The research was never the hard part
AI will hand you a hundred frameworks for free. It won’t tell you which lever to pull, whether you actually pulled it, or who owns the result. That’s the part that never got cheaper.
There has never been more good advice available, and it has never been cheaper to get. Ask a model for a pricing strategy and you’ll have twelve of them before your coffee’s cold, each one coherent, each one footnoted, each one probably fine. The research, the part that used to cost a McKinsey engagement, is now basically free.
Worth saying plainly, because a lot of people are quietly panicking that the machines are coming for the thinking. They’re not. They came for the looking-things-up. And looking things up was never the bottleneck in a $10–30M business. I’d go further. It never even made the list.
Three things that didn’t get cheaper
Watch what actually happens when a business stalls. It’s rarely that nobody knows what to do. Something else is missing, and it isn’t information.
- Judgment: which lever is yours. A model gives you a hundred moves. It can’t tell you which one is your constraint this quarter, because it can’t see your five leaders, your onboarding, or the discount that quietly became your price. That call is judgment, and judgment runs on context a general model doesn’t have.
- Accountability: who owns it. A framework has no skin in the game. It doesn’t feel the decision sitting unmade for six months. Somebody has to own the number and carry the cost of being wrong, and you can’t download that.
- The outside eye: who says the true thing. The model agrees with your premise. That’s most of what it does. It won’t tell you the problem is you, or your positioning, or the story you’ve been protecting. You wrote the prompt.
The number was sitting right there
A $20M brand I looked at had an agency reporting 11.7× on ad spend. The whole company believed it. The CEO wanted to buy more traffic and get more of that 11.7×. Notice where the information already was: in the building. The dashboard sat right there. Nobody was short on data.
I reconciled the “winning” campaign against what actually hit the bank. $918 in net sales against $600 in spend. Call it 1.5×, not 11.7×. The store was converting at under 1%, and the email flows that should have caught the buyers who bounced had been switched off during a migration two years earlier and never turned back on. None of that took research. It took someone willing to distrust the number everyone trusted and go check it against the money. That’s judgment, some nerve, and a spot outside the story. You don’t prompt for it.
So no, I’m not worried about the models. Let them get bigger. The more of them there are handing everyone the same hundred frameworks, the more the game comes down to the three things they can’t do: knowing which lever is yours, owning the result, being the one honest set of eyes in the room. That part gets rarer and more valuable, not less. At least that’s how I’d bet it. I could be wrong. But I don’t think I am.
Where is your company leaking revenue?
Twelve questions, five minutes, no email. A read on where to look first.