Forecasting without Fear How to keep the business informed and keep your cool NY SPIN December 15, 2015
Drivers, Challenges why you have to forecast, and why it s not easy what you forecast Refinement for refinement Cost of Delay when you forecast Sprints and releases Sprint forecasts Release forecasts capacity, velocity Delivery date forecasts Fixed cost forecasts scope, durations Forecasting in Kanban yesterday s weather, Little s Law Metrics updating the forecast 2
Drivers for forecasting budgeting initiatives cash flow and ROI trade shows investor relations regulatory compliance 3
What are we releasing, when? What are we releasing next? when?
Fear
Challenges: Work disconnected from vision Team overloading Context switching Inadequate training Low engagement Organizational and cultural barriers Lead times excessive 6
Not Commitment Forecast 7
Refinement Possibilities Domain Portfolio Themes CoD Business Value Initiatives Programs Features Stories Sprints 8
Cost of Delay Components User (feature) value Preferences, revenue impact Time value Decay of user value, upcoming deadlines, penalty for late delivery Learning value Risk reduction, learning technology / domain), new opportunities Reinertsen 9
Weighted Shortest Job First - WSJF A B C User 3 8 3 Cost of Delay Time Learning Total 8 8 19 3 2 13 3 2 8 Reinertsen 10
When you forecast: Sprints and Releases Program Level Team Level Sprints Sprints Leffingwell - SAFe 11
Forecasting Options: Capacity Velocity Little s Law Cost, Burn Rate Yesterday s weather 12
Sprint Forecast capacity based Candidates for this Sprint Tasks Dev CAPACITY for SPRINT Amit Betty Leslie Pat Steve Arthur Ordered Features Ordered Stories Estimate Tasks up to Dev Capacity (or 90 % of it ~ 200) 48 54 48 48 44 50 Dev Capacity 216 Test CAPACITY for SPRINT Charlie Sue Apala 48 54 36 Test Capacity 138 After Mike Cohn 13
Sprint Forecast capacity based Candidates for this Sprint??? Tasks Dev CAPACITY for SPRINT Amit Betty Leslie Pat Steve Arthur Ordered Features Ordered Stories Estimate Tasks up to Dev Capacity (or 90 % of it ~ 200) Features, Stories and Tasks forecast 48 54 48 48 44 50 Dev Capacity 216 Test CAPACITY for SPRINT Charlie Sue Apala 48 54 36 Test Capacity 138 After Mike Cohn 14
Estimating Real Capacity 15
Don t know Velocity? Back into Velocity Story Points Est. Task Hours Accept for Sprint As an actuary 3 64 As a plan participant 13 64 64 As a plan manager 5 96 96 As a plan sponsor 8 20 20 As a plan sponsor 8 As an actuary 8 60 104 104 408 284 CAPACITY dev team usable sprint size hours/day days 5 6 10 Implied Velocity 34 team capacity 300 hrs Mike Cohn Or : 8 s.p. per developer and tester: e.g. 5*8=40 Leffingwell - SAFe 16
Sprint Forecast velocity based Candidates for this Sprint Estimate Story Points for Candidate Stories and Select up to Average Velocity Ordered Features Velocity 3 latest sprints: 30, 40, 50 Ordered Stories After Mike Cohn 17
Sprint Forecast velocity based Candidates for this Sprint V=30 highly probable V=40 probable V=50 unlikely Features and Stories forecast for this Sprint Estimate Story Points for Candidate Stories and Select up to Average Velocity Ordered Features Velocity 3 latest sprints: 30, 40, 50 Ordered Stories After Mike Cohn 18
Theme: Sell Books Initiatives Features Stories Display categories Sgfghfithien see rty Show brands Sgfghfithien see rty Item List Search Purchases Show descriptions & prices Enter key words Populate Shopping Cart Payments Enter Card Number Bill Credit Card Sgfghfithien see rty Sgfghfithien see rty Sgfghfithien see rty Sgfghfithien see rty Sgfghfithien see rty Security New user can register Sgfghfithien see rty 19
Release Forecast velocity based Estimate Story Points in Backlog and Select up to Bank Velocity 40 Ordered Features Ordered Stories To release date x Sprints 6 = Story Point Bank 240 Velocity 3 latest sprints: 30, 40, 50 After Mike Cohn 20
Release Forecast velocity based Display categories V=30 highly probable 30 x 6 = 180 pts Show brands Populate Cart V=40 probable 40 x 6 = 240 pts Bill Credit Card V=50 unlikely 50 x 6 = 300 pts Features and stories Estimate Story Points in Backlog and Select up to Bank Velocity 40 Ordered Features Ordered Stories for this Release To release date x Sprints 6 = Story Point Bank 240 Velocity 3 latest sprints: 30, 40, 50 After Mike Cohn 21
Sizing points or cost Affinity / Analogy Is this feature larger or smaller than one we built before? Large Small Disaggregation What moving parts does this feature have? (not stories) Expert opinion Probably the least robust 22
When can you deliver? Estimate Backlog Derive Duration total story points Backlog # of sprints Calculate Schedule # = + 7 Velocity story points/sprint Backlog size: 450-675 story points Forecast Basis: 600 Jun Oct 16 23
Can you deliver by 7/12/2016? Size all Features Features Derive Points Calculate Sprints 7 Velocity Total of all Features = 450-675 story points start date end date 12/15/2015 7/12/2016 # of sprints: 15 Points Velocity Delivered 30 40 45 450 600 675 24
Fixed Cost Forecast 350 400 250 300 500 200 250 Select Features up to Fixed Cost 350 400 200 Ordered Features with Cost estimates Fixed Cost: 2000 25
Fixed Cost Forecast 350 400 250 300 500 200 250 Select Features up to Fixed Cost Features Forecast to be Delivered for Fixed Cost 350 400 200 Ordered Features with Cost estimates Fixed Cost: 2000 Try to decompose into Releases Re-estimate periodically 26
Forecasting based on burn rate C= sum of all feature costs $ B= burn rate $/day D= duration days Affinity estimation Standing team exists: C/B=D Result: duration in days No team(s) formed: Fixed ship date S C / (S NOW) = B Result: burn rate you will need -> capacity 27
Forecasting in Kanban yesterday s weather Increment 28
Forecasting in Kanban Little s Law Little s Law = = = 1.5 item/h WIP and Lead Time: from history D= = = = + Throughput Rate. / = 120 h =. = 200 real h 12/15/2015 + 200 = 7/2/2016 120 * 1.5 = 180 items Assumptions: 1. Work items are uniform type and size 2. Constant flow of work items 29
Feature Progress 3 4 5 6 7 8 9 10 Sprints -> Leffingwell - SAFe 30
Release Progress 31
So: Refinement Cost of Delay Sprint forecasts Release forecasts Delivery date forecasts Fixed cost forecasts Forecasting in Kanban Metrics 32
Andrew Kazarinoff Qualytic Consulting, New York Andrew.Kazarinoff@Qualytic.com 917-608-0016 33