations-shaped one before you brief anyone, which matters because the two need different specialists and different prep entirely.
The one thing that ruins even good preparation
You can do everything on this list and still waste the engagement if leadership disappears once it starts. I have watched a well-prepared business hand over data, brief staff, sort access in advance, and then have the managing director go quiet for three weeks because “the team’s got it.” Decisions stalled. The consultant kept working on assumptions that turned out wrong. The fix, when someone finally answered a question, took an afternoon. The delay had cost nearly two weeks.
Preparation isn’t just the paperwork before day one. It’s staying present enough during the engagement to answer the small questions quickly, because small unanswered questions are where consultant hours quietly disappear.
Frequently asked questions
How long should I spend preparing before an AI consultant starts?
Two weeks is realistic for most small and mid-sized businesses: enough time to sort data ownership, write a specific brief, arrange access, and brief the right staff, without dragging the process out so long that momentum dies before work even begins.
What’s the single biggest mistake businesses make before hiring an AI consultant?
Arriving with a vague goal like “we want to use AI more” instead of a specific, measurable problem, which forces the consultant to spend billable time doing discovery work you could have done yourself for free.
Do I need to fix my data before the consultant arrives?
You don’t need it perfect, but you do need to know where it lives, which version is correct, and roughly how messy it is, because a consultant will otherwise spend early days figuring that out at your expense.
Who should be involved in the preparation, not just the owner or director?
Include at least one or two frontline staff who do the task being examined; they usually know where the real friction sits far better than leadership does, and their input shapes a much more useful brief.
Official documentation
Related reading: How to Prepare Your Small Business for AI Automation (Before You Buy Any Tools) and How to Prepare for an Internship Interview: A Straight-Talking Guide.
Want the complete version? Read where I break down AI marketing.
