Corvane Analytics and the Office Lease Decision
Corvane Analytics, a 600-person data firm, has been remote-first since 2020. Its lease on 40,000 sq ft expires in nine months, and renewing would cost $2.4M per year. Internal metrics show output per engineer up 6% since 2020 and overall voluntary attrition down 3 points. But junior engineers with fewer than three years' tenure show a promotion rate that fell from 22% to 11%, and their voluntary attrition rose to 27%, against 14% for senior staff. Exit interviews mention mentorship twice as often as compensation. A messaging analysis found juniors sent 40% fewer direct messages to senior colleagues than they had in the office era. The CEO favours renewing a smaller 12,000 sq ft hub for optional use at $0.8M per year and mandating two in-office days per month for employees under three years' tenure. The CFO argues the hub is a half-measure that captures the costs of an office and the costs of remote work without the benefits of either.
1. Recommend a course of action and defend it against the CFO's claim that the hub is a half-measure.
Read the worked answer
Recommendation: take the hub, but do not treat the two-day mandate as the intervention. Fund structured mentorship - assigned senior partners, scheduled reviews, and protected time - and make the hub where that happens.
Why the junior data dominate. The 6% output gain is the weakest number in the setup. It could reflect process improvement, or it could reflect the departure of lower performers, which is exactly what a 27% junior attrition rate would produce. The junior signal is stronger for a different reason: it comes with a mechanism. The 40% drop in junior-to-senior messaging is a plausible causal path from remote work to fewer promotion opportunities, and it matches the exit interviews. Correlation plus mechanism plus a stated reason from the people leaving is a far better case than a productivity index.
Against the CFO. The half-measure charge assumes the only alternatives are a full office or none. That is a false dichotomy: the question is which office functions are worth buying. If the function the evidence implicates is junior-senior contact, the hub buys that function at a third of the full lease.
A rough cost check. If roughly 150 staff are junior, 27% attrition means about 40 exits a year. At even $30,000 per replacement, that exceeds the hub's $0.8M. The CFO's own logic favours the hub.
2. Is the two-in-office-days-per-month mandate defensible on the evidence, or should it be dropped?
Read the worked answer
It is close to indefensible as designed, and should be dropped - or rather, replaced.
The mandate has the shape of a policy that was reverse-engineered from a cost constraint rather than from the evidence. Nothing in the setup indicates that two days a month is the threshold at which mentorship recovers. The messaging data shows a 40% decline in junior-senior contact; there is no measured dose-response relationship showing what frequency restores it. A 24-day-per-year requirement is a plausible-sounding number that could just as easily do nothing.
There is also a fairness problem. The mandate singles out employees under three years' tenure, which converts a development problem into a compliance requirement imposed only on the newest and least powerful group. If in-person contact is valuable, senior staff must be in the room; mandating attendance for juniors while their mentors attend optionally will produce empty rooms and resentful juniors.
What I would substitute: a requirement that each junior engineer meets a named senior mentor at least twice monthly, by any means, with the hub available as the default venue, plus promotion-committee review of junior caseloads. This targets the mechanism rather than a proxy. The mandate's real content - that mentorship must be scheduled - is right; its instrument is wrong.
The CFO's critique is useful here even though I reject the conclusion: it correctly identifies that a hub plus a weak mandate purchases the costs of both models.
3. What evidence would most change your recommendation, and what result would flip it?
Read the worked answer
The evidence: a re-analysis of the junior promotion decline that controls for hiring composition and cohort size, paired with a pilot in which junior engineers are randomly assigned to a structured mentorship arm or a no-mentorship control.
Why the first half matters. The 22% to 11% collapse in promotion rates may not be a development failure at all. If Corvane tripled junior hiring after 2020, the denominator changed and promotion rates would fall mechanically while absolute promotions rose. If the re-analysis shows the decline disappears once cohort size is controlled, the entire motivation for the hub weakens and the CFO's position becomes much stronger.
Why the second half matters. Even with a real decline, my recommendation rests on a claim about mechanism - that contact causes promotion. A randomised trial is the only design that tests it. If mentored juniors were promoted at the same rate as controls, the hub is a $0.8M gesture and I would drop it.
What would flip me: a composition-controlled decline that persists, combined with a trial in which scheduled mentorship without any office component closes most of the gap. That would show the office itself is unnecessary, at which point I would fund mentorship and let the lease go.