https://www.growthbook.io/
Join Slack
Hi, I have some more questions - How can I achieve to know separate results by device id ? For examp...
# announcements
b
Hi, I have some more questions ā€¢ 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 ā€¢ 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
šŸ‘€ 1
w
ā€¢ 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
ā€¢ 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
Unlike most other A/B testing platforms all your data stays in your own data warehouse.
So yes you can get data into your data warehouse any way you like and just use growthbook for evaluating the results of the experiments
šŸ‘ 1
b
@white-fireman-22476 Got you. I thought when I split users by user id, it would ignore device id attribute.
@white-fireman-22476 Thank you !
w
@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 think what you would need to do here is to have two experiment rules. One where the targetting conditions would by to do it only for device_type=mobile. The desktop ones won't apply there so then put another experiment rule with targeting conditions be for device_type=desktop.
šŸ‘ 1
I believe you can give them both the same experiment tracking-key
b
I didn't think about that. I think that works.
w
And then see the results in the same experiment results page.
b
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
w
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.
By the way you would use the method above if you want the name of the variation different for mobile and desktop.
But if you are keeping it the same, you can use dimensions
b
Got you, I will look into this
w
Which is perhaps what you actually meant with your question.
b
Yeah, I think it is I just don't know how to say it haha
w
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.
šŸ‘Œ 1
b
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 ?
w
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.
b
Sorry, my bad. I mean does GrowthBook always require some document DB to be running ?
w
Yes
b
Thank you šŸ‘ šŸ‘
w
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.
b
Got you
78 Views
Open in Slack
PreviousNext
Powered by

GrowthBook Users

Open source platform for stress free deployments, measured impact, and smarter decisions.

Powered by