Ever since Bubble updated their pricing to the new Workload Unit (WU) system, there’s been quite a bit of confusion around how to ensure your app uses WUs efficiently. However, Bubble has finally released a guide on how to optimise your usage and minimise how much you get billed for.

Some of the information given in this guide is actually seriously helpful for optimization. For example, Bubble will only use WUs on conditionals if it actually runs the specific part of the conditional. What I mean by this is that Bubble will treat “and” conditionals in a specific way.

It will always perform the first condition, but if the first condition fails (i.e “is ‘no’”), it will not perform any of the secondary conditions, as there is no way for the condition to pass anyway.

This means that doing “Group is visible AND Do a search for:count >10” is usually more efficient than “Do a search for:count >10 AND Group is visible”, as the visibility check is considerably lighter than the search, if the visibility check fails, you won’t incur the cost of the search.

Another interesting thing to note is with Database Trigger events, even if they have conditionals attached to them (which would cause them to fail and not run) they will still use a tiny amount of WUs (in comparison to actually running) to “check” themselves. For instance, you have a trigger for “User is updated, but only trigger when before’s email is not after’s email”.

Even if the email addresses are the same, a small amount of WUs will still be used to perform the “trigger check”, even though nothing actually happened functionality wise.

I could spend ages listing out the interesting items in this guide, but I’d honestly advise giving it a look yourself, as you can read through items pertinent to your own optimisations. And don’t forget that the “logs” tab can really help you track down those big drains on your WU limits!

You can find the entire guide at this link:

https://manual.bubble.io/help-guides/maintaining-an-application/performance-and-scaling/optimizing-for-workload