estimation improvement in scrum
TRANSCRIPT
ROOTS - WIDEBAND DELPHI METHODExperts are getting specifications
Discussion happensEvery expert fills the form anonymously
Discussion happens about why estimations varyEvery expert fills the form anonymously
repeat
PLANNING POKERDiscuss the taskPick a card with a numberGive min and max proposals a soapbox timePick a card againRepeat until a consensus is reachedRecord a number
ATTRIBUTES OF A PLANNING POKERBiased opinionsDunning-Kruger effectAll are the experts in their field of knowledgeRelative estimations or real units estimation (days vsstory points or t-shirt sizes)Repeat until a consensus is reachedShare knowledge and cover dark corners is the goal, notnumber itselfnumbers can help to make forecasts
That's normal!We work as a Team because only together we knoweverythingCross experts communication is key. e.g. PO -> DevTeam,UI/UX -> POlack of expertise does not make your opinion lessvaluableplace to talk about the scope to align everyoneChinese whispers effect minimized
DUNNING-KRUGER EFFECT"If you're incompetent, you can’t know
you’re incompetent. The skills you need toproduce a right answer are exactly the skillsyou need to recognize what a right answer
is."
ALL ARE THE EXPERTS IN THEIR FIELD OFKNOWLEDGE
You know something I don'tYou share your knowlege. Now everyone knows it...Profit!
RELATIVE OR REAL UNITS ESTIMATIONHoursDaysIdeal Days/HoursStory PointsT-Shirt sizesParrots, carrots, bottle of beers, anything goes
A WORD ON UNITSBeginners tend to use t-shirts, ideal days and suchMature teams can easily juggle with story pointsStory point is super relative, ideal day is very natural
Yes, it is tough!
REACHING CONSENSUS
but "Discussio mater veritas est." someLatin dude said that
Repeat conversation until full consensusKeep up to 3 roundsOn 3rd - pick one biggestor On 3rd - take number by majority
HOW DO WE ESTIMATE AND WHAT IS A BESTPRACTICE?
Best Practice We
As acustomer, Iwant see myorders so thatI can figureout what Ibought.
... has provided an endpoint whichallows for enabling/disabling support tousers by providing a list of Subscriptionassignid's, as opposed to the currentendpoint which only accepts oneassignid.
Best Practice We
estimate effort nottime
estimate effort not time
minimalimplementation
sometimes we try to get bigger,better, more complex
one whole thing -full nine yard
5 cards = 1x Mockups, 2xInvestigation, 2x Implementation
IDEASReject card with no acceptance criteria!Start writing user stories! (bugs can skip that)Define personas, grow attached to themDiscuss user story instead of lifeless card at groomingKeep implementation in planningEstimate minimum implementation. Enough is enoughOn 3rd round pick a maximum instead of majorityDo not estimate task items. Only story itselfUse coffee break card if you want!Follow the roadmap. That is our short-term goaldefinition. Is that card is about our current priorities?