Open the last exec update you sent and count the sentences an exec actually needed. My guess is three, buried in the middle, surrounded by activity: shipped this, met with them, on track for that. The update was written to demonstrate effort, and execs can smell that in one scroll, which is why they skim it, which is why the one thing that mattered got skimmed too. I have written these for twenty years and read thousands more. The bad ones all fail the same way: they are activity logs. The fix is a fixed template with a brutal budget.
The four sections above the fold
The scoreboard. Five numbers at the top, each with a trend arrow and one line of context: a direction metric (the number your strategy says should move), the eval trend on your most important AI-touched surface, margin or cost per outcome, adoption of your current bet, and the top customer-signal theme. The rule that makes it work is that the five numbers never change week to week. The moment you swap metrics you have started metric shopping, and the reader knows it. A flat or falling number with an honest one-line explanation builds more credibility than a rotating cast of green ones.
Three calls I made. The heart of the update. Three decisions, each in three lines: the decision, the evidence behind it, and the reversibility (two-way door, walked through it; one-way door, here is why). This converts the update from an activity report into a judgment record, and it feeds your decision log for free. An exec reads three lines and learns more about how you operate than from a page of status.
One thing I need. One ask, and it must be a decision with a deadline and a default. "I need a yes or no by Friday on X; if I hear nothing, I will do Y." If you have no real ask, say "no ask this week."
What I killed or declined. One line: the request, the no, the reason. This is the least practiced and highest-trust section in the format. Execs see endless additions; almost nobody shows them the subtractions.
Below the fold and the agent draft
Everything else goes appendix-style with rules: no item over three lines, no item that requires action, and a standing link to the dashboard for anyone who wants raw numbers. The fold is a promise to the reader; keep it and they will read every update you send, break it twice and they stop. An agent can draft the mechanical 80 percent from your tools, but the 10-minute human pass is not optional.
How to use it this week
Take the update you are due to send and rewrite it into the format: five numbers picked once and never changed, three calls with evidence and reversibility, one real ask with a deadline and a default, one kill line, everything else below the fold. Send it, then watch the response pattern change: replies arrive sooner, they engage with the ask instead of the activity, and someone will ask about the kill line, which is often the first real conversation about tradeoffs an exec reader has had with a product team in months. Wire up the agent draft the following week. Format first, automation second, because automating a bad update just ships the badness faster.