How-to WSJF prioritization: How the workshop works

Prioritizing with the Weighted Shortest Job First. How exciting that is as an introduction! So let's put it this way: isn't it annoying when everything is urgent, everything is important and everything is too much? And then it really helps to prioritize with WSJF.

This article is the third part of a trilogy. In the first article I have the 1×1 of prioritizing illuminated, in the second article the WSJF.

Now it's all about the application. And with a practical example like this, what could be more obvious than using a real workshop? And then deduce what is transferable? Nothing, right?

A real example

Let's do that then. First a little background and of course I have to change a few details here because otherwise it's totally obvious with which customer the whole thing took place. It's about a company that does something with software. So far, so general. However, this company didn't used to do this, as software was really new territory. And so there is a "digital division" there.

Yes, you read that right. An area that does digital. The question is, of course, what the others in the company do. But that's not our topic right now. So since there is this digital division, you can imagine that it is allowed to work with the rest of the company. Must. Or to put it another way: everyone wants something from the digital division.

And that's exactly what made the work there so difficult. There were so many topics that were "in progress" that lists and lists were created and endless meetings were held to coordinate them. It was clear to everyone that having a common prioritization would avoid a lot of waste (for those with an affinity for lean: muda). We were welcomed with open arms with our idea of creating such a vote in two days.

About the workshop

Of course, this also meant that we had to deliver accordingly. Almost as much effort went into the preparation of the workshop as into the workshop itself (roughly 80 man-days for the preparation and 95 for the implementation and follow-up).

The room set-up

Around 45 people were involved, including 20 product owners, 15 stakeholders from the "non-digital areas", five vice presidents, one senior vice president and, fortunately, three people with whom I prepared the workshop and who gave me great support throughout the two days - it was a real collaboration across company boundaries! The workshop took place before Covid-19, so everyone was in the same room, incredible, right?

The agenda of the workshop

Day 109.30: Doors open
09.45: latest arrival time
10.00: Opening (by the facilitator)
10.15: The WHY of the workshop
10.30: Strategic outline for the business unit
11.00: The HOW of the workshop
11.20: Groups' and space setup
11.40: First estimation round - sort items into categories
12.30: lunch break of 60', followed by 30' of individual work time
14.00: "Bet on the horses"
14.15: Q&A on the portfolio
15.00: Second estimation round - estimate the top 5% Epics
15.30: Coffee break with time for chats and Mario Kart
16.00: Third estimation round - estimate the top 5% Epics
17.00: (En)lightnings
17.30: Adjourning
17.45: Free time/Operational Review with SVP, VPs, deputies, and reps
19:30: Dinner at "a nice place"
Day 209.30: Doors open
09.45: latest arrival time
10.00: Opening (by the facilitator)
10.15: Presentation of the current standings
11.00: Fourth estimation round - estimate the next 10%
12.00: lunch break of 60' initiated by 30' of individual work time
14.00: Energizer - get in touch with the latest product
14.15: Fifth estimation round - estimate the next 10%
15.00: Prepare for the presentation of the top 5% Epics
15.30: Coffee break with time for chats and Mario Kart
16.00: Presentation of the top 5% Epics
17.00: Flag risks, dependencies, and follow-ups
17.30: Closing
18.00: Debriefing with SVP, VPs, deputies, and reps
The agenda for the two-day workshop

From the sewing box of a moderator

Before I talk about the WSJF, I would like to share a few tips for such workshops - by the way, I am also happy to receive further advice.

  • Always plan for a buffer. Everything takes longer, needs changing times. Hence this beginning with "Doors Open".
  • Give people time to go about their daily business. It doesn't just last for two days. And so that people are more focused, it's good if they know the times when they can make phone calls, write, etc.
  • Plan generous breaks. Sometimes important things happen just when casual conversations are taking place. Break time is valuable time!
  • Have a programmatic defibrillator ready. Early afternoon is the circadian low point for many people, the low point of the day. Especially after a hearty lunch. Fresh air and something stimulating helps. At this workshop on the first day there was my favorite "Bet on the Horses": a big screen, Mario Kart and some joysticks. The fun was over and bets could be placed. That brings people together.
  • Plan a session with all participants to check whether the agenda is still appropriate or whether something important should be noted - here the item "(En)lightings" on the afternoon of the first day.
  • Always carry out a debriefing to get quick feedback and make adjustments - here at the end of the first and second day.

WSJF - how-not-to

There are no best cases, but there are a few things that are sure to lead to failure:

  1. Do not have delimited items (here: epics).
    Instead: The workshop involved preparation by many people. The product owners had also prepared themselves and created the epics using a template (based on this Benefit Hypothesis Statement) formulated.
  2. Investigate WSJF in too small a circle.
    Instead: Bring in as much specific expertise as possible. In this workshop, the stakeholders for Business Value, Time Criticality and Opportunity Enablement/Risk Reduction were important. If there are too few people, there is a lack of expertise and this will come up like a boomerang at a later stage and bring uncertainty back in.
  3. Determine WSJF in too large a circle.
    Instead: In preparation, identify the necessary stakeholders here with the POs. In my opinion, the VPs did not have to be there. Too many people make decisions more difficult.
  4. Estimate everything
    Instead: Estimate only selected items. Here, 180 were brought in and even the top 5% were just under 10 epics. It's better to work on the most important ones in more detail than all of them. This is work "on hold" and dilutes the assessment quality of the next activities.

Focus, focus, focus - round one

The announcement (agreed in advance, of course) that only the top 15% of epics would initially be estimated caused an uproar. This made it all the more important to develop a procedure that supports the selection.

To find the most important ones at the beginning, we used the following trick: Excel. Controllers hate this trick.

But it was precisely the opportunity to work with all groups at the same time that was the key to success. For the very first estimation session, groups were able to form in a self-organized way, the only requirements were that every table should be occupied, one seat should remain free at each table and there had to be at least one PO and one stakeholder in each round. The rest was self-organized, including and especially changes at the tables. The fact that everyone wanted to place as many topics as possible meant that the discussions were self-limiting.

In changing constellations, the groups drew epics and evaluated them according to the WSJF. The WSJF (according to SAFe) is made up of four variables:

how-to WSJF

For this first round, we have enabled a very simple prioritization (very easy to pre-configure using the drop-down menu in Excel): very high - high - low - very low.

first round estimates

The values could be selected at each table for each Epic. So several tables could estimate the same epic at the same time.

distributed estimations

The data was aggregated in the central sheet according to the arithmetic mean. To do this, simply translate the above scale into values from 1 to 4. The job size was estimated in advance by POs and devs.

aggregation of the estimates

The clever move here was that the epics that were considered important by more people were given more and higher ratings. And of course there were epics that were not touched at all. In this way, the focus arises almost by itself.

Who prioritizes what?

After an initial categorization was carried out in the first session, the top 5%, i.e. the top 10 epics, could be derived from this. Much more time was devoted to these. Once again, the tables were able to arrange themselves as required. This time it was clear at which table which epic would be discussed.

This demonstrates the strength of self-organization:
Business value tended to be assigned by stakeholders from the divisions, meaning that many stakeholders came together at one table at times. Time Criticality, Opportunity Enablement/Risk Reduction were valued more or less equally by the stakeholders and POs.

Rounds two and three

In rounds two and three, we estimated more precisely, i.e. not in the four categories above, but using the simplified Fibonacci sequence (0-1-2-3-5-8-13-20-40-100). First, we started with this for the Epic with the lowest "Business Value", for the Epic with the lowest "Time Criticality" and for the Epic with the lowest "Risk Reduction/Opportunity Enablement". The vote was also conducted using the Excel document mentioned above and here, too, we asked those people who were able to vote to do so. It was also easy to see that the most votes were cast for Time Criticality - everyone had an opinion on this 😉

The smallest value in each column is the reference for the others: "If this has a one in Business Value, then this has a three". This estimation in relations is in line with human thinking.

The estimates for the top 5% gradually filled up, sometimes faster, sometimes slower. It was interesting to see how people got into a "working mood" and busily moved from table to table with their shirtsleeves rolled up and their hair tied back. It felt like being on the trading floor back then with a telephone:

Rounds four and five

Anyone who has carefully studied the agenda will have noticed that on the second day the next 10% were estimated after the WSJF.

Why the next 10%? The idea behind this is as follows: With a lead time of over three months and a WiP of around 10 epics, an estimate of 5% is "enough" for at least three months of work.

So after three months, the epics that are between the top 5% and top 10% today would be estimated. Another 5% are another three months further into the future. The estimate of the top 15% therefore extends at least nine months into the future and thus two more such workshops.

If new epics appear in between, these can be included, as the "estimated stock" is always sufficient. At the same time, this stock should not be too large, as the conditions for a WSJF assessment are constantly changing.

Last but not least, it was a signal that the first 5% should receive as much attention as the next 10%. In this way, we also conveyed the principle of rolling planning.

A few more tricks and tips

Hopefully you can now imagine the rest of the workshop. I would like to emphasize a few points again: The two sessions at 5pm each are incredibly important. It may well be that the workshop process leaves gaps. And then that little red light that comes on is crucial. However, as this is also a point that is often used for discussions, I make the simple rule that everyone who submits a report is then responsible for ensuring that the point is followed up. Should, ought and could are thus largely eliminated.

These people also come to the operational review, where it is assessed whether the process enables the desired outcome. Those who come forward, which is another trick worth mentioning, are also thanked for doing so. They are celebrated and we have even managed to give them a performance bonus.

And virtually?

Yes, those were the days. All the people, bacteria and viruses in one room. And anyone who thought back then "it doesn't smell so fresh here anymore" thinks to themselves in 2022 "I think I'm breathing in another person right now".
Better virtual then. No problem!

Transferring this workshop to the virtual world is easier than you might think. What you need: A video platform (I think the three-dimensional, spatial ones are great, welo or similar), a collaboration document (I find the interactive ones, e.g. Office 365 or Miro great. Then create an excel file like the one I have shown above and set up the rooms. It is important to have a plenary room where someone from the organization is always available and can coordinate.

Send out the invitations. And that's it.

What tips do YOU have?

What experience do you have with WSJF?

What tips do you have for a successful WSJF workshop?

Everyone who takes part in the discussion here will be entered into a prize draw for a session in which we will work with you to develop the agenda for such a workshop.

How-to WSJF prioritization: How the workshop works
wibas GmbH, Vincenzo Parisi May 5, 2022
Share this post
Tags
Archive

Write a comment

Submit * mandatory field
My three most important learnings from Kniberg - how you can improve your Kanban