In 1900 there were fewer than 3,000 cars in France, and two brothers in Clermont-Ferrand had a tire company with an obvious problem. Tires only wear out when people drive, and almost nobody drove. So André and Édouard Michelin printed a small red book and gave it away. It told a motorist where to fill his tank, repair his car, sleep and eat. The commercial logic never needed to be printed: a driver who knows where the next good dinner is will drive to it, and every mile of that drive sells a little rubber.
The guide was excellent, and that is the part worth sitting with. It was not a trick or a coupon. Much of it was assembled from the reports of the company's traveling tire salesmen, who spent their days on those roads and knew them better than anyone, and it became so trusted that a star in it can still fill a dining room for a season. The advice was genuinely good, and it was written by the people who profited when you took it. Both things were true at once, and for the hungry traveler there was no conflict, because the traveler wanted to drive and the company wanted him to drive.
The Adviser Moves Into the Car
Last week Meta reported its third quarter, and one sentence in the call deserves a second reading from every marketing leader. Advertisers who adopted the recommendations of its AI business assistant drove almost 5% more conversions at the same cost per conversion and increased their advertising investment by 9% compared with similar advertisers who did not. The company is giving that assistant more of the wheel: it will plan, create and optimize campaigns and schedule tasks through a conversation, with previews and approvals. Nearly 10 million advertisers used its generative creative tools in the quarter, according to AdNews.
I do not read that sentence as a scandal. I read it as a guide. The conversions are probably real; the platform sees more of the auction than any advertiser ever will, and a recommendation to broaden a target or test a new format is often the right call. But notice which number sits beside the conversions. Not profit, not pipeline, not the advertiser's margin: investment. The adviser's success is reported in the same unit as the platform's revenue, because for the platform those are the same thing. More conversions at the same cost is a better trip. Nine percent more spend is more miles.
Every Mile Has a Price
The miles are not getting cheaper. In the same quarter Meta reported that ad impressions rose 10% and the average price per ad rose 14%, on revenue of $64.6 billion, up 26%. Its full-year capital spending range now sits between $135 billion and $145 billion. A company investing at that scale does not build an adviser to help you spend less. It builds one to help you spend more with fewer regrets, which is a real service and a different one.
This is the economic shape worth naming plainly. When the seller writes the guide, the guide optimizes the trip inside the seller's map. It will find you the better restaurant along the road. It will not tell you that you did not need the road at all: that the customers it is so efficiently converting were already going to buy, that the next dollar would earn more on your pricing page than in the auction, or that the quarter's real constraint is sales capacity rather than demand. Those questions sit outside the map, not because anyone hid them, but because nobody at the tire company is paid to ask them.
Keep Your Own Map
So the job is not to refuse the guide. Refusing good advice out of suspicion is its own kind of waste, and the advertisers who ignore these tools will mostly pay more for worse results while feeling principled about it. The job is to carry a map the guide did not draw, and to choose the destination before the conversation starts.
That map begins with the number you are actually buying, whether that is qualified pipeline, contribution margin or revenue from customers who were not already yours. When the adviser reports success in conversions and spend, the recommendation has to be translated into your number before it is accepted, and a recommendation that cannot be translated is unproven rather than wrong. The map also carries a ceiling, and the ceiling is set in a budget meeting by people looking at the whole business, not in a chat window by software that is rewarded for finding reasons to go further. A sensible traveler decides how far to drive before opening the book, not after the third recommended detour. And the map includes a test the platform does not grade: a region, a segment or a stretch of weeks held out on purpose, so that you can see what happens to your revenue, not your attributed conversions, when the spend goes away. If the answer is very little, the guide was showing you good restaurants on a trip you did not need to take.
None of this is adversarial. The best relationship with a platform adviser is the one a seasoned buyer has with a sharp salesperson on commission: listen closely, because they know things you do not, take the good routes gladly, and never let them choose where you are going or how far.
The Book on the Workbench
There is a later chapter in the Michelin story. The guide stayed free until André Michelin saw a copy propping up a bench in a garage, and the company started charging for it, on the theory that people respect what they pay for. Free advice gets used as furniture.
The platform advisers are free, and so they are easy to treat the same way: ignored until a campaign stalls, then obeyed because they happen to be open. The better posture is to treat them as worth what you would pay a well-informed consultant whose fee rises with your spend. That consultant can be enormously useful. That consultant should never be the only one at the table who knows where the trip is supposed to end.
The guide was written by the tire company. It was a very good guide. The person driving still decides where the car is going.
