Where does AI actually fit into your business?
The three lists to write first
A short guide for leaders of consumer businesses that already work. No tool comparisons, no vendor names. Three lists, one page, and a way to decide where AI can make the greatest commercial difference.
You know your business. The people selling AI do not.
Someone is in your inbox every day with a tool that is going to change everything. Your feed is full of it. Half of it contradicts the other half, and a good deal of it comes from people who have never run a business like yours.
They know their tool. They do not know your customers, your operations, or the decisions that drive your margin. And every one of them leads with the same thing: what the tool does, what it integrates with, what it costs.
You cannot judge any of that until you know what you want it to do.
That is the whole problem. You are being asked to evaluate answers before anyone has helped you write down the question.
You have probably tried something already. A tool, two systems joined up, somebody brought in to sort out the reporting. It half worked. The payoff never matched what went into it, and most of what went into it was time, from people who did not have any spare.
That is what starting in the wrong place costs, and it is not the licence fee.
So the decision gets deferred
Which is a reasonable response. Doing nothing feels safer than buying the wrong thing twice, and things are going all right.
The trouble is that the noise does not stop. Meanwhile the processes that used to work become harder and more expensive as the business grows.
This guide will not tell you which tool to buy. It gives you three lists to write instead, and they take an afternoon. What you end up with is one page, in your own words, saying what you want the business to do differently. It lets you tell the difference between an interesting AI idea and one that could genuinely improve customer experience, create capacity or unlock revenue.
The customers you have already got
What would make them worth more to you?
Almost every AI conversation starts at acquisition, because that is where the loudest tools are. In an established consumer business, the money is often sitting somewhere less exciting: in the customers already in your database.
So write down the things that would get them back sooner, spending more, or staying longer.
- Suggest the eyeshadow to the person who has just bought the lipstick, without somebody choosing it by hand.
- Send people recommendations based on what they have actually bought, rather than the same email to everyone.
- Remind people to reorder the thing they buy every eight weeks, at week seven.
- Spot the customers who are about to go quiet, two weeks before they do, and have something to send them.
- Answer “where is my order” without a person opening four screens.
What a good item looks like
Something you could do, that happens often, and that a customer would notice. “Better marketing” is not an item, and neither is “our returns are slow”. Turn the complaint into the thing you would do about it.
Write more than you think is reasonable. You are not committing to any of it yet.
The manual work
What eats up your week, or your team’s?
This is the easiest list to write and the one most teams underestimate, because the work has been there so long it has stopped looking like work. It looks like Tuesday.
- The stock report somebody rebuilds by hand every Monday, out of two systems that do not agree.
- Somebody exporting orders into a spreadsheet to work out who has not bought in a while.
- Four systems opened to answer one customer email.
- The same reply written from scratch, again.
- The invoice keyed in twice because the two systems do not speak.
- Stock counted in one place and recorded in another.
- Digging out the same email again to check when the new stock lands.
Put a rough number next to each one
Hours a week, to the nearest half day. Do not go and measure it. A rough number changes the conversation, because four hours a week is a fortnight a year, and most leadership teams have never said that sentence out loud about one process.
One thing to watch
Some manual work is judgement wearing overalls. If a person is quietly making a call while they do it, deciding what to reorder, deciding which order to chase, then the real item belongs on list three. Automating a decision nobody has examined only gets you to the wrong answer faster.
The decisions you make on a hunch
The ones where, honestly, it is last year plus a feeling
This is the shortest list, the hardest to write, and worth more than the other two put together. It gets ignored because these do not feel like processes. They feel like the job.
- How much of each line to buy for the season.
- How many seasonal staff to bring on, and when to start them.
- Which lapsed customers are worth chasing, and which have simply gone.
- Whether to refund or repair, when it is borderline.
- Which supplier is quietly costing you money.
- What to discount, when, and by how much.
If the list comes out short, try it from the other end. Which decision would you least like to justify, with evidence, in a room? Which one, if you were out by twenty per cent, would you still be feeling at the end of the year? And which do you make on gut and get right most of the time, because usually right means there is a pattern, and a pattern is something that can be evidenced.
You are not writing down what is wrong with your judgement. You are writing down where your judgement is currently unarmed.
Choose the opportunity worth acting on first.
You have got a page. You are not going to do all of it and you should not try. Circle the one item that would matter most if it were true this year.
What the page cannot tell you is how to actually do that one thing: what would have to be joined up first, what stops needing a person, and where AI genuinely earns its keep. There is an order to those, and taking them in the wrong one is the single most common reason these projects disappoint.
Bring your page to a focused first conversation
Bring the opportunities your team can already see, including the one you circled. We will pick the strongest one and agree what is worth acting on first. If AI can make a meaningful difference, I will tell you. If it cannot, I will tell you that too.
Immy Payton
I have spent the last decade where data, technology and commercial decisions meet, including nearly six years at Klaviyo, where I progressed from Senior Data Scientist to Senior Manager.
Alongside that, I helped move my family’s cashmere business online and built the systems that enabled it to grow from around £200k in annual revenue at fairs to a full year 2026 forecast of £4m. Payton Labs brings those two sides together inside ambitious consumer brands.
You know where the brand is going. You want AI to help it get there.
You lead an ambitious consumer brand with a strong point of view about where the business is going. Your team can see the processes that are becoming harder to run, and you want an experienced partner who can turn one of those opportunities into something that works inside the real business.
You bring the direction and the knowledge of your business. I bring the technical judgement and hands on execution to close the gap.