Five years ago, Kristina Kaganer stood in front of the e-commerce industry and told it to cut the crap. CRAP, as in Can't Realize Any Profit.

The thesis was rude and correct: brands were chasing top-line revenue like it was oxygen and ignoring the only question that actually mattered, which was whether any of it made money.

This is the part of marketing nobody wants to say out loud at the conference, because the conference is paid for by the people growing unprofitably. Revenue is a vanity number you can put on a slide. Profit is the number that decides whether you still have a job in eighteen months. Kaganer figured out early that most of the industry had quietly agreed to look at the first one and call it strategy.

"How do you know if you're profitable if you don't know what you're spending," she said on the show, and five years later she still says it with the patience of someone who has watched a lot of people not listen.

She has earned the right. Fifteen years buying marketing technology at Coty, AB InBev, and Publicis. A psychology degree underneath all of it, which turns out to be the most important line on the resume, not the least. And then, somewhere around 2018, a hard left turn that most strategists never make: she learned to code.

"I learned to code in about 2018 because I was tired of being told things couldn't be done," she told me. "And I knew they could."

That is the whole woman in two sentences. There is a particular kind of professional who hears "that's not technically possible" one too many times and decides to go check the math themselves. Most people don't. Most people accept the engineer's no as a law of physics rather than a negotiating position. Kaganer treated it as the latter, and the gap between those two instincts is, more or less, the entire story.

Now she's CPTO at Navigator, building what she calls a commerce media network, and on the side she shipped a consumer product in two weeks because her son's toy had an exposed API. We'll get to the toy.

First, the heresy. Everyone in the industry is measuring AI by hours saved and efficiency gained. Kristina thinks they're measuring the wrong thing entirely. AI, she argues, is not about automation. It's about access to knowledge you couldn't reach before. That distinction sounds soft until you sit with it. Automation assumes you already know the right process and just want it done faster. But most marketing processes aren't right. They're just old. Automating a broken process doesn't fix it. It industrializes it.

So before you automate anything, she has a single question that ends most projects on the spot:

"If you were to do this a hundred times over, would you be happy with the outcome? Because if the answer is no, automation is already not the right solution to your problem."

It's a deceptively brutal filter. Run it against your own org chart and watch how many "AI initiatives" fail it immediately.

Then she said the line that the whole conversation orbits around, the one about why the best marketing AI won't be built by the engineers. I pushed her on it, because it's a direct shot at every engineering-led startup on earth. Her answer is below, along with the part where I told her the 2020 strategist and the 2026 developer seemed like two completely different people, and she took the wheel of the interview and drove off with it.

Behind the paywall is where this one actually gets good. Anybody can read a flattering founder profile. ADOTAT+ subscribers got the receipts check, the part where we pull every confident claim Kaganer made and hold it up to the light. The "exposed API" that wasn't exposed at all but a sanctioned developer program with a company literally paying for her drinks. The 60,000 women leaving UK tech, a real number attached to a reason the data says is mostly a myth, and what the research actually blames instead. The Trade Desk swipe she hedged twice and landed anyway, aimed by her own former employer in an adtech knife fight that stopped smoking days before she sat down. The "brilliant Google study" that is neither brilliantly Google's nor quite saying what she thinks it says. This is the stuff that turns a nice interview into a useful one, and it's all on the other side of the wall. Free readers got the story. Plus members got the truth underneath it.