How we take decisions without managers and teamleads

We don’t have any middle management at sipgate. There is still a boss – bosses even, plural – and we look to them for strategy and long term planning, but they are not involved in day to day decisions. In our ~12 product teams no one is the boss of anyone else. This often prompts the question “So, how do you take decisions around here? Who decides things?”

That’s an excellent question that I had absolutely no answer for, the first time I got it. I had never thought about it, because decisions just … get taken. But how? And by whom?

Once I started thinking about it, I realized that the answer is not straightforward, because it’s always different groups of people that take any given decision. The guiding principle is:

We want decisions to be taken by the people who have the most information to make a good decision.
Sometimes that is a single person. More often it’s a pair or a team. Decisions that are more far-reaching but affect mostly one job role can be taken by this role’s community, e.g. all developers or all UX designers.

What happens when people can’t agree?

Our teams have many degrees of autonomy to take decisions about “their” product. Most decisions are taken together. When the team can’t agree the decision falls back onto the most appropriate role: The Product Owner has last say on product feature, time budget etc. Developers have last say on technical implementation etc.

A smart team and Product Owner take care not to force too many issues, because that always leads to trouble down the road. Resentful people are not doing their best work. Fortunately, we can usually agree at least on a general direction

What about company-wide decisions?

In most cases, the group that can take the best decision overlaps with the group of people that will be affected by the decision. The smaller the overlap, the more you need to consult with the people that will be affected.

A popular way to decide something with the input of many people is to discuss it in an Open Friday session. Let’s look at a decision from about 2 years ago:

At sipgate teams can go out on team events to foster team spirit. A typical team event is an activity (escape room, cooking class, minigolf, …) + restaurant + booze. Up to a certain amount, this is on the company’s dime.

Now, our colleagues in accounting had noticed a worrying trend and pitched the following session: “Last year, we’ve had 1 team building event in the whole year. Now it’s April and we’ve already received 5 team building events. We suspect some of these to be regular ol’ team events but we can’t and don’t want to decide which is which. We’d like to talk about how to handle this.

Wait a minute! “Team building event”? How is that different from a “team event”? Turns out, a team event is is your own private fun, whereas a team building event counts as company time, i.e. you can go home earlier if your team building day is longer than usual.

Anyway, at the time of the session, about 12 people from all over the company met and a short Q&A and some discussion we came up with a list of criteria so that every team can decide whether they are planning a team event or a team building event.

[A team building event is for teams that are in trouble. They typically last a day. You probably aren’t looking forward to them (see “team in trouble”). A team event is typically in the evening and you are happy to invest your own time because your team is awesome!]

Took less than 1 hour and those criteria still stand.

You made a money-related decision just like that?

Well, in this discussion a new colleague wanted to “delegate” the decision to a higher level: “Management has to decide that! That is not something that we can decide!”.

Not true. I could vividly imagine what would happen if we really bothered a higher level with that. The higher level would have been C-level, since there’s only 2 levels, C-level and everyone else. They would have looked at us like we’re crazy people and told us to decide for ourselves. They don’t want to think about minor operational stuff like that.

The same applies to at least 90% of issues that I’ve heard the “someone higher up has to decide that” bit by newbies. We absolutely can take many of these decisions and we do. We usually have more information and are also the ones affected.

It’s part of acclimatizing at sipgate to stop relying on higher ups. You can take your own decisions. You also have to. With great power comes great resonsibility.

What are the drawbacks of this distributed decision-making?

Problems arise, whenever it’s unclear who are the best people to take a decision. Depending on  personality some drop issues, others talk to a looot of people to try to reach a decision.

Another is that we are used to great autonomy. When team decisions are overridden by C-level it’s usually not taken well.

You might think that “You have to discuss, so this approach takes longer” is also an issue. I don’t think it  slows you down for two reasons:

  • A lot of decisions affect only 1 or 2 people and you can take those without asking anyone. You need a software tool for 50 bucks that will save you 1h per month? Buy it! The time to run this by anyone else is more expensive than the tool
  • Execution is faster, because we all agreed on what we want to do, instead of dragging our heals on executing something we had no say in and think is a stupid idea by someone ignorant

Would I trade this for a traditional top-down decision making process?

Hell, no! I admit, there were times (~2011) when I longed for someone to swoop in and just take a decision so that we’re done discussing. Now, I’m very happy that the powers that be deliberately created a decision vacuum until we had all learned to take widely supported decisions together.

Overall, I think this appoach makes us faster and results in much better decisions.

Addendum: I’ve just realized that this can only work because we’ve got a strong sense of “how we do things around here”.

PS: Like what you're reading? Join the Finding Marbles newsletter to get new blog posts in your inbox!

Share this article:

Enjoy the Silence

My proudest moment as a facilitator was when I did nothing.

Well, outwardly I did nothing. Inwardly I held the silence. I’m a chatty person. Shutting up is hard for me. Letting silence be. Enduring it. I counted from 10 to 20 in my head and when I had reached 20 I started over. I looked at them intently although they had technically already spoken about the problem. And then the magic happened!

The person that had talked before spoke up again. An additional thought on the topic that made her very concerned. We spend quite some time talking about what she had brought up. It was that important.

I don’t think she would have addressed the issue at all without that long pause. We would have nodded at each other in agreement, closed the topic and started another one, with that issue simmering on. Festering.

Very interesting things happen when you don’t fill an awkward silence with chatter. You’ll hear things that others wouldn’t have come forward with on their own. You’ll give others space to think through thoughts they hadn’t thought through before. It’s magical. Enjoy the silence!

What was your proudest moment as a facilitator?

PS: Like what you're reading? Join the Finding Marbles newsletter to get new blog posts in your inbox!

Share this article:

Mistakes were made – 3 things not to do when going agile

There are many awesome practices we do at sipgate that I love to share with others. But we’ve also gotten some things very wrong, especially in the beginning in 2010. Let me spill the beans about 3 “do not recommend” things for a change 🙂

Double duty Scrum Masters

All the first Scrum Masters were recruited from existing employees and they stayed in their old role for 50% of the time. I was 50% UX designer and the others were developers. This introduces a huge conflict of interest between SM duties and being part of the development team. And the short term “gotta make the sprint goal” interests will nearly always win against long term “team’s gotta learn and improve” interests. Can’t recommend.

Dedicated Scrum Masters are a far better idea.

Afternoon temp teams

Sometimes you need a few special to work on a small project for a short amount of time. We call these task force teams “temp teams”. In the beginning we had temp team members work in their “real” team in the mornings and the temp team in the afternoons. Yeah, multi tasking late that wastes a lot of brain cycles. Can’t recommend.

Nowadays temp teams work together all day. They finish the project faster this way and return to their “real” team.

1 shared backlog for all teams

We had 5 teams, ~3 products and just one backlog. The product owners wanted to flexibly allocate teams to products depending on their changing priorities. Didn’t work out to well. The teams couldn’t identify with any product nor establish testing or coding styles for themselves. A sprint later another team would continue working on “their” feature and “ruin” it.

Today each team only works on one product. They can make it theirs, learn its ins and outs and own successes and failures.

What are things that didn’t work out for you? What do you do now?

Share this article:

Activities to say farewell and reminisce in a retrospective

Teams go through stages. They form, clash, perform well, quarrel, perform, and so on. Until they  eventually disband. Tuckman called this stage “Adjourning”.

A while ago, Stephane asked for activities for a retrospective for an adjourning team. Here are some suggestions:

My team is awesome” would be a great opener.

Appreciative Inquiry” would also work well, if the questions were tweaked a little to serve the purpose.

If you’ve got a budget to take the team out to eat, “Retrospective Cookies” are an excellent option.

I guess it depends a little on whether you’d rather the team takes away a farewell lesson that they can carry over to the next team or you’d rather let them bask in their team spirit and appreciate each other.

What activities do you recommend for an Adjourning retro?

PS: Interested in retrospectives? Sign up to the Retromat newsletter to get related news and tricks!

Share this article:

The Missing Manual for Retromat: 5 tricks that might come in handy

Every once in a while I realise that others aren’t aware of certain features in Retromat. These feature are not exactly hidden, but they are not super obvious either. I’d like to share them with you, just in case there’s one you didn’t know.

0. Step through all activities in a phase with the arrows

For completeness sake, let’s start with the most obvious one. I won’t even count these as a trick: The arrows:

Because in a random plan the activtities will not work with each other, the arrows let you step through all activities in a given phase. Pick activities that fit with each other. If you’re a beginner, try this plan.

1. Click on a phase’s name to see all its activities

The arrows are too slow? You want to quickly scan all activities in a phase? Just click the phase’s name.

2. Check out a single activity by clicking on its ID

Let’s say you want to share a single activity with a colleague: Click on its ID and voila 🙂

3. Click those buttons!

Noticed the row of buttons?

This one generates a new random plan:

It’s supposed to be a wheel of fortune, in case you were wondering.

And then there’s the search buttons:

The left-hand button is just there to be a visible way to search for a specific ID.

The right-hand button lets you search for words and IDs. The search algorithm looks at  titles, summaries and descriptions. If you enter more than one word any one of them is enough for a match. It’s an “or” search, not an “and” search.

As soon as you enter a number as part of your search term, you’ll get the activity of the same ID among the search results.

I’ve been wanting to kick out the ID search button for years but have never really gotten around to it. It’s not really needed.

4. Change the ID to get the activities into a custom order

From the moment I first conceived Retromat I wanted plans to be easily shareable. That’s why each plan has an ID. And you can change the order of activities by changing the ID around. You can change the ID either in the browser’s URL field or by clicking into the Plan ID:


This comes in really handy, when you want to break out of the 5 phases. You see, Retromat is somewhat limited in that each activity can only belong to one phase, although many of them arguably fit into several. If you want to repurpose an activity for a different phase than I sorted it in, go ahead, change the IDs around and press enter!

5. Something completely different – An Easter Egg

There’s another way to forgo phases: Did you know that Retromat has a secret phase called “Something completely different”? You’ve got a 1 in 25 chance of getting a plan with just 1 single activity – one from this extra phase.

Wanna take a peak at these unconventional activities? Here’s a list of all activties in “Something completely different”. You’re welcome 😉

Were any of these new for you? Did I not mention something you would have liked to see? Tell me in the comments!

PS: Interested in retrospectives? Sign up to the Retromat newsletter to get related news and tricks!

Share this article:

Activities for Checking up on Action Items

There are a lot suggestions for Retromat that I can’t  include. Sometimes because the strict 5-phases format can’t accommodate them. One example: Anja Schwarzpaul developed the  following activities for “the new Phase 2” that I call “Phase 0” aka “the phase in which you check what happened to last retro’s action items”. So far, there are very few activities for this new phase described out there. That’s why I’m extra excited to have Anja’s permission to share these two with you!

Her reason for coming up with these?

I feel it’s important to analyze successful or completed experiments in at least as much detail as failed or incomplete ones. Success doesn’t just happen. There’s always a reason. Real life success example in my team: The phrasing was clear and concise, leaving little room for misunderstandings and making the item easy to follow.

And here are Anja’s activities in her own words:


Flow Chart

Use a good old fashioned flow chart to dissect a single action item. (Probably scales to 2 or 3 actions). Duration is flexible and largely depends on the number of questions.

Photo courtesy of Anja Schwarzpaul

From a start node, draw an arrow to a decision node labeled “Done?”or “Success?”. Now branch to “yes” and “no” paths along one or more boxes containing questions to be asked. Near the end of the diagram, merge both branches into a final box “Anything else?” and end in a final state.
Follow the path that the team indicates. If the result is ambiguous, use the “no” branch until just before the merge, then the “yes” branch.

You can either display the entire diagram at once or draw it as you go along. I outlined the start and decision nodes with a marker and sketched everything else with a pencil, in real time outlining only the path we used. This allows for adapting the question(s) to the situation.

Possible Questions:

  • Why did / didn’t it work?
  • How did / didn’t it work?
  • What could we have done to make it work?
  • What does it do for us as a team?
  • Is this something we can use / try again…
  • on a regular basis?
  • in a different context?
  • at some point in the future? (for non-continuous activities, e.g. release estimation)

Improve the Improvement

Suitable for 2 or more action items. Duration depends on the number of items and questions.

Write each hypothesis / item / experiment on a large-ish index card or sticky note. Lay them out on the table or stick them to a wall or board. Let the team rank them from most to least successful, top to bottom. Now ask a few strong questions to help the team analyze the outcome of the experiments. The goal is to get some general ideas of why and how experiments work, and put these ideas to use during the “decide what to do” stage, thus improving the improvement.

Possible Questions: What would have had to happen in order to…

  • make the least successful item come out on top?
  • reverse the order?
  • make all items an equal success?
  • move item <no.> move up a spot?

And maybe:

  • Under which circumstances would you not be able to rank the results?
  • How do you feel about the success to priority ratio?

If I ever have more time (fat chance…) I’ll figure out a good UI to include the new phase in Retromat. Until then, thank you Anja for sharing these with us!

PS: Interested in retrospectives? Sign up to the Retromat newsletter to get related news and tricks!

Share this article:

And what else?

Have you ever had to broach a difficult topic? You prepare, you make notes and you head for a 1:1. You start by talking about various smaller topics. You avoid the big fat whopper of a topic that you’ve been worrying about. Time is ticking by. You’re stalling. Then at the end of the – inconsequential –  conversation, there’s this little pause when you would usually leave and you have that inner debate on whether to mention THE TOPIC or not.

Well, I can’t help you with plucking up the courage to mention the topic. But I can help you if you’re on the other side of this potential conversation. That is, I’ve got advice for you if you would you like to hear about a serious problem sooner rather than later.

"And what else?” teases out difficult topics

It’s simple. Ask “And what else?”. Ask it often. It invites hesitantly shared information.

Oh, and yes, “And what else?“ is a better phrase to use than “Is there anything else?” It’s open instead of closed and errs on the side of implying that there is indeed something. So, sharing is more likely to happen 🙂

It doesn’t just work in this particular situation. “And what else?” is a great phrase in general, because it gives others the opportunity to reflect and dig a little deeper than the top of their heads. I picked up in an excellent workshop on solution-focused coaching by Sinnvoll Führen.

PS: Like what you're reading? Join the Finding Marbles newsletter to get new blog posts in your inbox!

Share this article:

Trump, sad, glad

There are a lot suggestions for Retromat that I can’t  include. Most commonly I decline because an activity is a variation of one that’s already in – just like the following one. I thought it was a fun one (gotta keep a sense of humour in trying times …) and asked its creator Leszek Blacha to publish it here so that you, dear readers, might use it. Those of you outside of the US, anyway.

Trump, sad, glad

This is a variation of “Mad, Sad, Glad” based on a certain Mr. Trump who has a distinct way of talking. Therefore the new categories are “Total disaster”, “So sad”, and “Great (again)”.

Let us know in the comments whether you got laughs for that one. Thank you for sharing, Leszek!

PS: Interested in retrospectives? Sign up to the Retromat newsletter to get related news and tricks!

Share this article:

So what? – Look for the real problem

Recently this statement raised my inner alarms: “We’ve got lots of problems! For example, nobody is pair programming.”

Why would this rub me the wrong way? That nobody is pair programming? After all, I am indeed a huge fan of pairing up. I witness this practice’s many benefits every single day at work. But no, that’s not the point. My alarms went off because “lack of pair programming” was presented as an actual problem. It’s not.

Let me repeat that: As much as I love pair programming, not doing it is not a problem in and of itself. Rather, pair programming is a possible solution to a host of problems an organization might be having such as:

And depending on the actual underlying problem there are different solutions available, one of which could be pair programming.

Nobody who’s not already a convert will start pair programming just “because”. Instead go looking for the actual problem. Ask “So what?” That phrase is magical and you can use it repeatedly. Just like there’s the Five Whys, dig deeper with five “So what’s”:

“Our problem is that nobody’s pair programming!”
“So what? Why is that a problem?”
“Nobody knows anybody else’s code. It’s 1 system = 1 developer.”
“So what?”
“Whenever a developer is sick or on holiday development in their area comes to a screeching halt. And there’s always someone sick or on holiday. Makes us super slow to release. And I dread the day someone quits.”
“Well, that does seem like a pickle…”

Okay, it weren’t five “So what’s” because I suck at making up examples but you get the point.

"So what" is a magical phrase to find an actual problem

This is not specific to agile practices either, though Agile folks have a reputation for dogmatism. Here’s a recent example from the field of web analytics: “We’ve got a huge problem: We can’t do cross-domain tracking.” Soooo …? What are the questions I want answers to and that we can’t answer because we lack this?

I’ve learned “So what?” in the context of Henrik Kniberg’s Cause-Effect-Diagrams and have since used it whenever I suspect that a “problem” someone presents is actually their (preferred) solution. It’s the same as with product features:

“Love the Problem, Not Your Solution”
Ash Maurya

PS: Like what you're reading? Join the Finding Marbles newsletter to get new blog posts in your inbox!

Share this article:

What is an Agile Mindset? Six years later

At the end of 2011 I wrote about what makes an agile mindset (in my opinion) and even made a fancy infographic about it:

It concentrates on how people think about their colleagues as humans vs. cogs; whether they have a growth vs. a fixed mindset; iterative product development vs. extensive planning and more. These are all still valid, but I can add another set of examples today.

The past few weeks I have often thought about how conversations changed at sipgate over the course of the years and why it is much easier and more fun to get things done than in the beginning. So here’s a list of behaviours and how they changed:

“It can’t be done!” -> “We have to take X into account”

In the beginning, many a discussion about potential features revolved around a central theme of what we can’t do. There seemed to be an awful lot of things we couldn’t do, despite the fact that we were working with software. If we created it in the first place, we can also change it. Made me so angry, I ranted about it here.

These conversations have been replaced by “If we want to do that we have to take care of X and Y. Oh, and Z will be tricky, too. From the top of my head we look at a two month effort. Is this feature worth two months to us?” A much more productive conversation!

“Clearly A is better!” “No, B!” -> “Can we just try it?”

We used to discuss options endlessly. Fruitless hypothesizing. Nowadays one of us will rather soon ask something along the lines of: “Can we try it out? (What would it cost to just try? Can we decide this or who else would need to take this decision with us? How easy is it to roll this back if it doesn’t work out? Who might we confuse with this?)”

And the number of theories you can easily put into practice to see what happens is surprisingly high. I’d estimate about 90% of the time we realize that, yep, we could just try it out without repercussions.

? -> “Where’s the value to customers?”

Here I’m not sure what was the focus before. I suspect there wasn’t any focus. But nowadays if you want anything done you’re better prepared to explain how it is of value to a customer. Otherwise, fat chance!

As my colleague Mathias so beautifully put it about how we build websites: “At first we designed desktop first, then mobile first, then content first and finally: Purpose first.” What is it that we want the customer to achieve on a given page? This approach makes decisions and trade-offs clearer and points you into the direction you need to take.

Summary: Appreciative and constructive

In general the whole company’s vibe has become much more appreciative and constructive. There are hardly any cynics left. Instead we’re pointing out what’s already going well and look for solutions where it’s not. Most days, anyway. Nobody’s perfect 😉
It is a highly satisfying way to work!

PS: Like what you're reading? Join the Finding Marbles newsletter to get new blog posts in your inbox!

Share this article: