The Return-to-Office Memo, Annotated

The memo landed at 4:51 p.m. on a Thursday, which is the single most reliable tell in corporate communications. Nothing good is ever sent at 4:51 on a Thursday. Nine o'clock Monday is for good news, because you want people talking about it all week.

Six paragraphs. The word collaboration three times, culture twice, energy once — and, in the fourth paragraph, sitting there like a shoe in a soup bowl, one sentence about badge data.

I have read a great many of these now and they are, without much variation, the same document. So here it is, annotated.

The memo¶

"After careful consideration, we're asking everyone to be in the office three days a week beginning next month."

"The energy of being together is something we've all missed. Our best work happens when we collaborate in person."

"We've heard a lot of feedback from you on this, and we've taken it seriously."

"We'll be using badge data to understand how the transition is going."

"We're confident this is the right decision for Halden and we're excited to see everyone back."

Six paragraphs, one mechanism, no numbers. And that's not sloppiness — a memo with numbers in it invites a conversation about numbers.

What the research actually found¶

Here's where the fake company stops and the evidence starts.

Yuye Ding and Mark Ma, at the University of Pittsburgh's Katz Graduate School of Business, examined S&P 500 firms that publicly announced return-to-office mandates, using Glassdoor data to track what happened to employees afterwards.

Two findings, and the pairing is the point. RTO mandates did not improve firm performance or shareholder value. And employee satisfaction fell — overall job satisfaction, work-life balance, and specifically employees' assessments of senior management.

Their stated conclusion:

That is an unusually direct sentence for a finance paper, and it is worth reading twice. The researchers are not saying mandates are merely ineffective. They are saying the pattern in the data fits control and scapegoating better than it fits any productivity story.

Then there's the cost nobody budgets for. David Van Dijcke, Florian Gunsilius and Austin Wright analysed 260 million résumés matched to company data, examining Microsoft, SpaceX and Apple:

Read the mechanism carefully. It isn't that people are unhappy. It's that the effect is strongest at the top of the tenure distribution — the longer someone has been there, the more likely they were to leave. And they left to direct competitors, which means the human capital didn't evaporate; it relocated to the firm across the road.

The memo says the mandate is about doing our best work. The best-documented consequence is the departure of the people who did it longest.

The turn: what isn't in the memo¶

Now the part where I have to be careful, because there's an argument here I believe and it is not what the research found, and blurring those would make me exactly the thing this desk exists to annotate.

What the researchers found: no performance improvement, falling satisfaction, senior attrition, and a data pattern consistent with control and blame-shifting.

What I think, which is inference and not evidence: the memo omits the largest relevant number on the balance sheet, and that omission is not random.

An office is a fixed cost, usually on a long lease, signed years ago by someone forecasting a workforce that would be in it. It is the purest example of a decision made once that keeps executing — the rent leaves monthly whether the desks are full or empty, and it is entirely unaffected by anything anyone decides today.

Which makes it a sunk cost, and therefore genuinely irrelevant to whether people should come in. That's the arithmetic. But a sunk cost of that size exerts enormous pressure on the people who signed it — quitting an expensive commitment requires confessing something, and "we have eleven floors and nobody in them" is a confession with a name attached to it.

I want to be scrupulous: I have no evidence that any specific company reasons this way, and neither Ding and Ma nor Van Dijcke and colleagues tested it. It's a structural argument about where the pressure sits, not a finding. Treat it as a hypothesis you can check against any particular employer's lease disclosures, not as something proven.

And there's a second thing memos don't say, which is more uncomfortable and better supported by the attrition data. A mandate that causes people to leave voluntarily reduces headcount without severance, without notice obligations, and without a layoff announcement. The Van Dijcke findings don't establish intent — but they do establish that mandates produce exits, and any executive reading that paper now knows it. Once a side effect is documented, continuing to describe it as a side effect is a choice.

That's the same grammar as a memo that reduces a workforce while describing a strategy, and the same grammar as nine minutes of true sentences that disclose nothing. It's all one dialect — the same one this site keeps translating — and once you can hear it the memos stop being surprising.

Who actually pays¶

Not the executives, who were coming in anyway and have the parking space.

The costs land on: the person who moved out of the city in 2021 on the strength of a policy; the parent whose childcare arrangement assumed two days at home; the disabled employee for whom remote work was the accommodation that made the job possible; the person on the lowest salary in the building, for whom a five-day commute is a pay cut measured in hundreds a month and hours a week.

None of those people appear in the memo, and it's worth noticing that the document has no vocabulary for them at all. It has words for energy and culture and collaboration. It has no word for this costs some of you materially more than others, because saying that would require quantifying it, and the whole architecture of the document is built to avoid quantifying anything.

Reading the next one¶

1. 1

   Count the mechanisms

   Go through and mark every sentence describing something that will actually happen — a system, a measurement, a consequence. Most RTO memos have exactly one, and it's badge data. Whatever that one mechanism measures is the real objective, regardless of what the other five paragraphs are about.

2. 2

   Look for the number that isn't there

   If in-person work improved output at this company, the memo would say by how much. Absence of evidence in a document that could easily have contained it is itself informative — especially when the same organisation quantifies everything else it cares about.

3. 3

   Check the lease against the mandate

   Public companies disclose property commitments. If a mandate arrives shortly before a break clause, a renewal, or a sale-and-leaseback, that's worth knowing — not as proof of anything, but as context the memo declined to provide.

4. 4

   Watch who leaves over the following two quarters

   The research predicts the senior end goes first, to competitors. If that's what you observe, you're watching a documented effect rather than a coincidence, and it tells you more about the policy's real cost than any engagement survey will.

5. 5

   If it costs you materially, say so in writing and in numbers

   "This adds £340 a month and seven hours a week" is a fact that has to be responded to. "I'd prefer to stay remote" is a preference that can be noted and filed. Same position, one of them enters the record as a cost.

4:51 on a Thursday¶

I went back and read the Halden memo one more time before filing this, which is a strange thing to do about a document you wrote yourself.

What struck me is that it is, sentence by sentence, entirely defensible. Nobody claimed a number that could be checked. Nobody promised anything that could be broken. The one enforceable thing in it — badge data — was mentioned once, in the middle, in a paragraph about transition support. It is a professionally competent piece of writing and I'd have signed off on it as legal.

And the research says the policy it announces will not improve the company's performance, will lower how people feel about working there, and will disproportionately remove the staff who have been there longest, to competitors.

Both of those things are true at once, and the gap between them isn't a lie. It's a document that was never required to be about the same subject as its consequences.

Send it at nine on a Monday, and someone might ask.

It's just business.