Why the smartest AI strategy starts with a question, not a purchase.

Here’s a classic piece of marketing wisdom: nobody goes to a hardware store because they want a drill. They want a hole. And if you follow the logic a bit further — they don’t even really want the hole. They want a deck. A place to put the grill. A spot to watch summer evenings dissolve into something worth remembering.

The drill is just a means to an end — a very loud, very specific end. The mistake happens when we forget all of that and just buy the most expensive drill we can find, hoping the deck somehow follows.

“Nobody needs AI. They need efficiency — and the gap between those two things is where most implementation plans quietly fall apart.”

Right now, there’s a lot of pressure to “get on board with AI.” Vendors are enthusiastic. Colleagues are enthusiastic. Trade publications are extremely enthusiastic. And somewhere in all that noise, a very practical question gets lost: what, specifically, are we trying to accomplish?

Start with the outcome, not the tool

The most common AI implementation mistake isn’t choosing the wrong product. It’s choosing a product before knowing what problem you’re solving. Organizations end up with shiny subscriptions and underwhelmed employees because nobody paused long enough to ask: where are we actually losing time? What’s the task that drains our afternoons and quietly eats our Fridays?

AI is genuinely remarkable at certain things — drafting and editing, summarizing long documents, answering questions about your data, generating first drafts of almost anything. It is considerably less useful as a vague gesture toward “digital transformation.” The more specific your problem, the more specifically AI can help.

How to actually find the right fit

Before you sign anything, do a little reconnaissance. Here’s a practical approach that tends to work:

  • Run a small pilot first. Pick a handful of curious, constructive people across a few teams. Give them a defined task and something to test. What you’ll learn in two weeks beats what you’ll guess in two months of planning meetings.
  • Find your advocates. In every organization, there are people who are quietly already experimenting. Find them. They know what works, what’s clunky, and where the sharp edges are. They’re invaluable.
  • Check what you already have. You may be sitting on AI tools you’re not using. Google Workspace includes Gemini. Microsoft 365 includes Copilot. If your team already lives in those platforms, start there — the integration is seamless and the price is already paid.
  • Test multiple options against your actual use case. Don’t evaluate tools based on demos. Evaluate them based on whether they make your specific workflow easier. The best tool is the one your team will actually use.

Reason first, product second

There’s a certain kind of buyer’s remorse that’s peculiar to technology: you pay for something sophisticated, implement it earnestly, and then quietly watch it not change very much. The culprit is almost always sequence. The tool came before the problem was defined.

Flip that order. Spend the first conversation with your team asking: where do we feel slow? What work is repetitive, time-consuming, or just kind of miserable? That’s your list. Then — and only then — go looking for a tool that addresses it.

AI in the right hands, applied to the right problem, can be genuinely remarkable. Fewer after-hours email marathons. Faster first drafts. Summaries of the documents nobody had time to read. Answers to the questions that used to require waiting for someone to resurface from a meeting. It’s not magic — but applied with intention, it gets surprisingly close.

“The deck, not the drill. The outcome, not the tool. Start there and the rest becomes a much easier conversation.”

So before the next vendor call, before the next software evaluation, take ten minutes with your team and just ask: what would make next week meaningfully easier? The answer to that question is your AI strategy. Everything else is just shopping.