Summary
By 2029, the tools that once separated big companies from small ones will cost the same for everyone. The power gap won’t move an inch. Five forecasts on what actually changes.
I like making predictions. A lot.
Who wins the Premier League. Elections, any country, I don’t care which one. What the Fed, the Bank of England, the ECB, and the Bank of Japan do next. Years ago, as a young economist, I used to go on TV to talk about this stuff. The central banks, at least, never the football. Nobody asks me anymore. I still do it anyway.
That’s a strange habit for someone in my day job, because clients pay us for certainty, and forecasting mostly means admitting you don’t have any.
I just finished Nate Silver’s book, The Signal and the Noise. He leans so hard on the researcher Philip Tetlock that it sent me back to Tetlock’s own book, Superforecasting, which I’d read years ago and mostly forgotten. Age, right?
Both books use an old story: the fox knows many small things, the hedgehog knows one big thing. If you actually score expert forecasts over time, the foxes win by a mile. But hedgehogs are the ones who keep getting invited back on TV, because one big confident idea sounds better than five small ones held loosely.
I know that game well. I used to play it. I was a hedgehog. Now I’m trying to be a fox.
And pricing, as an industry, is full of hedgehogs. Everyone’s got their one answer. Raise the price. Cut the underperformers. Fix the promo calendar. Buy the tool.
But here’s the thing that stuck with me more than the fox and hedgehog story. Most expert predictions are written so they can never be checked. “A significant opportunity exists to optimize price architecture.” You can’t ever prove that sentence wrong. That’s exactly why people write it that way.
Go back and count how many claims in the last pricing deck you got could actually be marked right or wrong three years later.
Silver draws a line I want to borrow. A prediction is definite: the earthquake hits Tokyo on Tuesday. A forecast comes with a probability and a timeframe: 60% chance of a major earthquake in southern California in the next 30 years. Seismologists live by that distinction. The USGS forecasts. It doesn’t predict.
So here’s one prediction and five forecasts. The prediction comes first, because it’s the one I’d bet the firm on.
By 2029, the gap in analytical ability between a $300M consumer goods company and a $30B one will basically disappear. The gap in negotiating power will be exactly the same as it is today.
No percentage attached to that one, which under my own rules means it can’t be scored. I’d rather admit that than dress it up. It’s an argument, not a bet, and the next two sections make the case.
Why anything changes at all
The math behind pricing has been around for 60 years. Airlines figured most of it out in the 1980s. So when a $300M food company prices badly, and most of them do, it’s not because the formulas weren’t available.
It’s because good pricing decisions used to eat up weeks of an analyst’s time. So companies only made the expensive calls when they absolutely had to, and the really expensive ones, like redesigning your whole price and pack lineup, just never happened. That’s where the once-a-year price increase came from. That’s where the promo calendar that’s a copy-paste of last year’s, locked in six months ahead, came from. Nobody sat down and designed that rhythm. It’s just what analysis used to cost.
That cost has dropped a lot. But the calendar built around the old cost hasn’t moved. Plenty of companies got new tools that make constant, up-to-date decisions affordable, and used them to do the exact same once-a-year price increase, just a bit faster.
Which gap actually closes
For 40 years, you had to be big to afford real revenue management. A data team. Market research in every region. Specialists. A project costing seven figures. Airlines had it. $30B consumer goods companies had it. The $300M food manufacturer didn’t, even though every dollar mattered more to them.
Take away that upfront cost, and the picture flips. But I want to be careful here, because the easy version of this claim is wrong.
The gap in analytical ability nearly disappears. Pricing sensitivity by retailer, ROI on every single promotion, simulators for your whole price and pack lineup, real net price by customer. A $300M company can run all of that today.
But the gap in negotiating power doesn’t close at all. A $30B company can hold its ground in a fight with a retailer. A $300M company gets its product pulled from shelves.
And in case that sounds like good news only for the small guys: the big companies were never held back by analytics either. Their problem is process and approvals. Forty people who need to sign off, and a planning calendar that can’t be interrupted for something as “minor” as new information.
Two limits, or this is just a sales pitch
The retailer controls the pace. You can’t change shelf prices every week in a grocery store, because you don’t own that shelf. So none of this is about changing prices constantly. It’s about being ready the moment a window opens.
Also, run the same promotion often enough and shoppers stop buying at full price. They just wait for the discount. Each individual promotion still looks like it clears the bar for a good return, because that bar is measured against a baseline that your own past promotions have been quietly dragging down for two years. So a model that checks each promotion one at a time keeps saying “yes, do it,” right up until there’s no full-price business left to protect. The first real AI pricing failure in this industry won’t be a broken model. It’ll be a model that got every single promotion right and still ran the business into the ground.
Five forecasts
Each one has a date and a probability I actually believe, not one that just sounds confident.
- By 2028, a $300M consumer goods company runs pricing at the level of sophistication a $3B company runs today. (90%) This is the floor. It’s already true at a few companies we work with.
- By 2029, most mid-market food and beverage companies will make a major price change outside their normal yearly cycle, driven by a model instead of a customer negotiation. (60%) This is the one I’ll actually track.
- By 2030, a major consumer goods company publicly walks back an AI pricing rollout, and the model gets blamed when the real issue was bad data. (70%)
- By 2029, most companies that buy one of these tools won’t have changed a single approval step to actually make room for it. (40%) I kept this under 50% because I’d rather be wrong. But it’s the one that would make everything else in this piece pointless.
- By 2029, “AI-powered pricing” stops being something you can lead with in a consulting pitch, the same way “cloud-based” stopped working. (85%) That one’s aimed at my own firm too. The AI was never the real advantage.
On scoring these
Five forecasts don’t tell you whether I’m actually good at this. You’d need a much bigger sample size to know that. So don’t treat this like science. Treat it like accountability.
I’ll publish exactly how each one will be judged, check in every year instead of waiting until 2029, and lead with whatever I got wrong.
Where I think this lands
Hedgehogs will keep doing this once a year and call it careful thinking. Foxes will be wrong more often, but in smaller, cheaper ways, and they’ll adjust faster. Over three years, that’s not a close race.
Ask me again in 2029.
