Here's a number that should stop any marketer cold: ten billion dollars.
That's what Meta told investors it expected to lose in 2022 revenue because Apple decided people should get asked before an app tracks them across the internet. One privacy prompt. Ten billion dollars. The stock dropped 26% in a single trading session the day they said it out loud.
I've spent years building businesses before I packed up and started traveling full-time, and I've learned this: when an entire industry's foundation gets rebuilt under its feet, it's worth paying attention to who adapts and who just keeps doing what used to work.
Performance marketing spent twenty years running on a simple idea. Every dollar could be traced to a click, every click to a sale. It worked because of two things: cookies that followed you around the web, and device IDs that followed you inside apps. Then Apple introduced App Tracking Transparency in 2021, and Google spent five years threatening to kill cookies in Chrome before finally backing off entirely this past April. Five years of "we're killing cookies," and then... they didn't. Chrome still runs cookies by default. Safari, Firefox, and Brave block them. Meanwhile Apple's changes never wavered, and they've been reshaping ad budgets ever since.
So what do you do when your measurement system gets pulled out from under you? Procter & Gamble gives us a real answer, and it's a good one.
By early 2023, P&G posted its first quarterly sales decline in five years. CEO Jon Moeller didn't respond by slashing the marketing budget. He told investors the company would chase "productivity" — getting more out of marketing without necessarily spending more on it. Then they showed their work. In the US baby care business, P&G cut media spend in half. Half. And reach went up 20%. Top-of-mind brand awareness climbed 26%. Sales in that business grew 10%.
Think about that for a second. They spent less and got more attention. How does that happen? Because the old system — chasing the last click before a sale — was rewarding ads that were easy to measure, not ads that actually built a brand. P&G had been pouring money into narrow, high-frequency digital placements because those were trackable, while the broader awareness-building work that actually moves people toward a purchase got undervalued. Cut the noise, and apparently the signal gets stronger.
That's a fascinating discovery, and it's forcing a methodology back from the dead. Marketing mix modeling — a statistical approach invented in 1949 — fell out of favor when digital advertising promised to track everything down to the individual click. Now it's back, and it's better, thanks to cloud computing and modern statistics. Google released its own version, called Meridian, in January 2025. Meta open-sourced theirs, Robyn, back in 2023. Both companies, worth noting, have an interest in a measurement method that happens to make channels like YouTube look better than click-tracking ever did. Nothing wrong with that — just worth knowing who built the yardstick.
Meanwhile, money is pouring into retail media — ads placed right where people shop, built on first-party purchase data instead of the browser tracking that's falling apart. US retailers are projected to pull in $69.33 billion from this in 2026, up nearly 18% from 2025. Globally, WPP Media puts the number above $150 billion. But here's the catch: more than 89% of the new growth is going to just two companies, Amazon and Walmart. A McKinsey survey of about 150 US media buyers found most of that retail media spend isn't new money at all — it's shifted out of other digital budgets.
So what's the actual lesson here? Measurement isn't truth. It's a tool, and tools go out of date. The companies doing well right now aren't the ones clinging to the old click-counting religion. They're the ones asking what happens if we measure differently, testing it, and adjusting when the answer surprises them.
That's not just a marketing lesson. It's a life one. The map that worked for twenty years isn't guaranteed to work for the next twenty. Question it. Test something new. See what happens.
Source: MarkHub24
See also our earlier piece: Your AI Isn't the Problem. Your Data Is.
