I find that as an early-stage startup founder, I’m intensely focused on building my product and getting users, which leaves little time for writing tests. If I do write them, they’re usually unit tests, but I rarely write UI tests with Selenium because they’re so brittle as CSS selectors and UI layouts change frequently.
Have you guys had a similar experience? Would you write UI tests earlier if there was a way to use AI to automate them and make them more robust?
if you are writing unit tests you are moving too slow
Ideally, you have devs that are faster when writing tests. But you might not have the brand and money yet to attract great talent, and in that case, it’s fine to sacrifice quality for speed. You will hate yourself for this decision in a year or two when you will have to feature-freeze for a year to do a full rewrite - but you would hate yourself more if you didn’t even get to enough customers to that a rewrite becomes necessary :)
At some point it is faster to write automated tests than to manually test that nothing breaks. Usually that point is when you start scaling your team.
Everyone here is making it be that writing some tests is gonna take forever but I would say that any project with more than 2 employees is going to save time in the long run by spending a week writing some automatic testing.
speaking as a startup advisor working for a VC… focus on the problem that is the biggest barrier to making money and growing… thats typically not UI.
you’ll know when its UI, because your data will show thats the point at which the greatest drop off (leakage is) and the only option to optimise that point is with UI/X solutions.
but i assure you, unless your business is about great UI experience, ppl are not visiting and interacting with your platform for that. i’m bias, but whatsapp’s UI sucks compared with WeChat, and yet its essential these days.
Depends on what you want to test. I would suggest the following two though:
-
There is nothing wrong with using chatgpt to guide your tests, while asking from users to provide feedback. Remember that, ultimately, you want something that is visually appealing and user friendly to your users and not necessarily to you.
-
Adopt some Kanban and Lean principles. Prioritise those aspects that would generate the most value and work on them only when is needed.
-
once i get it into production i like to write at least a few e2e UI tests on a chron to make sure the entire thing is up and alert if not.
This worked well for me in the past, super simple to use : https://ghostinspector.com/