Based on previous feature flagging systems we've b...
# ask-questions
b
Based on previous feature flagging systems we've built, we assumed we'd probably just evaluate all flags at the same time on our backend, providing everything from a single API method It seems like growthbook's SDKs are designed to evaluate more on the client-side, at the point a flag needs to be checked Is there a reason we shouldn't just evaluate them all at the same time?
f
We didn't want to require that GrowthBook be in the critical rendering path
so the SDKs do the feature evaluation locally, which means the features and rules can be cached
you can do the evaluation on the API side if you like
you can even mix and match - our variation assignments will be the same with any of our SDKs
b
Okay cool :) thanks! I wondered if it might somehow mess up the stats/reporting, but it sounds like as long as we track at the time we check the flag status, that's the main thing