Team Topologies Book Review

I recently listened to the audiobook version of Matt Skelton and Manuel Pais’ book Team Topologies. It was so good I bought the kindle version while I was still halfway so I could make notes and highlight the good bits (most of the book it turns out).

Team Topologies

The book takes several principals such as Conway’s Law and really applies them to business teams. This is something I’ve seen first hand. When several teams work on one large product the codebase becomes decoupled if the teams are given ownership of particular components. However, if teams are expected to work across the codebase the solution becomes monolith and the teams become a super squad.

In the book the authors argue that there are actually a very limited number of team types in a modern organisation. I don’t want to list them because I’d strongly recommend you to buy the book and read the descriptions for yourself. However, if the various types it was the concept of platform team which intrigued me the most.

I actually think Matt and Manuel underplay the huge value of a platform team. They discuss brilliant ideas about consumable APIs and documentation for product teams which consume them. However, a data driven business like mine I believe we should run far more platform teams and far fewer product teams. If we want our Product Owners to be able to innovate and prove the value of ideas quickly we need our data sources and components to be as plug and play as possible. This allows any product or concept to be built and tested very quickly. If all these services were owned and managed by platform teams, instead of falling down the gaps between product teams the solutions would be more robust and the lead times far lower.

If you’ve never given any thought to how teams are created and assigned areas of ownership then this is a brilliant book to read. If you’re not sure how your teams communicate and share information then this book is essential.

Is Velocity a Vanity Metric?

A few weeks ago I went to a talk at Agile Yorkshire by Tom Hoyland where he discussed the how he formed an agile team. In it he claimed that Velocity is a Vanity Metric.

If you haven’t read it then I highly recommend you pick up a copy of The Lean Startup by Eric Ries, this was my main introduction to the term. In the book Eric talks about how a Vanity Metric is any figure which is skewed to inflate the success of a product. One of the example he gives is Number of Registered Users, a figure which will clearly go up and up over time. A better metric, Ries argues, would be Number of Active Users. Obviously if you’re looking for a successful product you’re more interested in how many users are currently using it than how many people completed the signup process.

He claims that by basing metrics and KPIs on these Vanity Metrics is akin to giving yourself a pat on the back for building a successful product while sticking your head in the sand about why your business was losing money (or the startup running out of runway as he describes it in his book).

photo of woman looking at the mirror
Beware the Vanity metric! Photo by bruce mars on

So, to return to Tom’s question – is Velocity a vanity metric?

Most Scrum Masters calculate a teams’ velocity by taking an average over the previous sprints. The number of sprints varies but six (around 12 weeks of work) is the norm. Using this figure the the team can then calculate how far a particular story is away from delivery or epic from completion.

person typing on laptop
Photo by on

There’s no doubt that Velocity is a useful forecasting tool (although, as one of my colleagues pointed out recently that if you plan a sprint on your average velocity then you will, by definition, fail 50% of the time). However, is it deluding us into thinking we’re a successful team and detracting us from more accurate measurement?

I think this comes down to how you measure success. Most people would agree that judging a development team on how many features they can deliver is fairly narrow minded. A development team includes a Product Owner, their remit should be to develop a product not to simply crunch features. If that was their goal then they could simply create huge numbers of easy, yet useless features just to score points.

In this new age of DevOps a Development Team should use their PO’s expertise and take ownership of the success of their product. For them to be judged on how much work they produce, rather than how much success they’ve had is a limited measure. It reminds me of the from the book The Goal where they measured and optimised the workstations rather than asking whether the system was effective.

Therefore, although shocked when I heard it I agree with Tom. Velocity is a useful tool for the team to forecast. But if it’s chased as a KPI or used to measure the team then it is indeed a vanity metric and will distract the team from trying to improve their product. If we want to measure our teams’ success then we should look at the metrics of our products, not try to calculate some kind of Hours to Story Point ratio or chase an ever increasing Velocity. We should focus on Number of Active Users or Number of Requests via the API, this will measure the teams’ success, rather than it’s productivity. As a friend of mine is fond of saying, effective rather than efficient.