Magic Estimation – rapidly estimating huge backlogs

As the scrum product owner of the german Telefonica online shop I needed a backlog of 25 stories estimated quickly for an upcoming project – I had one hour. So I researched estimation methods that would eventually enable me to do so. With our usual approach of traditional planning poker and discussing and refining stories in the process, I would have missed that goal grossly as it usually takes us 10-15 minutes for estimating a mid sized story.

Recently a co-worker from another unit mentioned the concept of magic estimation to me, so I decided to try that approach for the challenge I was facing.

So how does it work?


I already had a rough version of the backlog put together on a confluence page as a draft, which I would eventually move into Jira. It consisted of the user story descriptions (e.g. „as a customer I want to X in order to Y“) and some bullet points for better describing the objective of that particular story. I printed the page and cut it into slices so that every story was available as a strip of paper. Then I glued these onto Post-it notes.

The day before the estimation I asked the team to prepare by reading the backlog on the confluence page and by reading a blog post explaining the concept of magic estimation that I thought was useful.

Estimation process

In our weekly backlog grooming meeting I put Planning Poker cards onto the wall in the right order, as table header – the usual 1,2,3,5,8,13,20 sequence. Then I randomly handed all the stories to the 7 team members – 3-4 stories each. I had explained the whole backlog to the team briefly and answered some questions, so every team member was able to visualize what the project was all about. The team members estimated their stories quietly and put their story cards underneath the corresponding planning poker card. Within 5 minutes my first complete estimate of the backlog was on the wall. We wrote the current estimation onto each card.

In the next step, the team members moved the cards around if they were not happy with the current position on the wall – no discussions allowed. The new estimation was written onto the card right after moving it. Some cards were moved back and forth a couple of times indicating that it probably wasn’t clear enough.

After 30 minutes the team was happy with the setting, the position of the cards was stable and I had the complete backlog estimated.

Mission accomplished!

Success factors

  • the team should be experienced in planning poker and have a mutual understanding of complexity and effort
  • the team should know in advance what the backlog is about, have a clear vision of the objectives of the project
  • The team should be familiar with the basic process of the estimation method
  • the product owner has all the stories prepared on story cards, so that the estimation can start immediately
  • questions should be asked and answered before initiating the estimation process
  • no discussions allowed during the estimation phase

3 thoughts on “Magic Estimation – rapidly estimating huge backlogs”

Comments are closed.