When will it be done?A forecast from your own throughput, not a guess.
Paste how many cards the team finished each week for the last couple of months and how many are left. Ten thousand simulated futures, each built from your real weeks, give you a date with 50, 85, and 95 percent confidence.
Your numbers
Around Oct 29, 2026. Half of the simulated futures finished within 7 weeks; only one in twenty took longer than 9. Average throughput in your data: 6 cards a week.
How to quote it
Say “8 weeks, Oct 29, 2026, with 85% confidence” and mean it. The 50% figure is a coin flip and should never be the committed date. Re-run every week as the backlog and the history change; the forecast is a living number, not a plan.
How the forecast works
This is a Monte Carlo simulation using bootstrap sampling. For each of ten thousand trials it repeatedly picks one of your past weeks at random, subtracts that week's throughput from the backlog, and counts weeks until the backlog is gone. Sorting the ten thousand answers gives the distribution; the 85th percentile is the number of weeks that 85% of futures finished within.
Because every simulated week is a real week you had, the forecast carries your team's actual variability. A team that ships 5, 5, 5, 5 gets a tight range. A team that ships 9, 1, 8, 2 gets a wide one, which is the honest answer.
Why not just divide backlog by average throughput
That gives roughly the 50% figure: the date you are as likely to miss as hit. Commitments need the tail. The gap between 50% and 85% is a direct measure of how unpredictable the flow is, and shrinking it (by limiting WIP and slicing work smaller) is usually more valuable than any single forecast.
Inputs that ruin it
- Leaving out the bad weeks. Holidays and incidents happen again.
- Mixing cards of wildly different size. Forecasting in counts works when work is sliced to roughly similar pieces.
- A backlog that keeps growing. Forecast the committed scope and add a percentage for discovered work, then watch it.
Forecasting questions
What does 85% confidence mean here?
In 85% of the ten thousand simulated futures, built by resampling your own past weeks, the backlog was finished within that many weeks. It is not a promise; it is a statement about how often a team with your history hits that date.
Why is the 85% date so much later than backlog divided by average throughput?
Dividing by the average gives roughly the 50% date, the one you are as likely to miss as hit. The gap to 85% is the cost of your variability. Teams that limit WIP and slice work evenly see the gap shrink.
Should I estimate cards in points instead?
Counting cards works when work is sliced to roughly similar sizes, and it removes an estimation meeting. If your cards vary by an order of magnitude, slice them smaller before forecasting rather than adding points.
Is the simulation random? Will I get a different answer next time?
The random draws are seeded from your inputs, so the same history and backlog always give the same forecast. Change either and the forecast changes with it.
Get the weekly numbers
without counting by hand.
Kanera’s Work Done view lists completed cards by week for a board or a whole workspace, and Table view exports it to Excel in one click.
No card required, 30-day Pro trial, free plan available after it