Cloudflare grew revenue 34% in Q1 2026 and cut 20% of its workforce in the same breath. That combination is what made CEO Matthew Prince's Wall Street Journal op-ed land so hard — this wasn't a struggling company cutting to survive:

"We haven't found another example in U.S. business history of a public company growing at more than 30% that laid off more than 20% of its workforce."

— Matthew Prince, WSJ, 20 May 2026

The line that made the piece travel was Prince's framework for who got cut:

"AI isn't coming for builders or sellers, but it is coming for measurers. […] The vast majority of those we laid off last week were measurers."

It is a clean line. It is also half a story. Six weeks ago I argued in Manager Mode that AI doesn't kill management — it makes everyone a manager. I still think that's directionally right. But Prince's data pulls in the opposite direction, and the productive question is which of these two readings actually survives contact with the next twelve months.

The answer turns out to be: both, but only if you split the role.

What Prince actually argued

Prince frames Cloudflare's cut through three categories — builders, sellers, and measurers. He's careful with the attribution:

"Drucker explores the different roles inside every business, which I would categorize as builders, sellers and measurers."

That phrasing matters. The trichotomy isn't a Drucker framework you'll find in The Practice of Management (1954). The labels are Prince's, traced loosely back to Drucker's broader thinking about organisational function. Worth flagging because it's already being reproduced as "Drucker's framework" — it isn't. It's a Prince framework with a Drucker pedigree.

His description of the cut:

"We cut middle managers across the organization because AI allows us to have more direct reports per manager while still measuring and mentoring our teams effectively. We consolidated our operations functions into a single group that can support teams across the business, using AI to gain specific expertise when needed. We significantly reduced our marketing team, which, like in most companies, was teeming with measurers. Across our finance team, we found opportunities to consolidate and automate."

The cut hit middle management, ops, marketing, finance. The lines spared were product and field sales. Prince's logic: AI compresses the cost of observing the organisation — dashboards, reports, status decks, decks-about-decks — far faster than it compresses the cost of originating in it.

That's the part I think is right.

Why "Manager Mode" was half right

Manager Mode made the argument that AI inverts the engineer/manager ratio: when each individual contributor can ship five times as much, every IC starts behaving like a small manager — coordinating sub-agents, reviewing diffs, judging quality, setting direction. The role shifts from production to oversight, and the share of management-shaped work in every job goes up.

Prince's data says the measurement share gets eaten. How can both be true?

They can both be true because management is two jobs in a trench coat.

The first job is building the team and its work — hiring, mentoring, setting direction, unblocking, designing systems, deciding what gets built next. That's a builder activity, dressed in a management title.

The second job is measuring how the team and its work are doing — gathering status, validating progress, summarising up, producing reports, comparing performance. That's a measurer activity, also dressed in a management title.

For a long time both jobs needed a human because both were expensive. AI compresses the second one by an order of magnitude and barely touches the first. Hence Cloudflare's specific cut: not "managers", but the measurer slice of management — exactly the people whose week was 80% dashboards and 20% one-on-ones. The managers Prince spared are the ones whose week is the opposite.

This isn't unique to management. Every role splits the same way. Marketing has builders (brand, creative, partnerships) and measurers (attribution decks, weekly performance roll-ups). Finance has builders (fundraising, M&A, capital allocation) and measurers (reconciliations, monthly close, variance analysis). Even engineering has measurers (status updates, RFC reviewers who never write code, the architect role that has slowly drifted from designing systems to documenting them).

What Cloudflare cut isn't a layer. It's a slice of every layer — and the slice was disproportionately concentrated in the functions Prince named.

A one-question diagnostic

Open last week's calendar and your sent items. For each working block, ask one question:

Was the output of this block new information (a design, a draft, a decision, a customer conversation), or transformed information (a summary, a dashboard, a roll-up, a status update)?

Add up the hours. The transformed-information share is your measurer percentage.

I've run this on three teams in the last fortnight. The numbers were uncomfortable.

• A staff engineer at 38%. Most of it weekly status, design-review comments without owning a design, sprint-retro slides.
• A senior PM at 61%. Roadmap status, exec readouts, dashboard maintenance, "what's the latest on X" Slack threads.
• An engineering manager at 78%. The other 22% was the part of the job that survived a year of AI-assisted dashboards.

Cloudflare's cut is what happens when a company applies that diagnostic at scale and acts on it.

The diagnostic doesn't tell you to quit your job. It tells you which 40-60-80% of your job is now contestable, and how much runway you have to move it.

This isn't one CEO

Prince's argument lands harder because it isn't unusual. The Korn Ferry Workforce 2025 study found that 41% of employees say their organisation has already slashed management layers, and 37% report feeling directionless without their old managers. Gartner predicted in late 2024 that by 2026, 20% of organisations would use AI to eliminate more than half their middle management. Their May 2026 follow-up added a cautionary note that the cuts often don't deliver ROI — the same gap between the honest 1.2× productivity number and the 10× promise — but the cuts happened anyway.

Anthropic's Dario Amodei made the most-quoted version of the argument in May 2025, warning that AI could "wipe out half of all entry-level white-collar jobs" within one to five years. Amodei softened the language considerably by mid-2026, but the early-career version of the measurer trap — the analyst slot, the associate slot, the junior management consultant slot — is exactly the role he was naming.

The pattern across all of these: AI eats new information being processed far slower than it eats existing information being repackaged. Anything whose primary deliverable is a transformation of someone else's input is on the line.

Prince's line is going to age well. So is Manager Mode. The synthesis is what matters: management as a layer is fine, but the measurer slice of every job is now contestable, and the people who recognise it first get to choose where they end up in the new composition.

Cloudflare just paid 1,100 people a severance to teach this lesson. The cheap version is to learn it on the inside before someone runs the diagnostic for you.