I imagine myself trying to buy a Jeep Wrangler.
At the first dealer lot, the salesman wants to know what I do for a living. Then what kind of driving I do, whether there are kids, whether there's a dog. I came to look at cars, and he hasn't shown me one.
He's doing his own little version of ethnographic research. He's also sold a lot of cars to guys who look like me, and he has a pretty good idea what I'm going to care about before I do.
So when we get to the car, he barely mentions the seats. (Do I value seats? Sure, the way I value brakes. I'd notice if they were bad.) He's showing me where the dog rides and how the dog gets in. Before long I'm imagining a car I could love with my dog, and a car my dog could love.
By the time I see the sticker, the question in my head is how much I'd pay for the kind of happiness the dog and I are going to find hanging out in this thing. Nobody's arguing with the number.
The second dealer opens with the retail price. Then he spends a while on how many hours of work went into the car, the tooling they built it with, the plant it came from, the date it shipped and the day it landed on his lot. He even starts talking about the seats, something about the alloy in the springs. All of it true, I'm sure. He could have walked me through the detailed cost breakdown of every seat, and it still would have meant nothing to me.
What do I do with that number? Haggle, I guess. Or more likely, I'll say "I'll think about it."
Too many of the agency proposals I've seen open (and end) the second way. Hardly any of them asked about the dog.
Why are agencies abandoning value-based pricing?
Because it asked clients to pay for value they couldn't see, and today AI makes that hard to hide. Promethean Research's State of Digital Services 2026 found the share of agencies using value-based pricing fell from 31% in 2024 to 18% in 2025, and the ones still using it grew more slowly than agencies on any other model. Promethean's own read is that AI exposed how much of the "value" was still execution, and clients caught on.
The value pricing fixes being touted are mostly relabeling. "Sell strategy" relabels the order. Value pricing puts a new price on the same order-taking, AI orchestration speeds up the hours, and productizing packages the orders. Relabeling doesn't change the fact that you're still talking about the thing that's losing value. A thing.
Tim Williams has the best-known version: split the work into Magic and Logic, do more Magic, and price the Magic on value. It's a useful split. But magic is value whose method is hidden on purpose, which makes it illegible by definition, and when buyers can't tell good from bad, the price sinks toward the bad (economists call it a market for lemons). AI now produces convincing magic for free, so put a price on yours and it reads as the counterfeit, or at best a $100/mo subscription to Claude.
Williams's own proof cases give it away. Butler/Till's media-buying agent and PMG's Alli platform are products priced on outcomes; that's Logic. Productizing doesn't escape the problem either, because only commodities can be sold as products. Huge tried it, in public. Your Next Org Model Will Sound Right for the Wrong Reasons
It doesn't matter what's inside the artifact that's being priced… if the client can't see it, then its price tag is trending towards zero.
When everything about the exchange says commodity vendor, the client reasonably concludes that what's for sale is assembly, and that whatever judgment came along with it is insignificant. You weren't hiding your judgment. The model you sold through was always about getting the client to pay the price.
Forgot Password?
Forgot Password?Is value-based pricing really an option for most of us?
Yes. I teach value-based pricing. But here's the real magic, what that first car dealer knew about me: it works, as the last step, not the first.
Years ago, at Reactive in Melbourne, they had a client who'd been very clear about her budget: $200,000. By the time the team finished the scope-value map, the cards added up to more than $300,000, and the departments lead were waging a silent war over whose scope would survive the cut.
Two hours before the walkthrough, Steph, the AD, was freaked out. Which hundred thousand do you take out, and who tells the client?
I told them to leave it all in. Let the client see what each piece was and they could decide what had the most value. They walked the client through all of it, card by card, what each thing was and what it would do, without a price tag on any of them.
Then Steph mentioned the bad news.
"But I want to do all of it!" the client said. "This is brilliant! Let me see how to get the rest of the budget."
She went from thinking in price (how much of my budget will they ask for?) to thinking in value (how great this would be to learn what the discovery will tell us).
Price has to be contextualized against value the client can understand and imagine the utility of. Ask yourself, what is the value of you using AI for the client? That's one example of where the conversation starts.
Next time a client asks whether this should cost less now that you use AI, ask yourself which car lot they're standing on.
Ask about their dog.
If this one hit close to home, let me know. [Set up a quick chat or shoot me a note.]
Or share it with another agency leader who is trying to sort out "What's next and what now?"
For what making judgment visible looks like in the room, read Why Your Agency Struggles to Get Paid for Judgment, and for the bigger picture, The 5% of Your Agency That AI Can't Replace.
To go deeper and ask questions, join me live on October 13
Forgot Password?Forgot Password?
