a p50 date is
not a commitment .
jollygig scheduling turns three-point estimates into a spread of finish dates, so a plan can say "about a 70% chance of nov 14" instead of pretending to know. plan with the middle, promise with the tail, and see what is moving the date.
30 days free · no credit card · one flat price per organization
one date hides most of what you know.
nobody plans badly on purpose. the tools just ask for a single number, and a single number can't carry how sure you are.
a single date hides the spread.
every estimate is a range in someone's head. when a plan shows one date, the range disappears, and the first time work runs long feels like a surprise instead of something you could see coming.
independence flatters the plan.
most simulations assume every task runs long on its own. real projects run long together: the same person, the same vendor, the same unproven library. treat them as independent and the long end of your forecast comes out thinner than it really is.
the longer versions: a p50 date is not a commitment and why independent estimates lie to you.
five small steps, then it keeps itself honest.
you draw the work as activities and dependencies, on a board or on a canvas. they're the same activities either way, so nothing lives in only one view.
- 01
three-point estimates
for each activity you give a best plausible case, a most likely case and a worst plausible case, in workdays. plausible, not imaginable. the shape is beta by default; triangular or lognormal when that fits the work better.
- 02
thousands of runs
we run the plan thousands of times, drawing a duration for every activity and walking the dependency graph on your team's calendar, holidays and partial days included. nothing is worked out by feel.
- 03
a spread, then dates
you get the whole distribution of finish dates, with p50, p75 and p95 read off it, and for each activity how often it sat on the critical path and how much it moves the finish. shown in words, not as a score.
- 04
correlation you actually elicit
you never type a coefficient. you tick what two activities share (an owner, an unknown, a technical area, an anchored estimate) and the forecast draws them together, the way they behave in real life.
- 05
actuals feed back
move a card to in progress or done and the real dates are kept. the forecast re-runs on what's left, and you can watch the chance of your committed date move from run to run.
three dates, three different jobs.
every run ends on some date. line the runs up and these are the points worth knowing. the illustration at the top shows where they sit.
your own expectation
half the runs finish by this date. a fair guess for planning inside the team, and a coin flip as a promise.
a comfortable plan
three runs in four finish by here. a reasonable date to plan around when a miss is easy to absorb.
worth promising outside
nineteen runs in twenty finish by here. this is the candidate for a date you give a customer or another team.
set a committed date and the forecast also tells you the chance of making it, and says so quietly when that chance drops below the level your team chose.
if your team is strained, your schedule should know.
link a workspace to a team in team health and the forecast sits next to that team's own anonymous signals, behind the same five-person floor as everything else. when the team is showing signs of strain, estimates made right now tend to run optimistic, so jollygig offers to widen the pessimistic end by 10%. it's one click, you can undo it, and it never names anyone.
you can use either product alone. together, delivery pressure and team strain show up on the same page.
how team health worksone flat price per organization.
up to 10 workspaces and every feature on this page. add as many people as need to see a forecast; it never changes the price.
use team health too and scheduling is 20% off, automatically, for as long as you have both. 30 days free, no card needed. see all pricing
honest answers about forecasting.
forecast your next project honestly.
30 days free. no credit card. cancel any time.