AI for small business: the realistic playbook

A small business does not need an "AI strategy". It needs one thing to stop eating its day.

Most of what gets sold to small businesses was designed for large ones and then discounted. This list is the opposite: built small from the start, and ordered so each system funds the next.

A friend of mine, and his print shop#

A few weeks ago a friend opened a small print shop. He came to me with two problems.

He needed a graphic designer. And he needed a subscription to invoicing software.

The third problem, which he did not say out loud, was that he could not afford either.

I suggested one subscription at twenty dollars a month. We sat together for two hours. We installed the app, and I taught him the basics of writing instructions. That was all.

We came out with a working invoicing app that he built himself without knowing a line of code. And he now designs his own posts.

The most important number in that story is not the twenty dollars. It is zero: his prior knowledge of programming, and his prior knowledge of design.

The order, and why it matters#

First: repeated replies. Eighty percent of your customer messages ask ten questions. Collect them, write final answers once, then make the reply instant. This saves more than anything else on the list and needs almost no technology.

Second: quotes and invoices. Turning a conversation into a formatted quote. The task business owners postpone most, and the main reason payment is slow.

Third: marketing content. After the first two, because marketing creates demand and demand you cannot answer is worthless. Details in marketing without losing your voice.

Fourth: connecting your tools. Getting the website form into your calendar and your WhatsApp without copy-paste. Start at your first automation in one hour.

Fifth: analysis. After the previous four have made your data tidy. Whoever starts here analyses a mess.

What costs more than it saves#

A tool for every problem. Five monthly subscriptions cost more than a part-time employee and need maintenance nobody on your team owns.

Automating an unstable process. If the process changes monthly, automating it means rebuilding it monthly. See when not to automate.

A chatbot too early. Before your answers are written down and organised. Details in a chatbot for your business.

A project with no owner. Any system without a named person responsible stops within two months.

The calculation that decides#

Before any tool, work out three numbers:

The second row is what gets forgotten. Monthly review hours are never zero in any system, and assuming they are produces a surprise in month three. There is another month-three surprise in token economics.

What this means for you#

If you own the business: start the first system this week. Pull a month of messages and count the repeated questions.

If you run a small team: assign each system to a named person before subscribing to anything.

If you sell to small businesses: the order above is the order in which they become willing to pay.

In closing#

A small business wins through sequence, not ambition. One system that works and saves hours funds the next; five systems at once all fail.

Try this today: open the last month of messages and count how often the same question appears. That number is your first system.

Common questions#

Where should a small business start with AI?

With repeated replies. Eighty percent of customer messages ask ten questions, and writing final answers to them once saves more than anything else while needing almost no technology.

Why not start with data analysis?

Because analysis needs clean data, and the four systems before it are what produce it. Starting with analysis is building the fifth floor first.

Which cost does everyone forget?

Monthly review hours. They are never zero in any system, and assuming they are means discovering in month three that the saving is far smaller than the plan said.

How long does this take to deploy?

Ninety days is enough for all five, on one condition: one at a time. Deploying them together is the most common reason small business projects fail.