All podcasts / Lenny's Podcast / Summary

This CPO regrets that product management exists | Tom Verrilli (CPO of Whatnot)

2026-08-02 - 84 min - source - Read full transcript
Lenny Rachitsky (host)Tom Verrilli

Key insights

Whatnot's PM philosophy is to not assume a team needs a PM by default; hire one only where there is a specific, provable need.
Tom frames product management as a trade learned through reps, not a qualification. The more engineers and designers are abstracted away from making product decisions, the more their own decision-making muscle atrophies, so Whatnot deliberately avoids the standard 'one PM per six engineers' HR ratio.
pm-scope-and-hiring
PMs at Whatnot are mapped to problems and six-month planning priorities, not permanently assigned to a team or an engineering manager.
Every six months leadership defines what has to be true for the company and assigns a DRI to each priority; that DRI is often but not always a PM. This keeps PMs building reps across varied problems and leaves teams without a permanent PM to build that muscle themselves when nothing critical is in flight.
pm-scope-and-hiring
The clearest anti-pattern in PM hiring is candidates who lead with alignment-driving and stakeholder management rather than technical or customer substance.
Tom calls this 'politics' as a specialty and says it is trending down hard in what Whatnot looks for. What is trending up is demonstrated systems thinking: candidates who can describe both the macro end-state and the specific, fast way they would validate it, and who can point to things they built and decided on rather than existing systems they merely maintained.
pm-scope-and-hiring
AI has sharply increased IC leverage, letting senior product people replace work that used to require dedicated data science and engineering support.
Tom describes pulling nuanced cohort data himself via internal Hex threads, in minutes, that used to take a data scientist one to two weeks. He also queries the codebase directly through tools like Claude instead of interrupting engineers to explain how something works, which he says has cut his data-science conversations dramatically even as he spends far more time in the underlying data than before.
ai-leverage-in-product-work
Whatnot pushes senior product leaders, including VPs, directors, and Tom himself, to spend the large majority of their time doing individual-contributor work rather than pure management.
Tom estimates he personally spends about 50% of his time on IC work; managers on his team spend 90%+. The rationale: staying hands-on keeps decision-makers connected to ground truth, avoids the 'yo-yo' cycle of review-reject-rework, and lets one senior person make the tradeoff a split team would otherwise fight over politically (he cites merging Twitch's warring discovery and ads teams under one PM as the fix for exactly this).
ic-over-management
'Hire great people and get out of their way' is wrong in Tom's experience; the better model is verify-then-trust.
He describes Whatnot founder Grant clearing his calendar mid-review to sit with a team and go line-by-line through tickets, code, and data until he genuinely understands a system, rather than delegating and assuming competence. Tom argues you cannot make good macro decisions without periodically going deep on the micro details yourself.
founder-led-culture
Top-down, founder-led decision-making works well specifically when leadership stays close enough to ground truth to be right; it becomes micromanagement only when leaders manage from above without that grounding.
Tom says most junior engineers and designers are energized, not resentful, about working directly alongside a CPO or CEO who is genuinely informed, because it removes layers of alignment meetings. The failure mode is a leader issuing top-down direction without having done the work to actually know what's true.
founder-led-culture
'Play the accordion' - the practice of alternating between zooming out to reason through implications and zooming back in to ship the smallest testable thing - is Tom's core operating discipline.
He illustrates it with Whatnot's listings decision: skipping traditional e-commerce listings works for live-selling AirPods, but zooming out reveals it breaks search and discoverability for new buyers; zooming out again on the fix (forcing all sellers to write listings) reveals it would crater seller throughput. The team has to keep alternating between these zoom levels rather than settling on either a rigid roadmap or pure iteration.
systems-thinking
The core PM skills Tom sees as durable in an AI-heavy era are understanding the customer, the business, and the tech well enough to translate between all three, not storytelling, alignment, or roadmap theater.
He agrees with Lenny's 'PMs are best positioned to thrive in AI' thesis but adds that most PMs have not been rewarded for these skills over the last five years, and that the AI-era hiring bar (a hands-on case study plus live verbal defense at Whatnot) exposes candidates whose thinking 'decays' once they can't rely on polished framework delivery. He credits Marty Cagan's term 'product theater' for this failure pattern.
pm-scope-and-hiring
A recurring career mistake Tom names in himself is trusting population averages without checking whether a small user segment depends entirely on the thing being cut.
He gives the example of deprecating a feature used by only 3% of users company-wide, without realizing it is 100% of the workflow for a distinct group, comparing the downstream damage to 'turning off the power at a mall in the lead-up to Christmas' for the businesses that depend on it.
systems-thinking
Whatnot's live-commerce model deliberately trades off search and structured listings for speed, betting that most retail demand is actually low-intent and browsing-driven rather than high-intent search.
Tom cites that e-commerce has never exceeded roughly 20% of US retail spend after 30 years, arguing most shopping resembles mall browsing more than targeted search. He sees agentic commerce as complementary for high-intent purchases, not a competitor to the social, live-browsing experience Whatnot is built around.
systems-thinking

Books referenced

Media referenced

Companies

Techniques and frameworks

Summary

Tom Verrilli, Chief Product Officer at Whatnot and formerly longtime CPO at Twitch and a growth PM at Twitter, joins Lenny to unpack a line he published that surprised a lot of people: "we regret that product management exists." Tom is careful to clarify this isn't an anti-PM stance but a forcing function - Whatnot refuses to default to the industry-standard ratio of one PM per roughly six engineers, and instead only staffs a PM where there is a specific, demonstrable need. The reasoning is that PM work is a trade built through reps, and every time an engineer or designer is abstracted away from making a product decision because "that's the PM's job," their own decision-making muscle goes undeveloped. In the last two years, Whatnot received over 31,000 PM applications and hired one; the company currently runs about 20-22 PMs against a business processing enormous transaction volume.

Much of the conversation traces how this plays out operationally. Rather than permanently attaching PMs to teams, Whatnot runs a biannual planning cycle where leadership defines what must be true over the next six months and assigns a DRI to each priority - sometimes a PM, sometimes an engineer or designer. In hiring, Tom says the clearest red flag now is candidates who lean on "alignment" and "stakeholder management" language, which he calls politics dressed up as a PM specialty; what he's looking for instead is demonstrated systems thinking, spanning both macro end-state reasoning and the concrete, fast steps someone would take to validate it. He credits AI directly for raising the bar here: PMs can now pull nuanced cohort data themselves through internal tools in minutes rather than waiting a week or two on a data scientist, and can query a codebase through an assistant like Claude instead of interrupting an engineer, both of which mean fewer, more senior generalist PMs can cover more ground than a larger junior bench could.

A second major thread is Tom's insistence that senior product leaders, including himself, stay heavily in individual-contributor work rather than shifting fully into management once promoted. He estimates spending roughly half his own time as an IC and says Whatnot's managers spend 90%+ of theirs there, arguing this keeps decision-makers connected to ground truth and avoids the "yo-yo" cycle where a PRD bounces between a junior PM and a reviewer who already knows the answer. He explicitly pushes back on "hire great people and get out of their way," describing instead a verify-then-trust culture where Whatnot's founder Grant will clear his entire day mid-review to sit with a team and go line-by-line through tickets, code, and data until he genuinely understands the system before trusting a team's judgment on it.

Tom's most durable mental model is what he calls "playing the accordion": stretching out to reason through the broader implications of a decision, then compressing back in to ship the smallest testable version, and repeating that motion continuously rather than settling into either a rigid multi-year roadmap or unstructured iteration. He walks through this with Whatnot's own listings tradeoff - skipping traditional e-commerce listings works great for a seller holding up AirPods on a live stream, until you zoom out and realize new buyers expect search to work, and then zoom out again to notice that forcing every seller to write a listing would crater how much they can sell per hour. The conversation closes on Tom's Twitter years (nine heads of product in two years, and the lesson that real product-market fit survives almost any organizational dysfunction), a Fail Corner story about trusting population averages that hide a dependent minority use case, and a lightning round covering his book and media picks, his life motto ("I'll figure it out"), and a live California spiny lobster he bought, shipped overnight, off a Whatnot seller's dockside livestream.

Notable Quotes

"In the last two years, 31,832 people applied to be a product manager at Whatnot. We hired one." - Tom Verrilli

"If you truly find product market fit, like if you manage to bottle lightning, it doesn't matter how badly you screw up the organization of it. It's huge." - Tom Verrilli

"Averages mean nothing to the individual is probably the thing that I've really scarred by... it turns out you like blow up the use case of that group of humans." - Tom Verrilli

"I don't think there is one way to do product management, and I don't think there is one way that AI will shape the industry." - Tom Verrilli