Taylor Holiday did the thing nobody does: he showed his receipts. CTC's AI spend, broken down by department, year to date. $31K a month, $276 per person on average. And here's the part he didn't see coming: the biggest spender isn't engineering. It's client delivery, at 44.5% of total spend.
His growth team (the ones he calls Prophit Engineers) run client P&Ls, forecasts, creative strategy, media allocation and measurement. They burn through $13,759 a month, nearly half the whole AI bill. The pattern he keeps clocking: AI spend concentrates where revenue gets generated. Costs follow the money, which is exactly what you want.
The other numbers are smaller but the spread is the interesting bit. Prophit Engine at $320 per head, Technology at $222, Retention at $80, TikTok Shop at $94, Finance at $56. Heaviest per-capita users? Operations and leadership. Nobody mandated it. Adoption just spread on its own once a few teams proved it worked. Taylor's honest that he doesn't know if $31K is high or low for a company his size. He's still figuring out the right expectations. But he published it anyway, partly because @marty_kausas did the same and he found it useful. More of this, please. Nobody learns from a screenshot of a victory lap.
Right then, onto product. @andrewjfaris owned up to giving the same useless advice a million times: "make more good products." It's true. It's also worthless. What products? Vertical or horizontal? For what customers, what channels? What happens if the launch flops? Everyone knows product matters. Almost nobody has an actual framework for doing it.
His answer is @MehtabKarta, who he reckons has a clear framework for everything in ecom. Karta's big thesis lands hard: most "marketing problems" are actually product problems. You can be a 300 IQ marketer and still die on a bad product. On Faris's pod, Karta laid out 11 steps his portfolio companies follow. A few that stuck out: pushing spend past your product's local maximum is just wasted money, so know where that ceiling is before you scale. He runs a product-channel fit checklist before launching any SKU. He won't touch a DTC product that can't carry a $60 to $80 CAC, which quietly kills off a load of thin-margin nonsense before it starts. And he rates IP as the most defendable moat in ecom. The comic example: a "conservative mom swimsuit" line that fixed a brand's dead summer season. Right product, right gap, revenue out the other end.
The thread that ties Taylor and Mehtab together: stop treating spend as the lever. Taylor's data says money flows to where revenue is made. Mehtab says spend past the product's limit is dead money. Same lesson from two directions. The tool, AI or ad budget, only works if there's something underneath it worth amplifying.
@ChrisLangSocial is speaking next Wednesday on how he built Fresh Chile to a million email subscribers. The framing is the useful bit even without the talk: a big list isn't the goal, a list that generates revenue is. His seven-step system leans on selling access instead of newsletters, turning every post into a doorway, and building moments instead of campaigns. Whether the system holds up is another question, but the reframe is sound. A million dead emails is a vanity metric with extra storage costs.
Elsewhere it was quieter and less commercial. @codyplof was enjoying the World Cup footage of travelers trying American chains for the first time, which is the kind of wholesome filler the timeline needs more of. @youderian posted from @ecomfulco Live 2025, mountain biking with the line "bonds aren't forged in the conference room. Work hard, adventure often." Fair enough. And @moizali went fully off the ecom reservation with a couple of pointed political jabs about Schumer and Israel, which is his prerogative but isn't moving any product.
Bottom line is this: the operators worth reading yesterday were all circling the same truth from different angles. Your spend, your list, your AI bill, none of it matters if the thing underneath is weak. Build the product, then point the money at it. Not the other way round.