Agile Testing book club: Everyone is a Tester

If you’ve even dipped your toe into the online testing community, there’s a good chance that you’ve heard Agile Testing by Janet Gregory and Lisa Crispin as a recommended read. A couple weeks ago I got my hands on a copy and thought it would be a useful exercise to record the highlights of what I learn along the way. There is a lot in here, and I can tell that what resonates will be different depending on where I am mentally at the time. What I highlight will no doubt be just one facet of each chapter, and a different one from what someone else might read.

So, my main highlight from Chapters 1 and 2:

everyone is a tester

Janet and Lisa immediately made an interesting distinction that I would never have thought of before: they don’t use “developer” to refer to the people writing the application code, because everybody on an agile team is contributing to the development. I really like emphasizing this. I’m currently in an environment where we have “developers” writing code and “QA” people doing tests, and even though we’re all working together in an agile way, I can see how those labels can create a divide where there should not be one.

Similarly surprising and refreshing was this:

Some folks who are new to agile perceive it as all about speed. The fact is, it’s all about quality—and if it’s not, we question whether it’s really an “agile” team. (page 16)

The first time I encountered Agile, it was positioned by managers as being all about speed. Project managers (as they were still called) positioned it as all about delivering something of value sooner than possible otherwise, which is still just emphasizing speed in a different way. If asked myself, I probably would have said it was about being agile (i.e., able to adapt to change) because that was the aspect of it that made it worth adopting compared to the environment we worked in before. Saying it’s all about quality? That was new to me, but it made sense immediately, and I love it. Delivering smaller bits sooner is what lets you adapt and change based on feedback, sure, but you do that so you end up with something that everyone is happier with. All of that is about quality.

So now, if everybody on the team should be counted as a developer, and everything about agile is about delivering quality, it makes perfect sense that main drive for everybody on the team should be delivering that quality. The next step is obvious: “Everyone on an agile team is a tester.” Everyone on the team is a developer and everyone on the team is a tester. That includes the customer, the business analysts, the product owners, everybody. Testing has to be everybody’s responsibility for agile-as-quality to work. Otherwise how do you judge the quality of what you’re making? (Yes, the customer might be the final judge of quality means to them, but they can’t be the only tester any more than a tester can be.)

Now, the trick is to take that understanding and help a team to internalize it.

What the Bug!? (An attempt at knowledge sharing, two ways)

When I was making the transition from waterfall style projects to agile teams in a previous company, one of the main things I struggled with was the loss of the testing team as we all became generic “software engineers”. We all still did the same testing tasks, but none of us were “testers” any more. There were a lot of positive effects from the change, but I kept feeling like without dedicated testers focused on improving our testing craft, we’d stagnate.

What I was missing, as I eventually realized, was a testing community. A recent episode of the AB Testing podcast talked all about building a community of testers, which brought all of this back to mind. A few of the ideas Alan and Brent talked about I had actually tried at the time. One in particular was the idea of highlighting major fails of the week in a newsletter, even offering prizes for the “winner”. Back in 2016 I heard a similar idea at a STARCanada talk, where the engineering group at AppFolio would send an email to everybody in the company for every bug they found describing what was learned, again emphasizing that finding these bugs was a positive thing, not a blame game. (Sorry I can’t find now who gave that talk; if it was you let me know!)

The reason the idea stuck with me at the time was primarily because I had started to notice that as our newly agile teams specialized in subsets of what was previously a monolithic product, we started to loose visibility on bugs that other teams ran into. Different teams were getting bit by the same issues. It didn’t help that the code base was also old enough that newer people would run into old bugs, spend potentially hours debugging it, and then hear “oh yeah, that’s a known problem.” (My response to that was to scream “It isn’t known if people don’t know it!” silently to myself.)

Here’s what I tried to do about it:

What The Bug!?
Honestly, the part of the idea I’m most proud of might be the name

I wanted teams to start talking more about the bugs they found, both so that others could learn from them and so that we could all tune our spidey senses to the sorts of issues that were actually important. There wasn’t much appetite for an email newsletter—people didn’t seem to read the newsletters the company already had—but we ended up trying two alternatives, one of which was pretty successful and one of which really wasn’t.

Building A knowledge base

The first idea was to solicit short and easily digestible details about every production bug that got logged into our bug tracker. Anybody who logged a bug would get an email asking them to answer three questions. The key was that the answers had to be short—think one tweet—and written at a level that any developer in the company would be able to understand. Bonus points for memorable descriptions. The questions were roughly:

  1. What was the bug?
  2. What was important about it?
  3. What one lesson can we learn from it?

The answers were linked back to the original ticket and tagged with a bunch of meta data like which team found it, the part of the system it was found in, what language the code was in, and any other keywords that would make it easy to find again. The idea was that if I was going to start writing something in a new language or working on a new part of the system, I could go look up the related tags and immediately find a list of easily digestible problems that I should stay alert for. I think it was an okay idea, but there were issues.

First problem: People were pretty bad at writing descriptions of bugs that were short, but also useful and interesting. It not only took a lot of creativity, but in order to do it well you also really had to examine what was important about the bug in the first place. The example I used as a bad answer to the second question was “This caused an error every Tuesday”; What caused what kind of error every WHY TUESDAY!? This was especially problematic for the third question, where often the answers that came back were “testing is important” or “we’ll test this next time”. True, but shallow. What I was really hoping for would have been “There are different kinds of Unicode characters, so always consider testing different character planes, byte lengths, and reserved code points”. To really make the knowledge base as useful as it could have been, it would have needed committed editors who would talk to the people involved and craft a really good summary with them.

Second problem: The response rate was pretty lousy. It might have been that targeting every production bug was just too much. Not everybody is going to see something interesting in every bug, and not everybody is as interested in learning from them. That was part of the culture that I wanted to change—I wanted everybody else to be as excited about these bugs as I was!—but it wasn’t going to happen over night.

Third problem: It might seem minor, but the timing of the email prompt coming the day after the bug was logged was often just too soon to have digested what the real lesson learned was. This turned up as a problem as I asked around about why people weren’t taking part. They just didn’t have the answers yet.

All of this created a chicken-and-egg problem. Until people saw the value of this project, saw interesting summaries, and got excited to contribute their own experiences, we wouldn’t get the content we needed to build that interest or excitement. And, in all honesty, though there was a conscious effort with this to make an accessible library of bugs compared to the technical JIRA tickets, we were still basically asking people to log a bug in one database every time they logged a bug in another database. We needed something more active and engaging.

Welcome to the What The Bug!? Show

At the time we already had a weekly all-hands meeting every Friday afternoon where anybody could contribute a segment to demo something, talk about something interesting, or anything else they wanted. I was doing short segments on quality and testing topics every few weeks to try to promote testing in general, but it was largely a one-man show. A fellow tester/developer that I was working with on the What The Bug!? knowledge base had the idea to take just a couple minutes each week to present our favourite bugs of the week.

Turning a passive library of bugs into a weekly road show was a big success. We were basically getting the benefit of a bug-of-the-week newsletter with the added bonus of an already established live audience. Again, the branding helped to sell it, and because the two of us both embraced the idea that any production bug could be turned into an 2 minute elevator pitch for something interesting to learn, we were able to make pretty fun presentations. We at least got laughs and had people thinking about testing if only for 5 minutes, but the real highlight for me was that afterwards I had people come up to me and say “you know, I found something last week that would make a good segment next time.”

This was possible in part because she and I had both spent that time soliciting user-friendly descriptions from people logging bugs, so we knew what had been going on across all the teams for the last couple weeks. We had a lot to choose from for the feature. It would have been harder to do without that knowledge base, being limited to only the bugs we knew from the teams we worked directly with. I suspect, though, that once the weekly What The Bug!? segments built up enough momentum that people took on presenting their own bugs, the need for the email prompts and the knowledge base would fade away. An archive of the featured bugs could still be a useful resource for new people coming on board, but it would no longer be the primary driver.

Where are things now

I left the company shortly after starting What The Bug!?, but I recently had a chance to check in with an old colleague and inquire about where these two manifestations of the idea ended up. Unsurprisingly in retrospect, the knowledge base has largely fizzled. The combination of asking people to volunteer extra writing work and the low ROI of a poorly written archive doomed it pretty early on. On the flip side, bugs are still a regular topic at those weekly meetings, although I’m sorry to report that they dropped the What The Bug!? branding. If anybody else wants to try something similar, feel free to use the name, all I ask for is a hat tip. (A quick google of the phrase only turns up some etymologists and an edible insect company, but if someone else had the same idea in the context of software, do let me know and I’ll pass the hat tip on.)