Every small business we audit is paying for software it does not fully use, leaning on knowledge that lives in one person's head, and wondering whether AI is supposed to fix any of it. You do not need a consultant to start sorting out which of those is true for you. You need five questions and an honest hour.

This is that hour. Five questions, in order, from the one you can answer in thirty seconds to the one that takes real reflection. Answer them in sequence. Write the answers down. By the end you will have a rough map of where your money is leaking, where AI might actually help, and where it would be a waste.

This is not a quiz with a score. It is the same opening sequence we use when we sit down with a new client in the Portland metro or Southwest Washington, minus the parts that need us in the room.

Who this is for

If you have a full IT department and a software budget someone actively manages, this is too basic for you. It is written for the owner-operator running a business of roughly one to fifty people, no dedicated IT person, and a stack of subscriptions nobody has reviewed in a year. If that is you, keep reading.

The order matters. The questions move from concrete to abstract on purpose. The first one is something you can answer off your bank statement. The last one requires you to be honest about how your business actually works. Do them in order and each answer feeds the next.

Question 1: What software are you paying for that you cannot remember the last time you logged into?

Start here because it takes thirty seconds and it almost always finds money. Pull up your last three card statements or your accounting export and read the recurring charges line by line. The tool you bought for one project two years ago. The seats you never reduced when someone left. The two products that do the same thing because two different people each picked their own and neither knew about the other.

The dollars are real, and they are usually the cheapest win available to you. But the bigger point is what an unused subscription tells you. It is a decision that made sense once and never got revisited. Where you find one, there are usually two more nearby.

What to do with the answer: cancel the obvious dead weight now, this week. Flag the maybes, the tools you are not sure about, and carry them into question four, because the reason they are still on the bill usually lives there.

Question 2: What task does someone on your team do every week that they personally hate?

This is the AI question, and it is phrased this way deliberately. Not "where could AI help," which is too abstract and where most people guess wrong. The better question is what someone on your team hates doing every single week.

Hated, repetitive work is the best early candidate for AI, for two reasons. The fit is usually real: repetitive tasks with a clear input and a clear output are exactly what these tools are good at. And the adoption is willing. People do not fight you when you offer to take the worst part of their week off their plate. They help you make it work.

Good candidates look like this: drafting the same handful of customer emails over and over, summarizing inbound messages, reformatting data between two systems that do not talk to each other, producing the first draft of a recurring document.

Here is the trap. Some people hate the parts of their job that require judgment, hard conversations, or decisions. AI does not help there, and pointing it at that kind of work is how you end up with a tool nobody trusts. The test is not "is this hated." It is "is this hated and repetitive and mechanical." Repetitive points toward AI. Judgment points away from it.

Question 3: If your most experienced employee left tomorrow, what would you actually lose?

This surfaces tribal knowledge, the way of doing things that lives in one person's head and nowhere else. It is the highest-stakes operational risk most small businesses carry, and the one they think about least, because the person is still here and everything still works.

Sit with the answer honestly. If it is "a few passwords and some contacts," you are fine. If it is "the whole way we handle X," you have a documentation problem, and you have it now, before you have any AI problem at all.

The good news is that this is fixable, and it is one of the places AI genuinely earns its keep. Capturing a process, turning one person's method into written steps, building the documentation nobody has ever had time to write: that is real, useful work these tools do well. We are not going to tell you every business needs to drop everything and address this. But in the ones where a single person quietly holds the keys to how the place runs, it is usually the first thing we would protect.

Question 4: Who in the business decides what software gets bought, and who has to live with that decision?

This is the misalignment question. When the person who buys the tools and the people who use them are different, you get shelfware: software bought by someone who liked the demo, used by people who were never asked whether it fit their work.

Remember the maybes you flagged in question one? This is usually where they came from. The unused subscription is the symptom. The gap between buyer and user is the cause.

The answer also tells you something useful going forward: who needs to be in the room the next time you change anything. If your front-desk staff will use a tool every day, they should have a say before you buy it, not a surprise after. Fixing this one is not about software at all. It is about who you ask before you spend.

Question 5: When you imagine "using AI" in your business, whose job changes?

This is the honest question, and it is the one most "we should use AI" conversations never actually reach. "Using AI" is not a plan. It is a slogan until you can name a specific person, a specific task they do, and how their week is different afterward.

So try to name it. Whose job changes? What part of it? What do they do with the time that comes back?

If you cannot answer, that is not a failure. It is information. It means you are earlier than you thought, and the right next step is not buying a tool. It is watching your own operation for a week with these questions in mind. If you can answer it, with a real name and a real task, then you have the start of an actual project rather than a slogan.

What to do with your answers

Group them. Questions one and four are usually a software cleanup you can run yourself: cancel the dead weight, fix who you ask before you buy. Questions two, three, and five are where AI either earns its place or does not, and they are worth more thought than a single sitting gives them.

If your answers point somewhere specific and you want a second read on whether AI actually fits, the first conversation is free. Book a 30-minute call. No pitch, no upsell pressure, and we will tell you plainly when the answer is that you do not need AI for this.

If you would rather have these five questions as a one-page PDF to take to your next strategy meeting, grab it below.

Five questions every small business should ask about their software, IT, and AI spend