Agile Estimation and Planning (That Doesn't Suck)

Agile Estimation and Planning(that doesn’t suck)

Craig Drayton@craigdrayton

think sharp elabor8

think sharp elabor8

Why are we estimating again?So we can monitor deviations from the

planned scope and schedule?

So we can measure performance,and hold people to account?

think sharp elabor8

Agile planning has gone horribly wrong.

think sharp elabor8

Two genuinely useful reasons to forecast

Make investment decisionsBased on our best estimates of value, cost and risk,

should we do this thing? When should we do it?

Meet date expectationsAre we likely to deliver by a date that matters to us?

If I find out that we’re not, I can take action.

think sharp elabor8

Estimation and forecasting:The least interesting thing about building a product

Do you practice incremental delivery -you’re working on small batches in iterative cycles, maximising feedback while limiting cost and risk?

Do you have modern Product Management practice - you’re driven by outcome (not output), and you’re

prototyping and experimenting to discover what works?

… well, just keep doing that!

think sharp elabor8

Oh, you’re still here?

If estimating cost and forecasting delivery dates is important to you, there’s good news…

… there are ways to do it that don’t suck.

think sharp elabor8

What is forecasting, anyway?

Forecasting is using past data,to calculate the likelihood of future outcomes, assuming that the future looks like the past.

The most popular agile forecasting method today:

Estimating in Story Points,Forecasting based on Velocity.

think sharp elabor8

What’s wrong with velocity?Average October day in Melbourne:Cloudy, light winds, high of 19.7°

4 October 2016 in Melbourne:

think sharp elabor8

Monte Carlo Simulation1. Start recording the time that you

complete each work item.

2. Calculate how much time passed between each successive completion (‘Takt time’)

3. Marvel at your newfound ability to predict the future

think sharp elabor8

Monte Carlo Simulation4. Simulate your progress through your backlog.

NOW

Backlog

18 Days

think sharp elabor8

Monte Carlo Simulation5. Do this thousands of times.

NOW

18 Days

15 Days

21 Days

18 Days

20 Days

think sharp elabor8

Monte Carlo Simulation5. Calculate the proportion of simulation runs that finish on or before each date.

This is the probability of completion by that date.

think sharp elabor8

But what about estimation??

think sharp elabor8

Do all my stories needto be the same size?

think sharp elabor8

Know why you’re forecasting.Misuse of forecasting is harmful.

If you forecast using an average value,you’ll be wrong half of the time.

Forecasting only works if yourfuture is similar to your past.

Key takeaways

think sharp elabor8

OK, how do I get started?

People have built spreadsheets that do the math & simulation for you:● https://goo.gl/EaJjFr● http://bit.ly/SimResources

Or, talk to me about getting a free beta account with Mazzlo, a predictive analytics and forecasting app for agile delivery.

craig@mazzlo.co

elabor8.com.auThank you!