Topics
Estimates ask "how long will this one card take?" and are usually wrong. Phase durations ask "how long does any card sit here?" and are usually right.

Per-item estimates are theatre. Per-phase durations are measurement. The difference shows up in your sixth month, when one matches reality and the other doesn't.
The classic shape: someone asks "how long will this card take?" and the team answers with a story-point or an hour count. The number is anchored on the optimistic case (no blockers, no rework, no surprise scope) and the team almost always under-shoots reality. The tool dutifully tracks the variance and the team's confidence in the tool decays.
The deeper problem isn't optimism. It's that items vary too much for per-item estimates to be useful. Every wedding shoot is different; every weld is different; every batch has its quirks. Asking the team to estimate them is asking them to forecast variance they can't see.
A phase duration is the typical time any card sits in a given phase. Not how long a specific card will take - how long the phase tends to hold whatever lands in it. The intake phase: 30 minutes regardless of what kind of project comes in. The QC phase: 2 hours per item, give or take. The deliver phase: same day. The numbers come from observation, not guess.
Production work is, almost by definition, similar items moving through the same phases repeatedly. Each item varies, but the phase shape is stable - that's what makes it production rather than projects. Phase durations exploit this: instead of forecasting per item (impossible), you forecast at the phase aggregate (tractable). Five items in shoot * 4 hours per shoot ≈ 20 person-hours of shoot work in flight, regardless of which weddings they are.
How to bootstrap the numbers
Don't start from estimates - start from observation. Run two weeks without any phase duration set. Production Board records when each card enters and leaves each phase; at the end of week two you have an empirical median per phase, which is your starting duration. Refine quarterly.
Two failure modes. Too coarse: "production" as a single phase tells you nothing useful - the average across intake + shoot + edit + deliver is a meaningless midpoint. Too fine: ten phases of fifteen minutes each makes the team spend their day moving cards. The sweet spot for most production teams is 4-7 named phases, each with a duration on the order of hours-to-days. If the phase typically takes minutes, fold it into a checklist on the previous phase.
Phase durations enable two kinds of forecast that per-item estimates cannot. End-to-end time per card: sum of all phases ≈ how long a typical card takes from intake to delivery. Capacity check: items in phase × phase duration ≈ work hours in flight at that stage. The first lets you tell a customer when their thing will be done; the second lets you tell yourselves when the next intake starts to back up. Neither requires the team to estimate any specific card.
FAQ
Free plan, no credit card. We host in Germany. You can export and delete everything self-serve.
Read next
WIP limits, plainly: why constraining work-in-progress speeds the team up
What WIP limits do, when to set them, and the numbers that work.
Read
Production board vs project management: knowing the difference
When work is *flow* through stages, not a *plan* with deadlines.
Read
Best lightweight productivity boards for small teams
Curated picks for two- to five-person teams.
Read