The first thing I tell every business owner who comes to us about AI is this: don’t buy anything yet.

That probably sounds strange coming from someone who sells AI services. But I have seen what happens when businesses skip this step, and it is not pretty. They buy something, they deploy it, it does not match their actual problem, and six months later they are sitting across from me explaining why AI “didn’t work” for them.

It did not work because they started with the solution. We start with the problem.

The three questions we ask first

Before we talk about any tool, any service, any technology, we ask every new client three questions. The answers tell us more about what a business actually needs than any demo or proposal ever could.

Question one: What is the most expensive repetitive task in your business?

Not the most annoying one. Not the one your team complains about most. The most expensive. That means time multiplied by volume multiplied by the cost of getting it wrong.

A dental practice spending two staff hours a day on appointment confirmations and reminders, that is expensive. Not because each confirmation takes long, but because it is two hours, every day, from trained staff who could be doing patient-facing work.

A law firm where a paralegal spends three hours on initial intake calls every week, capturing the same information in the same format for every new inquiry, that is expensive. Not because intake is complicated, but because it is a process. And processes can be systematized.

When I ask this question, most business owners give me the wrong answer first. They tell me about the thing that frustrates them most. Which is usually not the most expensive thing. So I push: if you had to put a dollar figure on it, what is actually costing you the most?

That answer is almost always where we start.

Question two: Where are you losing customers to slow response times?

This one hits differently. Because most business owners know the answer, but they have made peace with it.

“We miss some calls on busy days.” “Leads that come in after hours don’t always get a same-day response.” “We’re slow getting back to quote requests sometimes.”

Each of those is a version of the same thing: a customer who was interested, who reached out, who then waited, and who may have moved on by the time you got to them. That is not an operational inconvenience. That is revenue walking out the door.

I was talking to a plumber last year who had a handle on his conversion rate from answered calls (around 60%), but had never thought about what was happening with the unanswered ones. He was missing roughly eight to ten calls a week. At an average job value of $400, and assuming half of those callers would have booked, that is potentially $800 to $1,000 in lost revenue every week from calls that just went unanswered.

He had been running his business for seven years and had never quantified that number. Once he did, the conversation about AI phone handling took about four minutes.

Question three: What report or insight do you wish you had every Monday morning?

This question reveals a lot about where a business is in its data maturity, and what the right lever is.

Some business owners immediately have an answer. “I want to know which clients are at risk of churning based on their activity.” “I want to see our weekly revenue by service line without having to pull it manually.” “I want to know which of my team members is generating the most billable hours.”

Those businesses are ready for the data work. They know what they want to know. They just do not have the infrastructure to surface it automatically.

Other business owners stare at me blankly. “I don’t know, I look at whatever my system shows me.” That is also useful information. It tells me that data visibility is not the urgent problem. Something else is.

The answer to question three tells me whether we are talking about team skills, data infrastructure, or operational agents. And it usually confirms or redirects what the first two questions surfaced.

Why these questions matter more than “what tool should I buy”

The AI tooling market is crowded and moving fast. There are hundreds of platforms, dozens of approaches, and no shortage of vendors who will tell you their product solves everything.

The problem is that most businesses shopping for AI tools have not clearly articulated what they are trying to solve. They know they should be “doing something with AI.” They have a vague sense of where the friction is. And so they buy something that sounds vaguely relevant and hope it works.

It usually does not. Not because the tool is bad, but because there was no specific problem matched to a specific solution. The fit is wrong.

The three questions get you to a specific problem. And a specific problem is something you can actually solve.

The mistakes we steer people away from

Over the past few years working with businesses on AI implementation, we have seen the same mistakes repeat themselves. Here are the ones I see most often.

Shiny tool syndrome. Someone saw a product demo, or a competitor mentioned a new tool, or LinkedIn showed them an article about the latest AI platform. They buy it before they have assessed whether it solves a real problem in their specific business. The tool is often impressive. But impressive is not the same as useful.

Trying to automate everything at once. A business decides it is “going all in on AI.” They identify fifteen processes to automate simultaneously. They build nothing properly because they are spreading attention across too many things. Six months later they have fifteen half-built automations and nothing running cleanly.

The businesses that get real results start with one thing, run it properly, and then expand. One agent, deployed well, working consistently, is worth more than fifteen agents that kind of work.

Ignoring team readiness. This is the one that costs the most. A business invests in AI tooling without preparing their team to work alongside it. The tools get deployed but the team does not trust them, does not use them properly, or actively works around them. Nothing changes.

Team readiness is not about training courses. It is about starting with something that makes your team’s lives easier, demonstrating that it works, and building confidence from there. The first AI deployment in any business should make someone’s day noticeably better. That is what creates buy-in for everything that follows. Before any deployment conversation, it helps to be clear on what actually changes for a team when AI comes in — and what stays the same.

The right order

We recommend three phases, in this order.

Start with advisory. Before you spend money on tools or services, spend time getting clear on what problem you are solving. What is the highest-value thing to address first? What does success look like? What does your current process look like, and where are the gaps? This does not need to take months. A focused conversation often gets you there in a few hours.

Then pilot. Pick one problem, deploy one solution, and measure what changes. Not a proof of concept. A real pilot with real volume, real output, and real measurement. You want to know whether this works in your business, with your team, for your customers. Not whether it worked for someone else.

Then scale. Once you have a working model, replicate it. Expand the use case, add additional processes, bring in more of your team. Now you have a foundation that you know works, and you are building on something solid instead of guessing.

Most businesses try to skip straight to scale without doing advisory or pilot properly. The ones that follow this order end up with AI that actually works in their business.

What a first engagement with us looks like

I want to be direct about this because I think a lot of people assume there is a big commitment involved before they can have a useful conversation.

There is not. A first call is genuinely just a conversation. No pitch deck, no proposal, no sales process. We ask you the three questions. You tell us what is actually going on in your business. We tell you honestly what we think the right first move is.

Sometimes the right first move is working with us. Sometimes it is something much simpler that you can do yourself. Sometimes it is doing nothing yet and getting your processes documented first.

We would rather give you the honest answer than sell you something that is not right for your situation. The businesses we work with best are the ones that trust our advice even when it means starting slowly.

If that sounds like a conversation worth having, book a call. We can usually get you clear on the right first step in under an hour.

Book an Omni Advisory call

Related reading: 3 AI investments that pay off in year one, are you actually ready for AI agents?, AI automation vs an AI workforce — the difference matters, the complete business owner’s guide to AI agents, and how to know if your AI is actually working once you’ve deployed it.