9 points that make follow-through more likely
tl;dr Small, concrete, team has control over it, clear first step, responsible person, follow-up date, success criteria; there are only a few AIs, visualize
Retrospectives are only meaningful if they result in change. Sometimes this change is sparked just by everyone reaching a better understanding of everyone else’s perspective. More often we reach change through experiments, i. e. by trying something new for a set period of time. Obviously, you would like your experiments to improve things, but they won’t always. You have to try out many things to find the ones that are an improvement and discard the rest.
Experiments come in two flavors:
- Action Items (AIs) – concrete todos; usually one-time-tasks
Examples: “Invite the devops team to our refinement meeting”
- Rule changes – how the team handles their interactions, routines, rituals or events; usually on-going and repeating
- “everybody will answer these 3 questions in the daily standup”,
- “we will groom upcoming stories every Wed 3pm”,
- “we will prepare the product demo the day before the review”
I’m sloppy and use “action items” to mean both types – yes, also in this post. My recommendations apply to both types.
Okay, so we’re in a retrospective and try to come up with good experiments to try out. But what is a “good” action item?
For me, a good action item is something that has a high chance of actually being implemented by the team. You don’t get brownie points for coming up with ten AIs. You get points for those two AIs that you actually carry out and observe the results of.
Great. And what exactly increases the chances of follow-through for an AI? Glad you’re asking!
Aim for small experiments. Go for the smallest change that could possibly make a difference. Small changes are easier to agree on. They have a higher chance of actually being implemented, because they are not such a big effort. If an experiment works: Great! If it doesn’t you haven’t invested much and can try something else. Rinse and repeat for continuous improvement.
Don’t underestimate the power of baby steps! Small changes quickly add up to big improvements. It’s like compound interest. Additionally, people experience that change is possible and gain momentum. Change is like a muscle: It gets easier with practise. Eventually big changes also become possible.
Let’s assume, the team has a high-level goal such as “We want fewer failed stories”. There are many different ways to get closer to this goal. The team decides to go with limiting work in progress, e. g. “Work on fewer stories in parallel.” I would not be happy with this. How few is “fewer”?
How about: “We’ll work at most on 2 stories simultaneously”. Better. It’s EASY TO CHECK whether or not the team fulfills this.
In Control of the Team
Now it’s time to check if the goal and their strategy to reach the goal is within their control. In any given system there are
- parts that the team Controls
- parts that the team can Influence and
- parts that they can’t change. But they can Adapt to deal with them better
Make sure that the goal and strategy reflect where the teams stands regarding “CIA”. It’s okay to pick a goal that they can only influence as long as their plan is about who and how to influence.
When you’re clear on that, ask them for the first step.
People often lose momentum, when they don’t know exactly how to start. Aim for a concrete change in behavior – including events that will trigger the behavior such as “During our daily standup, we’ll make sure that we work at most on 2 stories at the same time”.
Owner aka Responsible Person
Who is going to take care of this AI? Either by doing it themselves, by finding other people to implement it or by reminding people. For our example this could be: “Timm will add the WIP-check to our standup-checklist”.
If there are no volunteers for an action item, then it is obviously not important enough to the team (right now) and is discarded. Being explicit about not having the capacity or desire to do something is important information. Know thy(team)self 😉
If nobody voluntunteers, “What would have to change so that you would volunteer?” is an interesting follow-up question.
For todos, this is straightforward, until when will it be done.
Rule changes often need a longer period of time to see them in action, before you can review them. So how long will the team try a new rule? When the trial period is up, the team review the rule to see if it solved their problem.
And how do they know that they solved the problem or at least improved? What are their success criteria? “Gut feeling” is an okay metric in my book as long as the team is explicit about it.
So far, all the points were for a single AIs. The next one applies to the set of AIs that come out of a retro:
One of Few
I once heard someone boast that they got 17 AIs out of their last retro. They thought that that wos a good thing. To me, it’s not. There’s a proverb “Those who hunt two rabbits will catch neither”. If you have too many goals you will reach fewer of them than if you had a small number to begin with and are able to focus. Out of a 60-90 minute retrospective we will typically get 2-3 action items. Anything more than 5 would make me very skeptical.
Last but not least:
Find a way to keep the experiments on everyone’s minds. Some ideas:
- Big AIs can become stories in the Sprint backlog
- Visualize 1-time-todos on the team board
- Have a running list of ongoing experiments
- Maintain a “Working Agreement” to list all current team rules
- Calendar reminders
- Added to existing checklists
Keep AIs in (the vicinity) of / on other boards and (digital) documents that the team use on a daily bases.
Summary & Mnemonic
It feels like I’ve got a prime opportunity to coin a mnemonic here. Like “INVEST in Good Stories” this could be “Great Action Items are …”
SCIFORMOV? – Small, Concrete, In Control of Team, First Step, Owner, Review Date, Metric, One of few, Visual Reminder
I’m not impressed. What does SCIFORMOV even mean? And yes, I’ve played around with alternative names for the various criteria. See the “tl;dr” to see the first way I phrased them 😀
What if I change around the order of the letters? I don’t want to change the order in the article, because it reflects the chronological order in which you address these when writing down an AI during a retro. But I guess I can ignore the order for the sake of a mnemonic. So, here we go. I give you… Drum roll please:
Forgive me, it’s late and I think I’m hilarious 😉 And yes, I realize that’s not how you spell “Moscow”. Anyway, great* action items are MOSCOV FIR. You’ve heard it here first.
* “Great” as in “have a high chance of being implemented”
And for full disclosure: I have no idea what the average follow-up on AIs is. For my teams it’s between 60-80% of the action items. Dropped action items often “belong” to problems that sorted themselves out in other ways. So I’m not aiming for 100% follow-through. I think 80% would be cool. But even at 60-80% the teams are happy and improving. So there’s that.
What the average follow-through in your teams? Are you happy with that?