Leadership1 publisher2 min readPublished
A $29.5 billion waste figure anchors the case that product problems start above the product manager
Rob Versaw of Dynatrace argues that capable product managers look weak inside systems they did not choose. The Pendo estimate behind the argument is an industry-scale figure. It sizes the category; it cannot audit one company.
The Board Room · Leadership desk
What happened
- Pendo estimates that 80 percent of features in the average cloud product are rarely or never used, and priced the build of those features across publicly listed software companies at about $29.5 billion.
- Rob Versaw, who works in product strategy at Dynatrace, wrote that companies complaining about product quality arrive with the remedy already chosen, usually a brief to hire much better product managers.
- He names five repeating patterns instead, including restricted customer contact, decisions taken above the team, and a ranked priority list that resets every fortnight.
- The argument leans on Deming's estimate that 94 percent of trouble belongs to the system and is management's responsibility, leaving 6 percent to special causes.
- It sets Marty Cagan's distinction against current hiring practice: an empowered team is given a problem to solve, a feature team a list to build, and firms recruit for the first while running the second.
Compiled by The Board RoomSomething wrong?How this is made
Why it matters
- decision An executive who accepts this reading spends the quarter on one of two things: a search, or a rewrite of who approves roadmaps and who sees the P&L. Only the second changes what the next product manager inherits.
- exposure Raising the hiring bar without changing the environment exposes the employer to paying twice, once for the search now and again for the replacement about 18 months later.
- constraint The $29.5 billion can justify opening a review of feature waste. It cannot measure any single company's share of it, so the internal case still has to be built from that company's own usage data.
- capability Because four of the five patterns are questions of access and approval, they can be tested against records a company already keeps. Checking those records costs less than commissioning a talent assessment.
Pendo's figure is the only priced quantity in the argument [1]. If cost per feature were uniform across used and unused work, $29.5 billion of unused build implies roughly $36.9 billion of total feature spending across the companies counted [14]. The column does not say how many companies that is, or over what period the spending accrued [13].
"It is not a talent failure. It is the accumulated output of an operating model that keeps sending capable people to build the wrong things," Versaw wrote [4]. He works in product strategy at a software vendor and published on a contributor platform [2], so a skeptic can fairly say the piece moves the fault upstairs from the function he represents. The supporting authority is thin in the same direction, and the author concedes it: on Deming's 94-6 split, he wrote, "You don't need the exact split to accept the principle" [8].
What survives that objection is narrow and factual. Whether product managers can reach customers, and whether they can see revenue, margin, cost-to-serve and retention, is a matter of record inside any company that cares to look [5][6]. Marty Cagan's distinction gives the hiring version of the same check: firms interview for commercial judgment and conviction, then hire someone whose actual job is backlog administration [9].
The two remedies differ in what they cost when they are wrong. Hiring against an unchanged operating model buys about 18 months, on this account: "Put a strong product manager in that environment and they will look weak. Give it 18 months and they will be somewhere else, because the strong ones always have options" [10]. Moving roadmap and pricing decisions back to the team costs executive attention in the current quarter, and it makes the next hire's performance readable.
The founder-mode objection points at the same architecture. Since Paul Graham's essay, hands-on executive direction has had a respectable name. Versaw's argument is that it is a style of involvement. It is not a decision-rights model. So a company can call itself product-led while running founder mode and leave the contradiction unresolved [11].
The AI budget is where the fifth pattern bites this year: AI-native delivery is mandated while tooling and token budgets are capped [6]. The column cites the 2025 DORA research, drawn from nearly 5,000 technology professionals, for the finding that AI amplifies what is already in a team without fixing it [12].
What to watch
- Publication of the company count and period behind Pendo's $29.5 billion would make the figure comparable with a single company's R&D line.
- The next DORA report, and whether the amplification finding holds on a larger sample of teams using AI in delivery.
- A company reporting product outcomes before and after it moved roadmap approval back to its product teams.