Get Testing out of your Tech Debt


Technical debt and testing have a long and entangled history. Across many organizations, teams struggle to define “technical debt” and what should fall into the “tech debt” bucket. Testing commonly suffers the fate of being categorized as tech debt, and consequently isn’t prioritized. Defining tech debt, and even rebranding it, can help your team to prioritize testing and reduce the negative stigma around tech debt.


Hi everyone, my name is Carly and today I'm going to talk to you about getting testing out of your tech deck. Okay, first, just a quick introduction. I'm a software engineer, I work at Galileo, which is a small health tech startup in New York City. Previously, I worked at Haven and Zocdoc, which are also health tech companies. Do a lot of JavaScript, TypeScript, React. I love automation, which is why I'm here at this testing conference today, really excited. And in my spare time, I like to get outdoors. Okay, so let's start off with talking about sort of the state of tech debt, what I've witnessed on teams, how people are thinking about tech debt so far. I often see a roadmap, something like this, that product managers and engineering managers will work really hard to put together, figure out exactly what the very top priorities are for this team. And then there's usually some like 15-ish percent capacity that's dedicated to tech debt. And that's what I'm going to focus on today, that 15% capacity. Let's talk about what falls into there. I oftentimes see like scaling will fall into this 15% capacity, testing might fall into there, upgrades and maintenance will be categorized as tech debt, bugs will be categorized as tech debt. And it sort of starts to feel like it's the junk drawer of software engineering, where anything that isn't like explicitly put on the team's roadmap is kind of dropped into this tech debt 15% capacity drawer. And the team tries to get it done within that, within those constraints. The reality is, though, I just don't think you're going to fit all of that stuff into just 15% of your team's capacity. There's actually, you know, a lot of work to do here. I've been on teams who have tried, I mean, I've tried before to fit all that in. And the result I see is that it really slows down your team. Your team starts to suffer because the tech debt is just growing and is never really being fully addressed. So it's, of course, bad for your business when your team can't ship features quickly. It also has these second order effects, I think, of like developer happiness starts to suffer because people don't usually want to work on teams that are saddled with a ton of tech debt. So you can have problems with morale and retention. You can either even like have problems with maybe hiring because people often don't want to join teams with a ton of tech debt. Okay, so it's obviously not great to be in that place where you're just mounting with a ton of tech debt. But how do you get out of there? I think a solution is to apply some more structure and rigor around how your team decides what qualifies as tech debt. And what I've seen work well is this one philosophy that I've come up with, or, you know, that I've learned from others, which is that tech debt is issues that primarily impact the developer rather than the user. So that means things that are going to make the developer more productive, make the developer more efficient, rather than are directly sort of to improve the user experience. And the kind of like less obvious follow on to that is that I think feature preservation is in the scope of building features and should not be categorized as tech debt. And when I say feature preservation, I'm talking about things like automated testing and monitoring, things that make this whole system aware that a feature exists, and prevents, you know, a feature from disappearing or breaking without the developer team knowing about it. The reason I categorize this as not tech debt is because if these features are to break, the real impact falls on the user. Like it really is the user that suffers if a feature breaks. And so I think it's very much in the user's incentive to have feature preservation built in from the start. So let me just double click on this philosophy. When you get a ticket or a project from a product manager, you usually have some very explicit scope built in, which is to make that new thing work. And then there's kind of some implicit scope hidden underneath, which is to not break the old things. And if you were to break the old things, the person that's going to suffer, again, is the user rather than the developer. And so for that reason, I think it's really important to have product manager buy in that things like automated testing and monitoring need to be in scope of the feature, because the users will suffer if you don't have that. Okay, cool. So now we have that philosophy. Let's try and pressure test it against some more real life situations. And as we're going through these, I want to again focus on who is benefiting from solving this problem. First one, let's say you have a ticket to fix a bug, like fix this broken tooltip. This one I categorize as for the user, and so it's not tech debt. It's probably not the developer who's suffering from this broken tooltip. It's probably the user. And so it should be prioritized against all the other user concerns. How about a code cleanup ticket, where we are going to refactor some huge authentication file into multiple modules. This one I'd say is a great example of tech debt. This is for the developer. This is probably in order to help developers move faster, understand things more quickly. And so this is a great use of that 15% capacity. Testing, like writing tests for a new model. This is feature preservation, and so it's not tech debt, and it really needs to be baked into the estimation of the original creation of the feature. And scaling, like making a user service work for 10,000 people. Again, this is actually for the user, and so I'd say it's not tech debt. Despite the fact that this feels like a very technical project, doing this work isn't going to make the developer more productive or more efficient. It's really to serve your users better. And so I'd say this isn't tech debt. This, again, should be prioritized against other user concerns. Okay, so I recognize that I've sort of portrayed this scenario as very black and white, but it's not going to be, right? There's going to be lots of things that are maybe good for the user and good for developer productivity. And so it'll take judgment and nuance to apply this philosophy, for sure. But as you're thinking about it, I think just concentrate on the two ends of this spectrum, like tech debt being primarily for developer productivity, not tech debt being anything that's really for the user's benefit. And hopefully, with this structure, you can start to climb your way out of the tech debt hole and use that 15% capacity more effectively. Okay, thank you. That's my talk. Just a quick mention that Galileo is hiring, so definitely reach out to me if you're interested in that. And you can find me on Twitter at Carly Jo. Thanks.
8 min
15 Jun, 2021

Check out more articles and videos

We constantly think of articles and videos that might spark Git people interest / skill us up or help building a stellar career

Workshops on related topic