Say No to Grow, one year on

AI Review

Twelve months ago we argued the brands that win are the ones doing fewer things, properly. AI hasn’t changed that. It has just made the line much harder to hold.

A year ago we published a piece called Say No to Grow. The argument was not complicated. The brands that win are rarely the ones doing the most. They are the ones doing fewer things, properly. Fewer SKUs, managed well. One marketplace, run hard. A range the merch team can actually forecast. We pointed at Liquid Death turning down the deals that didn’t fit, and we made the case that strategic refusal is a competence, not a weakness.

I still believe every word of that. But a year is a long time, and this one has put the theory under real pressure. Not because the theory is wrong. Because saying no has got harder.

What’s changed isn’t the cost

People will tell you AI has made everything cheaper. I would be careful with that. Some jobs are quicker, the savings are real but smaller, and someone still has to check the work. Cost is not really the thing that has changed.

What has changed is access. AI is more available than it has ever been, and the barrier to entry has dropped through the floor. A year ago, taking on something new meant a business case, a budget, maybe a hire, a project with a start date. Today, anyone with a browser and twenty minutes can have a go. That sounds like progress, and in places it is. But it has changed the conversation inside your business, and not in your favour.

The boss, the colleague, the competitor

Here is how it shows up.

Your CEO has noticed. The next time you push back on taking something on, the next time you say we do not have the capacity for that, you will get the same question back. “Can’t AI just do it?” On the surface it is a fair question. It is also very hard to say no to in a meeting, because the honest answer has caveats, and caveats lose to confidence every time.

Someone on your team has noticed too. There is at least one person in merch, in retail, in ecom, who has been experimenting. They have built something. And they will show you. “Look what I made.” And it will look good. It will look like proof that this is easy now, that the thing you called a quarter of work is really an afternoon.

And your competitors have noticed. They are using AI as well, or at least saying loudly that they are. So now there is a third voice, the one that says if we do not do this we fall behind. The temptation to stay ahead by saying yes to everything is strong, and it is exactly the temptation Say No to Grow was written to warn you about.

Three pressures. The boss, the colleague, the competitor. None of them existed in this form a year ago, and all of them push the same way. Say yes. Do more. AI makes it possible.

Accessible is not the same as good

So here is the thing I most want you to take from this piece.

A low barrier to entry is not the same as a high standard of output. They are two completely different things, and the last year has blurred them on purpose.

Accessible does not mean good. The fact that anyone can now generate a forecast, a report, a product feed, a campaign, does not make any of those things right. The colleague who showed you what they built showed you a demo. A demo is not an operation. The version that looks impressive on a Friday afternoon, on one tidy example, is not the version that holds up across four thousand SKUs, a messy catalogue, a real trade meeting and a peak you cannot afford to get wrong.

AI has made it easy to start things. It has done almost nothing to make it easy to finish them well. And “well” is the entire game. It always was.

This matters more now, not less

This is where the original argument gets sharper, not weaker.

Because AI is now so accessible, everyone in your business can point it at the operation. That cuts both ways, hard. Point a capable AI system at a focused operation, one with a tight range and product data that is clean and consistent, and you get real leverage. A small team punches well above its size. Point that same system at a sprawling one and you do not get relief. You get the mess, faster, and now produced by more people, because the barrier to producing it has gone.

AI is a multiplier. It does not care what you point it at. Accessibility just means more hands on the multiplier. If your operation is in good shape, that is a genuinely brilliant thing. If it is not, you have handed everyone in the building a faster way to make it worse, and a cover story that they were being productive while they did.

Picture two brands in the same category.

Similar revenue, similar headcount.

The first spent the last two years saying no. They held their range tight, killed the SKUs that never earned their shelf space, ran one marketplace properly instead of three badly, and kept their product data clean because they had less of it to keep clean. Nobody writes case studies about the SKUs you didn’t launch.

The second said yes to everything. Every channel, every territory, every range extension a supplier waved at them. More ambitious on paper. More going on.

Now AI becomes accessible to both of them, on the same day, to every person in both teams.

The first brand gets leverage. The second gets a faster, more democratic way to be confused. Same technology, available to everyone, completely different outcome. And the outcome was decided years before either of them typed a single prompt. It was decided every time one of them said no and the other said yes.

We have written before about product data being the real differentiator for agentic commerce, and that argument has only sharpened. When an AI agent is doing the shopping, your product data is your shop floor, your sales assistant and your window display all at once. Brands with clean, focused data are about to look very good. Brands with sprawl are about to find out exactly how much of it they were carrying, because the AI will surface every contradiction in it, at speed, in front of a customer who is really an algorithm.

Saying no is now quality control

So no, Say No to Grow has not aged out. The reason behind it has changed, and the new reason is the one that matters.

A year ago, saying no protected your team’s time. Today it protects something harder to see. It protects you from mistaking activity for progress. The CEO’s question, the colleague’s demo, the competitor’s noise, none of those is evidence that you should do the thing. They are evidence that the thing is now easy to start. Easy to start has never been the bar. Easy to start is how brands end up with four systems and a trade meeting nobody trusts.

Saying no is now an act of quality control. It is you, the operator, holding the line on the difference between what is possible and what is good, at the exact moment everyone around you has stopped seeing the difference.

The question for this quarter

So here is the question I would put to your leadership team this quarter, the next time someone asks why you are not doing more with AI.

Not “can AI do this.” Almost certainly it can, in some form, badly, on a good day. The better question is whether it will hold up across your whole operation or only in the demo. And the harder one: if we pointed the best AI available straight at our business tomorrow, would it make us faster, or just make our mess move quicker?

If you are not sure of the answer, that is not an AI problem. It is a focus problem. And the fix is the same as it was a year ago. Decide what you are going to be genuinely excellent at, and say no to the rest. No matter how easy the rest has become to start.

Grow by doing less. It was true last year. AI has only made it harder, and more important, to actually do.

This is what OX is for

Everything in this piece, the pressure to say yes, the demo that does not survive a real catalogue, the question of what good actually looks like, is exactly the sort of thing we get into at OX.

OX is the community and event series we built for operators. Finance, merch, commercial, ops. The people who carry the weight of every yes, and who have spent years underserved by an events scene built for everyone but them. It came out of more than sixty conversations with brand-side leaders, and it is deliberately practical. Real operators, real problems, no slide decks pretending it is all simple.

If you are the one being asked “can’t AI just do it”, come and have that conversation with people facing exactly the same thing.