Everybody Nods

I’ve sat in a lot of kickoffs where everybody agreed, but not everyone believed. The deck goes up, the date is on the slide, someone says this is going to be great, and around the table five people are each thinking what about the API limits, this doesn’t account for the data model changes, will other customers actually want this, that date is too tight. None of it gets said. Four months later the thing slips, and the slip gets treated as a surprise even though most of the room could have called it on day one.

The encouraging part of that story is that nobody had to go find the information. It was already in the room, on day one, for free. Five people knew most of what the project would eventually teach everyone the expensive way. That’s a distribution problem, not a talent problem, and distribution problems have fixes.

Asking the right questions is good. Asking them at the right time is great. In the age of AI-assisted building we keep hearing that the building isn’t the hard part anymore. What to build, and whether to build it at all, is. That should make the conversation before the work more valuable, not less. But most of the gates we have are built for momentum: get to the thing, ship it, iterate. That’s a good strategy for two-way doors. It’s an expensive one for the decisions you only get to make once.

And the format itself works against the conversation. The person who pokes hardest at the PRD ends up with a reputation for being difficult, so they poke less next time. Half the room is reading the doc during the meeting because they’re heads-down on something else. The review is a process formality rather than a real weigh-in. And people who didn’t get heard the last three times stop raising things. That’s what happens when the only channel for disagreement is speaking up, out loud, in front of everyone, with a deadline already on the slide.

Those voices are the whole point. Early friction between teams makes better products. A shared understanding of scope produces sharper questions, finds holes, and surfaces opportunities nobody had budgeted for. I used to go get it by hand, pinging people individually on Slack or spinning up a survey, and it worked, but the answers ended up scattered across five places and I wanted them in one.

So I built Daylight.

One card, one project. Somebody writes a sentence about what the project has to do and by when, sets how many people will answer, and sends a link. Each person answers three questions on their own: how good is this today, how good does it need to be by the delivery date, and will it land, as a percentage. They can attach up to five doubts, chosen from a short list or written in their own words. Once everyone has answered, the results open for the whole group at once. It’s free, it’s anonymous, and it takes about a minute.

Send one before your next kickoff and tell me what you find.