The two ways small businesses get AI wrong

There are two ways to be wrong about where your business stands on AI, and from the inside they feel identical. They both feel like knowing.

The first: further ahead than you are

The pattern is consistent enough to be recognisable.

• There is a subscription, possibly two, one of which nobody can remember signing up for.
• Three or four people use it, mostly to tidy up emails and make things sound a bit more formal.
• Nothing is connected to the systems the business actually runs on — the bookings, the quotes, the customer records.
• The owner does not really use it.
• The word “agent” appears in meetings, and if you stopped the room and asked for a definition you would get five different ones.

That business believes it has adopted AI. What it has adopted is a vocabulary. The tell is that if you switched every subscription off on a Monday, the only thing that would break is people’s email tone.

I am not sneering here. It is a completely reasonable place to end up. Nobody sat down and decided this — it accumulated, the way most things in a busy business accumulate.

The second: further behind than you are

The mirror image, and I meet it just as often. A firm that has genuinely automated something real — a quoting process, a follow-up sequence, a reporting job that used to eat a Saturday — and is convinced it is a decade behind, because they spend twenty minutes a day looking at other people’s published highlights.

Nobody posts the workflow that half works. So the comparison is always your ordinary Tuesday against everyone else’s best day, and it is not a fair fight. This is the more corrosive of the two, because it stops people starting at all.

Both come from the same place

You can only benchmark against what you can see. What you can see is your own business, the four competitors you happen to know, and whatever an algorithm decided to show you this morning. That is not a sample. It is a keyhole.

This framing — that misjudging your position is fundamentally a visibility problem, and the fix is measurement from outside — is one Allie K Miller makes about enterprises, and it travels down to small businesses completely intact. Arguably it lands harder, because a large company at least has departments who can disagree with each other. You have your own head and a busy week.

So measure something

Firstly, pick criteria that do not move — the same questions asked the same way, so this month is comparable to next. Secondly, make it external, because the point is to see past your own keyhole. Thirdly, keep it fast; a measurement that takes a fortnight will not be repeated, and one you never repeat tells you nothing about whether you are improving.

That is what we built our free audit to be: fixed criteria, applied from outside, in about a minute. It scores how manual your operation still is, reads your homepage the way a stranger would, and puts real competitors beside you.

And because I would rather you trusted the number than took it on faith, the scoring rules are published — the weights, the bands, and what happens when something cannot be measured. Overall: you do not need to guess where you stand. Guessing is the expensive part.