Hi, I have some more questions • How can I achieve to know separate results by device id ? For example, In the same experiment, I want to have separate result of desktop device and have separate result of mobile device • Can I use my own service/tool to put data into data source of GrowthBook's metrics or Am I required to use some GrowthBook tool ? For example, I want to create my own tool to save purchase record into the metrics data source

• How can I achieve to know separate results by device id ? For example, In the same experiment, I want to have separate result of desktop device and have separate result of mobile device

You would need to pass an attribute for device type when you make the growthbook instance: https://docs.growthbook.io/features/targeting#targeting-attributes

@white-fireman-22476 Got you. I thought when I split users by user id, it would ignore device id attribute.

@white-fireman-22476 Got you. I thought when I split users by user id, it would ignore device id attribute.

ah I think I misunderstood you here.

I didn't think about that. I think that works.

And then see the results in the same experiment results page.

I am just wondering. Apart from your method above, does GrowthBook have a way to make sure that when split user by user id 50/50, in each 50, it will eventually devide into mobile/desktop 50/50

Growthbook hashes by the feature-id and the split-id randomly. So both mobile and desktop should end up around 50/50 given enough users.

Got you, I will look into this

Which is perhaps what you actually meant with your question.

Yeah, I think it is I just don't know how to say it haha

Looking back at your question, I think it was clear enough. I was just momentarily read it wrong to be that you wanted something more complicated than you actually need.

Actually, I have one more question about self-hosting In doc, it gives docker-compose containing

mongo

and doc says we should not use it on prod Does it mean that on prod, we are required to replace that

mongo

with some real document DB and, otherwise, GrowthBook will not work ?

It can work ... its just not recommended to use a mongo docker image running on a single machine as it might not be so robust.

Sorry, my bad. I mean does GrowthBook always require some document DB to be running ?

Yes

Thank you 👍 👍

It is where the feature flag, experiment definitions, growthbook users, etc get stored. If the mongo container goes down or the machine dies and you do not have a backup of the data directory, then you might lose those things.

Got you