Matthew Mamet

Steve Kaufer ran TripAdvisor's product review personally. My first session: he asked which OTAs were the largest buyers on meta. I didn't know. His reply: "Aren't you supposed to know that?"

I spent the rest of the two-hour meeting writing a HiveQL query from the back row. When the room cleared I walked over and gave him the answer. All I got was a wry smile.

Full format + calibration rules at matthewmamet.com/product-review-meeting/

Private invite-only CRO luncheon in Boston last week. Consensus: organic and paid traffic is already dropping to agentic search. Three things to do - qualify harder on the traffic you have, test by channel before page, and keep human judgment in the testing loop. AI generates hypotheses fast but does not know why your last ten tests failed.

@matthewskelton@mastodon.social This is the kind of constraint that forces the architectural conversation that should have happened three years earlier. Capacity ceilings from a single hyperscaler have a way of making multi-cloud religion look a lot more practical.

A friend who is CPO at a PE-backed ad tech company told me the standard his firm set: come back when everyone on your team is a solid A. The B+ player is the dangerous one. Good enough that the case for action is never obvious. Not good enough that the team performs at the level you need. And everyone else already knows who the drag is.

When we expanded at EverQuote, we launched several verticals at the same time instead of sequencing. The supply-side cold start problem does not get cheaper when you run it in parallel. You have to build a real flywheel in one vertical before the next. We learned this the expensive way.

Etsy Q1: GMS per active buyer rose year-over-year for the first time since late 2022. That is not an operational metric, it is a strategic choice. They decided the acquisition problem was not the main problem and optimized the existing base instead. Worth watching.

When the steam engine got more efficient, people thought demand for coal would collapse. It tripled. More efficient engines meant more machines. More machines meant more fuel.

As AI makes building easier, demand for people who know what to build - for whom, and why - does not shrink. It compounds.

The constraint has shifted to judgment. That is a different job than what most PMs were trained for. It is not a smaller one.

@_elena@mastodon.social Just saw this announce from Ghost today and made me think of this thread. https://ghost.org/changelog/digital-public-good/

@designative@mastodon.social Intent documentation is the right frame. The analogy I keep coming back to is the difference between a spec that documents what to build and one that documents why. Agentic systems need the why layer baked in or they optimize toward the outcome while drifting from the constraint set the team actually cared about.

I was on a call last week with someone running partnerships for a free leadership curriculum for girls 11-15. Famous founder. 300+ partner schools since 2023. No price barrier.

Still fighting a last-mile distribution problem most paid products would envy having.

National sign-off creates permission, not momentum. Each state, each regional council, each after-school program makes its own decision.

Brand helps. Free helps. Neither solves it. Distribution is its own problem.

In 2009 I was charging $800 a day, four days a month per client, hard cap at four clients. Nobody called it fractional.

The structure was already there: day rate, monthly retainer of days, hard client limit, ongoing strategic involvement without a full-time seat. The category came 15 years later.

The four-client cap was intuitive, not calculated. Concentrated senior judgment has a carrying capacity. Above a certain number, the quality drops before clients notice.

The real risk is not the content going dark, it is the growing gap between what is preserved and what is discoverable. Indexing has always lagged archiving, and AI summarization is now consuming the archive without necessarily surfacing the original source.

@_elena@mastodon.social Ghost is the one I'd push hardest on that list. I moved my practice blog there about a year ago - setup is straightforward, you own your subscriber list outright, and the SEO tooling is solid enough that I stopped needing a separate layer on top. The one honest caveat: the theme customization requires comfort with code. For most writers though, the defaults are good enough to ship immediately.

Six weeks into an engagement, I asked the product team why each roadmap item existed. Nobody could explain more than four of twelve without saying "it came from sales" or "leadership wanted it." That is not a roadmap. That is a parking lot. I have seen this at almost every company I have walked into - fractional, FTE, advisory. The first real work is not building a roadmap. It is excavating the one that already exists and asking what happens if you stop doing half of it.

Saying "No" to a bad idea is not hard. What is hard is saying "Yes" to something and defending it for six months while priorities shift around you. The PM who becomes Dr. "No" out of self-preservation is not exercising discipline. When the business stops engaging with product and starts going directly to engineering, you have not protected the roadmap. You have made yourself irrelevant. The question is what you said "No" to when it cost something.

The most common growth mistake in scaling B2C companies is not a strategy failure. It is a measurement failure. A channel works, the team doubles spend, CAC holds for two quarters, then collapses. They saturated the addressable audience and had no measurement to see it coming. The signal is not CPL. It is the ratio of new-to-repeat visitors in the converting cohort. When that ratio flips, you are six weeks from CAC blowout.

@hbons The can analogy is better than the fruit one because the decay is invisible until you open it. Most teams don’t find out their software has expired until a dependency breaks or a new hire tries to set up the dev environment and can’t.

The expand vs. cut framing maps perfectly to what I see advising growth stage companies. The ones treating AI as a cost reduction play are the same ones who treated every previous efficiency gain as a cost reduction play. They did it with offshoring, with automation, with cloud migration. It is a pattern of strategic laziness, not a response to the technology itself. The companies using AI to ship into adjacent markets faster are pulling away.

@david_chisnall The expand into new markets logic is exactly right, and I've lived it. At every consumer marketplace I scaled, a productivity gain meant we could finally go after the adjacent segment we'd been ignoring. The companies using AI as layoff cover are telling you they ran out of ideas before they ran out of people.