3 July 2026 · 5 min read

Before You Buy Another AI Tool, Answer These Five Questions

Most AI spending in small businesses goes to tools nobody uses by month two. Five questions that separate a real fit from an impulse subscription.

There is a pattern we see constantly: a business subscribes to an AI tool after a convincing demo, the team uses it for two weeks, and by month two it is just another line on the credit card statement. The problem is almost never the tool. It is that nobody asked the boring questions before buying.

Here are the five we ask before any tool enters a client's stack.

1. What task, exactly, will this replace?

Not "it will help with marketing". A task: "it drafts the first version of our product descriptions" or "it answers delivery questions in chat". If you cannot name the task, you are buying a capability you have no plan to use. Capabilities do not save time. Replaced tasks do.

Write the task down with a number next to it: how many times per week it happens, and how long it takes today. That number is what the tool has to beat.

2. Who owns it on Monday morning?

Every tool that works has one person who runs it, checks its output, and improves how it is used. Every abandoned tool lacked that person. Before you buy, name the owner. If the answer is "the team will share it", the honest translation is "nobody".

This matters double for AI tools, because their output needs judgment. Someone has to decide what good looks like and keep the tool aligned to it.

3. Does it work with what you already have?

A tool that requires copying data in and out by hand has hidden costs that eat the time it saves. Check the boring plumbing: does it connect to your chat channels, your storefront, your spreadsheet, your storage? A slightly weaker tool that fits your stack beats a stronger one that lives on an island.

4. What happens to your data?

You will be feeding this tool customer messages, pricing, maybe financials. Read enough of the terms to answer two questions: is your data used to train someone else's models, and can you get everything out if you leave? Neither answer should be a surprise later.

5. Have you tried solving this with what you already pay for?

The unglamorous truth: a large share of the tasks that businesses buy new tools for can be handled by the general-purpose AI they already have access to, plus one hour of setup. A written prompt with your rules, your tone, and your examples is a tool. It costs nothing and it is yours.

New tools earn their place when the task is high-volume, the integration matters, or the specialized tool is genuinely better at the job. That is a real bar to clear, and plenty of tools clear it. Just make them prove it against question one.

The pattern behind the questions

All five questions are the same question in disguise: are you buying a solution to a task you have defined, or a feeling of progress? Tools bought as solutions get used. Tools bought as progress get abandoned.

This is the thinking we install in teams during our training programs, and the tool audit is a standard part of the AI Opportunity Diagnostic. For how this plays out in content and creative work specifically, see AI for Creative & Content Teams.