Engineering culture audit

Where we looked, and how we protected the people we spoke with.

The subject is a post-IPO scale-up with ~250 engineers across eighteen teams, six business units, and two engineering hubs. We were given access to hiring data, aggregated performance review summaries, three years of engagement survey results, and four years of promotion and attrition records.

We did not audit specific managers or specific teams in this report. The patterns surfaced are organization-wide; calling out individuals would obscure the structural picture. All interviews were conducted under strict confidentiality. No comment in this report can be traced to any individual.

How we gathered evidence.

• Reviewed three years of engagement survey data, including free-text comments.
• Analyzed promotion and attrition rates by team, level, and tenure.
• Conducted thirty-one confidential interviews across all levels — engineers, EMs, directors, two VPs.
• Spoke with eight former engineers who left in the past twelve months.
• Reviewed published engineering values, manifestos, and onboarding documents.
• Sampled twenty anonymized performance review documents to characterize feedback norms.

What we observed.

A. The hiring bar has held, but the calibration has drifted.

Pass rate on the senior-engineer loop dropped from 14% to 9% over two years — appearing to indicate a strengthened bar. However, interviewer-level analysis shows the loop is harder primarily because two specific interview stages have become idiosyncratic. Engineers describe the loop as "passing depends on who you get" — a calibration problem, not a bar problem. The risk: rejecting strong candidates while still admitting weak ones.

B. Review practices have softened.

Performance reviews in the past two cycles show a marked compression of ratings: 71% of engineers received the same rating ("meets expectations"). This appears to be a well-intentioned response to a previous over-stratified cycle. The effect: high performers feel invisible; struggling engineers receive no signal to improve. Engineers describe this as "the year nothing was said."

C. Decision-making has slowed and become hierarchical.

In a sample of twelve significant technical decisions over the past nine months, nine required VP-or-above sign-off. Eighteen months earlier, the same kinds of decisions required director sign-off at most. This is not the result of any deliberate change; it is what happens when no one is sure who can decide. Engineers describe a "wait for the next staff meeting" pattern — clarity is delayed, and then everyone executes on a deadline.

D. Ownership has eroded at the team boundary.

Cross-team work has become harder over the same window. Teams describe a "not my problem" attitude as more common than it was. This correlates with team-level OKRs that reward focus on internal work over collaboration. It does not correlate with any change in interpersonal dynamics; people still like each other.

E. Senior engineers feel unseen.

The leavers we spoke with shared a common phrase: "I wasn't sure what I was working toward." The career framework is documented but not used in practice. Promotion criteria for staff and principal levels are vague, especially for engineers who do not lead teams. Combined with the review-softening above, this produces a "what does anyone actually want from me" feeling.

The culture has not eroded. It has gone quiet.

The patterns we observed are second-order effects of healthy reactions to specific past problems — over-stratified ratings, individual decision-making bottlenecks, hiring with too much variance. The corrections were correct. The corrections were also un-recalibrated.

The result is a culture that has lost its signal-to-noise ratio: harder to know what good looks like, harder to know who can decide, harder to know what you are working toward. This is fixable, but only with deliberate action. Cultures do not re-calibrate on their own.

What we would change, in priority order.

1. Recalibrate the interview loop.

Re-run interviewer calibration with a structured rubric. Identify the two stages that have drifted and rebuild them. Expect a one-quarter project, owned by a single director.

2. Restore signal in performance reviews.

Move from compressed ratings to explicit, written feedback that names specific strengths and gaps. Train managers in how to give it. Accept that some engineers will be unhappy in the short term — that is the point.

3. Publish a decision-rights document.

Name, for each category of technical decision, who can decide. Make it visible. Update it quarterly. The document itself matters less than the act of agreeing on it.

4. Tie OKRs to cross-team value.

Re-introduce at least one cross-functional commitment per team per quarter. Reward delivery on it explicitly.

5. Make career progression legible.

Refresh the career framework with specific examples per level. Publish promotion criteria for staff and principal in terms of concrete contributions, not adjectives.

A sequenced plan.

Weeks 1–2.

Form a small calibration working group — three senior ICs, two EMs, one director. Charter is to recalibrate the senior-engineer loop.

Weeks 3–4.

Draft the decision-rights document with the executive team. Publish v1, mark it as living, schedule a quarterly review.

Weeks 5–8.

Manager training on written feedback. Run a parallel performance-review exercise with a small sample to test the new feedback model before the next cycle.

Weeks 9–12.

Refresh the career framework. Publish v2 of decision rights with what was learned. Begin cross-team OKR work for the next quarter.

This is twelve months of work compressed into a ninety-day kickoff. The first quarter establishes momentum. The remaining three quarters do the patient work of making the changes stick. We are available for the kickoff and for the first manager-training cycle.