Estimate Discovery, Not Dates You Don't Own
- YT :: https://www.youtube.com/watch?v=obpXJDkzgBY
- Original title :: Stop Giving Dates You Don't Own: What to Say Before You Estimate
When pressured for a delivery date on unknown work, don't volunteer one. Commit to a short discovery, give a range bounded by explicit assumptions, and make whoever needs a date today choose which risk or scope they're trading for it.
The problem: a guess becomes a promise
"Can we ship this in 2 weeks?" - asked before anyone knows the scope, the integration, or what another team owns. Saying "2 weeks" isn't an estimate, it's a date you don't own. Once it lands on a plan, the team defends the number instead of learning the work, and the guess becomes your credibility problem.
Questions like "who owns the data?", "what if it fails?", "who signs off?" are not details - they are the work. Without scope, assumptions, capacity and dependency owners, an estimate is theater. You can own the quality of a forecast; you can't own an untested integration, another team's approval, or an unmade product decision.
1. Estimate discovery first
"It depends" is true and useless. Instead: "I don't have enough information for a date. I can give you a one-week discovery plan, the two assumptions that matter, and a range by Friday." Now you're committed to producing evidence - the cheapest proof (data migration, vendor constraint, identity-provider support) that prevents an irreversible bad commitment. A ticket title isn't a technical baseline, a mockup isn't a design, a sales-call sentence isn't a release plan.
2. Give a range with conditions
"With current scope and no identity-provider changes, 3-5 weeks after discovery. If the migration needs backfill or security changes, it's longer. We'll tighten the range after Friday's proof of concept." That's a forecast plus an update rule, not hedging. A range without boundaries (3-12 months) is useless; one with a scope boundary, two real assumptions and a proof point lets people decide.
3. Force the trade-off into the open
If they still need a date today: "Tell me which risk you want us to accept or which scope to remove." A deadline is useful when it forces a choice - cut scope, add capacity, accept risk, or move another priority. A date with no trade-off gets absorbed privately: overtime, skipped tests, weak rollbacks, unbudgeted support.
Forecast, not promise
Bad estimates don't wreck credibility because dates move; they wreck it because the movement is unexplained and arrives as a surprise. Own the evidence, name the assumptions, give the range, set the next decision point - and stop volunteering as the future fall guy. The video is bracketed by pitches for the author's career coaching.