One of the key needs in lean, scrum, and other agile processes is for continuous improvement. We constantly review how we do things to do them better. The most common method of doing this is the retrospective. After 5 months of writing the Baby Steps to SOA series, I decided that I wanted to review what I had done and figure out how to do it better. Of course, since this is an agile-related blog, I wanted to share this experience with those of you out there so you can learn about how a retrospective works, and how you can apply it to any work you are doing.
Choosing a retrospective model
The purpose of a retrospective is to look back at a period of time (the sprint, a project, the year) and review your challenges and how you think you should overcome them. If you aren’t familiar with agile retrospectives and are looking to start, I highly recommend browsing the Agile Retrospective Resource Wiki to find different suggestions and tools you can use for reviewing the work you’ve done. There is also a great article that was done recently on How to Run an Effective Retrospective which is worth a read for those of you looking to improve your existing retrospectives.
At first, I ran a lot of Start/Stop/Continue retrospectives. The Start/Stop model had the benefit of making people think differently about new versus existing things. However, over time, the lines became blurred. Most of the starts usually had an associated stop, and organizing our actions was a little difficult. More recently, I have enjoyed using the Plus/Delta model. This focuses more on what is going well, and what needs to change. Essentially, this model merges the Start/Stop together into Delta, and the Plus category covers the Continue items.
For this blog retrospective, we will be using the Plus/Delta model to review what worked, and what needs to change.
One of the keys to learning from your actions in Lean’s Build-Measure-Learn (BML) cycle is obviously the measuring component. For any work you are doing, there is likely some form of measurement that is being done. Story points completed, bugs fixed, dishes cleaned, hours worked, how late the project is, number of views to your blog, etc. When doing a retrospective, you should make sure to bring these measurements in and review them so you can compare against where you were before and where you are now. Of prime importance with measurements is discussing the why of the metric. Numbers might go in a negative direction, but there could be a very reasonable reason why. For example, the number of story points completed in an iteration might go down, but it could be because the client wanted the team to focus on defects, infrastructure, and training prior to a launch. The low number of story points isn’t a bad thing, it’s just reflective of the focus of the iteration.
While writing the Baby Steps to SOA, I’ve been watching the stats provided by WordPress and reviewing how users are getting to my blog posts on SOA, versus how they read other topics on the site. Here’s the relative ranking at the time of writing for the posts:
- Step One – Analyze and Plan (105)
- Step Three – Three Tiers for the Website (79)
- Introduction (56)
- Step Five – The Move to a CMS (44)
- Step Two – Measure It (42)
- Step Four – Single Sign-On (39)
- Step Six – Data Services (25)
- Step Seven – Centralizing eCommerce (17)
- Step Eight – Sharing the Business Tier (15)
- Step Nine – Moving beyond the website (9)
- Step Ten – Riding the ESB (3)
The Why: In general the older the post, the more views it has. That being said, a post that is even newer than Step 10 has already doubled the Step 10 post in views. Also, Step Three and Five have jumped up the list because of their focus on popular topics. Reviewing search keywords and referrers shows that most of the traffic is coming through search engines, and not users flowing from one post to another. This confirms the suspicion that popular topics will garner more views, not a new learning. What was interesting was that after 10 years of discussions on the topic of SOA, there were still users using the SOA keyword to find content, however, not many. SOA is not a hot topic these days, and therefore doesn’t drive a lot of views on that keyword alone.
Now it’s time to start reviewing the Pluses and Deltas of the series…
- Learned a lot about different elements of the SOA process, and what others are writing out there.
- Experience gained in writing lengthy series (longest series I’ve ever written). I imagine that some of the problems I experienced would be very similar to authors who are writing a book, trying to get chapters completed, and going through a review cycle.
- Provided a continuous subject for the blog over a long period of time (5 months). Every few weeks, I knew what I was going to write about.
- The outline set out at the beginning provided structure as to what would come when.
- Regular cadence provided a dependable source of the topic. Although the cadence was shifted to 3 weeks after an initial 2 week cadence, by trying to stick with the cadence I knew that any readers following the series would be able to know when the next step was to be published.
- Got the chance to create something on a subject I am passionate about.
- Repetitive intro tag deters readers. Because the introduction to each blog post was basically reviewing what had already happened, many of the posts had the same teaser content. Given that search engine traffic was the primary source, this made a lot of the posts look very similar (other than the title). Near the end of the series, I started changing out the intro to be more focused on the content of the blog, and less on the previous steps taken.
- The Evolutionary Roadmap ate up a lot of real estate. While the roadmap allowed users an easy navigation tool to jump from one step to another, it meant that the first page was essentially an introductory paragraph followed by a long list of links. Given that very little traffic has been brought in by users navigating on that roadmap, I would probably avoid it in the future.
- While interesting to myself, using a topic that has been out of circulation in the hype machine for almost a decade does make it difficult to spread the information to users. By changing the keywords to be less SOA focused and more focused on the elements of each step and the eventual goal of a distributed system would likely have been a better idea.
- The length of the series seemed to be a bit too long. While I did want to focus on the entire iterative process from spaghetti-code to Enterprise Service Bus, I doubt most readers would desire to consume a 10-12 step series. Restructuring as a WIKI, Reference document, or as a small book may have been a more appropriate medium.
- No comments. I did not engage the readers enough to get them to want to interact or ask questions. For a resource-oriented post, that is not as big an issue, but it does raise the question as to whether the content was helpful or not for anyone.
I am going to consider writing another series in the future, but for now, I think my biggest lesson learned is that I want to keep things contained. Any upcoming series will probably be only 2 or 3 blog posts in length, and I will probably focus a little more on ALM in the upcoming months, and less on architecture.
Additionally, I want to learn how to write better to engage the audience. I would like to get to know my readers, and find ways to help them. I have always enjoyed it more when social interaction is involved then simply sending out data into the stream.
I also want to find a way to keep the idea of the cadence in topics present. With the series done, what will take it’s place in the cadence? My initial thoughts are to go back to a 3-week cadence and try and replace the SOA spot in the schedule with an ALM focus. The other two weeks would focus on Agile and Sitecore, respectively. This would allow me to keep the idea of having a schedule for topics, and also make sure that I am providing content for three of my key interests.
I hope you all had as much fun reading the series, and the retrospective, as I did writing them. Let me know if you have any Plus or Delta to add to the list above!