Playtesting: maintaining high product quality without QA

One aspect of ClassDojo’s engineering culture that tends to surprise people is that we don’t have dedicated QA engineers or testers. We use a variety of strategies to keep our end-to-end feature development fast, continuous, and high quality. These strategies include a strong culture of writing automated tests, along with regular playtesting sessions.

What do we do instead of QA?

At ClassDojo, we prioritize truly continuous deployment of our backend monolith, which means removing as many manual blocking steps as we can, and we’ve built systems to do so safely. Engineers have end-to-end ownership of features, and we rely heavily on writing automated unit and integration tests to ensure that we’re not pushing code that causes unexpected behavior or errors. Our canary containers help automatically roll back any commits that cause errors, and we monitor our logs carefully.

However, as thorough as we try to be, automated tests and monitoring can’t catch everything. To supplement our automated systems, we hold regular playtesting sessions to serve as manual testing as well as context sharing.

What is playtesting?

Playtesting is simply manually testing the products and features the team has built. You may have also heard of the term “dogfooding,” which refers to actually using your product as a real user. Some folks on the team are able to dogfood as legitimate parent users since their kids’ teachers use ClassDojo. However, since the majority of us are not teachers in a classroom nor parents connected to a real classroom, we do intentional playtesting sessions instead.

How do we run a playtesting session?

Who should join?

Individual teams might playtest new features right before shipping, in which case the whole team should join. We also schedule sessions that are open to anyone at the company so teams can share their recent features and solicit feedback from the wider group. For the best insight and context sharing, we encourage including folks from various functions, such as engineers, designers, PMs, marketers, and customer success agents. Playtesters don’t need to prepare anything ahead of time, or even have any knowledge of the flows. They just need to have the right type of device if it’s a mobile feature that’s only released on one platform.

The process

The first step is scheduling a time for the group of testers to get together and play through certain flows. Usually 30-45 minutes is sufficient, depending on how large or complex the feature is. Ahead of time, the product owner should prepare a set of loose guidelines for how to run through the flows, including any necessary setup steps such as installing an alpha build or turning on certain feature flags.

At the beginning of the session, the facilitator or product owner gives a brief overview of the flows and setup, then all the playtesters simply go through the flows independently on their own devices and accounts. Doing this synchronously instead of async lets us quickly identify whether an issue is widespread or a particular edge case, and answer any questions on the spot. We typically use an Asana board where playtesters can add cards for anything that comes up — not just bug reports but also general product feedback and points of confusion.

We reserve 5-10 minutes at the end of the session to go through the cards and make sure the issues are clear. From there, the product owner and team can prioritize them at their next prioritization meeting.

What are the benefits?

Our playtesting often finds bugs in obscure edge cases not covered by automated tests. With an app and user network as complex as ours, it’s nearly impossible to cover all use cases with test fixtures, so it helps to have a variety of people testing on their own real-world accounts.

Playtesting is one of our best methods for cross-team and cross-functional context sharing. Screenshots, product specs, and demos can only go so far in conveying what a new feature really involves. Having teammates actually get their hands on the features is a great way to share what’s being built. It’s also valuable to get fresh perspectives from folks who perhaps hadn’t explored that product area before. It’s like having an in-house focus group to give feedback.

If you have ideas on how we can improve our playtesting or manual testing strategies, reach out! We’d love to hear from you.