--- title: "Nobody Could Go On Holiday Without Something Breaking" source: "https://virtualddd.com/facilitating-archdes/skill-matrix-knowledge-transfer/" date: 2026-11-10 authors: ["Cosima Laube", "Andrew Harmel-Law", "Kenny Schwegler"] tags: ["facilitating software architecture and design", "team collaboration", "team autonomy", "social dynamics", "psychological safety", "community of practice"] type: "Facilitating Story" episode: "32" youtube: "https://youtu.be/z5y8VYkxkVk" --- # Nobody Could Go On Holiday Without Something Breaking There's a particular kind of complaint that shows up in retrospectives and never gets solved. Someone says they can't take time off properly because there's no way to hand things over. Someone else says they'd happily help, but they wouldn't know where to start. Everyone nods. The topic gets a sticky note, maybe an action item, and then the next sprint arrives and nothing changes. The complaint isn't about process — it's about knowledge sitting in exactly one head, and nobody having a shared way to talk about that. In this episode, Cosima Laube tells Kenny Schwegler and Andrew Harmel-Law about a group of four platform teams she coached during the early pandemic, where that exact frustration finally got addressed. What makes the story worth listening to isn't the tool she used — a skill matrix, hardly novel — but how she got the team to build it themselves, and what happened when they insisted on adding something she hadn't planned for. ## Four Teams, Critical Infrastructure, and Too Much on Everyone's Plate The setting was an IT organisation of around 400 people going through a SAFe transformation. Teams were being reshaped, roles were shifting, and some people were still carrying old titles alongside new ones. Then COVID arrived and everyone went fully remote. Cosima joined as an agile team coach for a group of four teams responsible for the platforms other developers depended on — infrastructure, tooling, what you might call developer experience. Some of it was genuinely critical infrastructure, which meant there were eyes on this group from outside. Her mandate wasn't just to facilitate ceremonies. It was broader: find where the teams could benefit from help. So she started by talking to people individually, and the same themes kept coming up. Too much on my plate. What happens if I get sick. I can't hand over my work well enough to actually switch off on holiday. New students join and it takes forever before they can contribute. Within a few weeks it was obvious to her that knowledge transfer was the problem to address. What she did next is the interesting part. ## She Had a Tool and Deliberately Didn't Introduce It Cosima knew what she wanted to try: a variation on a skill matrix. Listing the skills, tools and methods the team cared about, and letting people place themselves on a scale. She could have brought it in fully formed. She explicitly didn't. "I didn't want to come in and say, 'Oh, I heard from a few folks on the team that this is an issue. Here is a tool that I would like to apply, and so please do it.'" The reason was practical, not ideological. She was an external coach. Whatever they built had to be light enough for them to keep running after she left. Something imposed from outside tends not to survive the person who imposed it. So she worked outward in stages. First, two people who knew the team well — the former team lead who was transitioning into an architect role, and one other key person. Then some of the more influential team members, one-on-one, where she asked a version of the same question each time: "Do you want to hear my idea? It's not finished yet, and I would like to have your opinion on it." She knew from experience that people were far more open in those conversations than in dailies or group discussions. Then, once she had a rough shape, she made an open invitation in a team meeting: anyone who wanted to help finish it could join on a specific date. The group that showed up co-created it, presented it back, and asked everyone else to contribute. By the time it came to actually using the thing, buy-in wasn't something she had to fight for. Everyone had either shaped it or had the chance to. ## The Axis She Didn't Plan For The team agreed quickly on a four-point scale for capability. From "I've heard we use that and have no idea what to do with it" through to "I know it well enough to teach a student." In between: "I could do it if I had to, because a colleague is off sick" and "I know what I'm doing and it goes well, but I couldn't teach it." Then the team asked for something Cosima hadn't proposed. They wanted a second axis: motivation and joy. Because the honest picture wasn't just about who *could* do things. Some people said outright, "I don't want to do Kubernetes at all. I'm doing it because I inherited it, but I would be glad to give it to somebody else." Others said the opposite — that if you took a particular technology away from them, they'd think about leaving the company. So they built a second four-point scale for how much energy each skill actually gave the person holding it. Suddenly the matrix wasn't just a capability inventory. It was a map of where knowledge was stuck, where it was wanted, and where the two didn't line up. Andrew picked up on this immediately — that the team had added the dimension of desire alongside capability, and that this was the part he was going to steal. ## Where She Refused to Budge Kenny asked the question that anyone who has tried to facilitate while also having expertise will recognise. You're brought in *because* you know things. So what happens when the team heads towards a solution you know is worse? Cosima's answer was specific rather than philosophical. There were a few places where she held firm. The scale had to have four points, not five. "So that there's no tendency to the middle possible, so that you really need to decide." People asked more than once whether they could have a neutral option. She said no. The anchor points were negotiable; the absence of a middle wasn't. The list of skills was another intervention, though a softer one. The first draft was enormous, and she let it get enormous on purpose — she wanted to hear what everyone thought was worth discussing. Then she started pushing back on granularity. Is that really a thing we should talk about, or is it too small? Doing X in tool A probably isn't its own line item. Otherwise you end up with fifty rows that nobody will ever fill in. And she guided how they started. With twelve people and multiple single points of failure, the temptation was either to fix everything at once or to fix one thing and call it done. She worked with the incoming architect to identify the riskiest areas — the ones where exactly one person knew a technology and nobody else had a clue — and they picked five or six to work on. Pairs, mostly, and one trio. Each pair worked out their own learning situations: not just talking someone through it, but finding real tasks and modelling sessions where the learner led and the expert sat in the back seat. "I think I would have intervened if they had said, 'We would like to address everything at once,' or just one piece." ## Why It Didn't Feel Dangerous Kenny pushed on safety — was there ever a moment where she thought one wrong move would break the trust she'd built? She couldn't recall one, and her explanation was about the groundwork rather than the facilitation. The one-on-ones had established that people could say difficult things to her. Some of those conversations were seventy-five percent complaining, and her response was consistently to reflect it back and then ask: what would you like to be different? "I cannot guarantee we can change it, but I really value your opinion and I'm here to help you and us structure how to move on." Two other things mattered, and neither was hers. The former team lead was widely trusted and respected, and he made a specific public commitment: the matrix would not be used to rate people. There would be one final round of reviews under the old system, and the skill data would play no part in it. That distinction — this is how we see ourselves in relation to each other, not a list of who's expendable — was what made honest self-assessment possible. And the SAFe transformation, for all its reputation, had shifted incentives from individual goals to team goals. Growing the team's shared knowledge base became one of the actual team goals. As Cosima put it, reflecting during the recording: "I don't know if I would have had the same power if we wouldn't have this team goal." Without it, knowledge transfer is the thing you do when you have time, and you never have time. She moved on to a different department after a while, then a third. She kept hearing that the matrix was still in use, tweaked here and there. ## What Actually Made This Stick - **The co-creation wasn't a nicety, it was the delivery mechanism.** Cosima had a working tool in mind from early on. What she spent her effort on was getting the team to own it, because she knew she was leaving. Ownership was the requirement, not the courtesy. - **She sequenced the conversation from private to public.** One-on-ones first, because that's where honesty lived. Then a self-selecting working group. Then a full-team invitation. Each stage widened the circle of people who'd already shaped the thing before anyone was asked to commit to it. - **The team's addition was better than her original design.** Capability alone would have produced a competence audit. Adding motivation turned it into something people wanted to be honest on, because it gave them a legitimate way to say "get this off my desk" and "don't take this from me." - **She held the line on structure, not content.** Four points, not five. Reasonable granularity. A manageable number of areas to start. She let the team decide what mattered and how to describe it; she refused to let them build something that would produce mush or collapse under its own weight. - **Somebody with authority had to explicitly disarm the tool.** A skill matrix in a reorganising company is a firing list unless someone credible says otherwise, out loud, in specific terms. The incoming architect did that. Without it, the whole exercise would have produced polite fiction. - **Organisational incentives did work the facilitator couldn't.** Team goals rather than individual goals meant learning time was legitimate. Cosima's reflection that she might not have had the same leverage otherwise is worth sitting with — some facilitation problems are actually incentive problems. The part that stays with me is the second axis. Cosima came in with a tool designed to answer "who knows what," and the team said that question wasn't quite right — you also need to know who *wants* to know what. That's not a refinement, it's a different tool. And she got it for free, because she showed up with something unfinished and asked what was missing. Which raises an uncomfortable question about how most of us introduce practices. If you arrive with something complete and well-argued, the best you can hope for is compliance. What would you have to leave deliberately unfinished to find out what your team actually knows about their own situation? ## Further Reading **Worth exploring** - *The Skills Matrix* as popularised in *Management 3.0* by Jurgen Appelo — a practical treatment of competency mapping as a team-owned artefact rather than a management instrument. - *The Skills Matrix *Miro template created by Cosima Laube: [https://miro.com/templates/skill-matrix-for-tech-teams/](https://miro.com/templates/skill-matrix-for-tech-teams/) - *Team Topologies* by Matthew Skelton and Manuel Pais — the platform-team-enabling-other-teams setup Cosima describes is exactly the territory this book maps, including the cognitive load problem behind "too much on my plate." - *Turn the Ship Around!* by L. David Marquet — on the mechanics of moving authority to where the information is, and why permission has to be made explicit before people will use it. - *The Fearless Organization* by Amy Edmondson — the research behind why the incoming architect's promise that the matrix wouldn't be used for ratings was the load-bearing part of the whole exercise.