When I force a version. Either via ?forcevariation= in the URL or via match in a UTM via the rules on the feature in growthbook. Should I expect the trackinig event to send again and with the new version the user is put in via one of the forced rules?

I have a question about SDK connection in GB panel. I have outdated JavaScript SDK 1.1 on website and right now we rise version to 1.6.1 directly on our website. But how this change is corelated with settings in GB panel? What will happen if Ieft SDK version here with

1.1

or is a risk that something stop working properly in our A/B tests after I change this to

1.6.1

? Should I have an attention for something after this change to verify if everything work properly?

qq: in the enterprise license is there a way to keep users from seeing the SDK keys? I've tried to play with the user roles but even when I create a role that only should have access to the ideas bit, the users continue to see the SDK keys

Hi! I have a question regarding this part of documentation

We have many environments (prod, stage, about 50 dev boxes) that need to be completely independent. This means we need to be able to deploy GrowthBook separately for each environment (and for dev boxes — about 50 times in total). We need to understand how this would affect the cost, if at all.

Thank you in advance

Hi team, to confirm, can we verify the winner experiment directly with GrowthBook dashboards and no need to add something like Google Analytics? I'm using Unleash and we need to make this connection to define the winner experiment. I'm sorry if I'm not on the right channel for asking this question.

Hello! We're relatively new to GrowthBook and keep getting timeout errors for several metrics. I've navigated to "Edit Data Source" and set the

Request Timeout

field to

0

and the problem persists: every time I update metrics for an experiment several metrics fail with the error

Timeout: Request failed to complete in 120000ms

. What can I do about it?

when we create a new sdk connection, it asks for language and the version, does the API payload differ in some cases? I wanted to have 1 connection across all our clients, is that something I need to address?

Hello, I inadvertently fumbled something when entering variation IDs in my experiment which is now providing a warning that that database has returned 4 variation ids. I already verified the correct IDs and started a new experiment phase, but the message persists. Is there any way to dismiss this message, or is there something else I need to do to resolve this?

afternoon folks. I am facing an error when trying to create metrics of any kind using fact_tables: This started to happen now. A while ago everything was running smoothly. Important to note that I can query data normally using SQL Explorer, and when I click the "test query" button it correctly displays the data in the table

Hi team, I have a question about metric scope in experiments. I’ve created a metric in GrowthBook (e.g., for a

btn_click

event in BigQuery). When I use it inside an experiment, the results only include users who have seen the experiment. However, I’d like to see all occurrences of the

btn_click

event across all users, even those who weren’t exposed to the experiment. Is there a way to configure a metric or query so that it includes all event data, not just experiment-exposed users? Thanks!

Is there a way to see what the actual failing query was towards mongodb somehow? I see errors because we use azure comosdb for mongo and they indicate I "just" need to create an index for order by's . thing is that I can't find what table, what field 🙂

Hi Team, We’re noticing some unusual activity on our account, even after removing all SDK connections and confirming that none of our services use this account anymore, we’re still seeing around 150k–200k CDN requests per day. Could anyone please help us identify and block or exclude these old requests so that they no longer count toward our usage quota? Thanks in advance for your help!

Hello, We’ve encountered a serious issue that seems to be connected to your recent implementation changes. After investigating, I noticed that the response format for /api/eval/apiKey has been completely changed. As a result, our app can no longer parse the experiment data, which has broken all our experiments. This issue is critical and quite costly for us. Could you please revert to the previous response model and release a fix as soon as possible? Thank you in advance for your quick assistance.

Hello, I would like to know why I am finding differences in metrics compared to other platforms, specifically GA4. Metrics on this platform show up to twice as much as on Growthbook. What could be causing this? In one specific experiment, I am using “purchase” and “ARPU” as goal metrics. (in GA4 there ismore than 6K users)

image.png

I'm trying to speed up my experiment assignment query by using the

{{experimentId}}

sql template. When I save it I get the following error though. I'm assuming because the experimentId is empty when it tests the query?

I'm setting a Java SDK for GrowthBook, but not able to proceed further in the setup steps as it shows not connected here. Any idea why this is happening?

Hey community. I cannot make remote evaluation work. I've deployed Growthbook Proxy. Made everything work (stream updates, client-side evaluation). However, I cannot enable "Remote Evaluated" in the SDK Connection settings. Seems like the

PUT <http://localhost:3100/sdk-connections/sdk_4xn1tmhe5p13z>

request body still contains

remoteEvalEnabled: false

. Therefore, I am getting 400 error "Failed to get features (remote eval required)" in the response from

POST <http://localhost:3300/api/eval/sdk-NrZe4lDsZmzA3Q6M>

(my proxy). Growthbook and Growthbook Proxy are on the latest tag (4.1.0+). Is it a known bug? Did not find anything on Github or so. I would appreciate your help and suggestions 🙂

I am using the REST API to update experiments, in particular to add new phases when the number of variations change. For example, I start an experiment with 4 variations, and after some time I see that two variations are clear losers, and so want to start a new phase with the other two. First, is this an ok practice, and if not, what are the alternatives? If this is an ok practice, how should I structure the call to update the experiment? I have tried updating with a new variations array that only includes the remaining variations, and adding to the phases array with an object that has a trafficSplit that only includes the two remaining variations. This seemed to work, however after some time, I got a warning in the Growthbook admin that it was receiving data that still included the original 4 variation IDs instead of just the remaining two.

Hey all, I'm interested in testing out the Vercel Integration, however the only way to get it set up is to create a brand new account through Vercel's dashboard, rather than connecting it with my existing account. This seems like either an oversight or a complete inability on my part to figure it out 😅

Hello Everyone, All of sudden our CDN requests increased to millions, without any deployment? What could be possible reason? We have checked, there is no multiple initialisations Our current implementation is we are using growthbook as provider to openfeature. BackgroundSync : true This spike started on 27 Oct, and from IOS only, Investigating but didn't find anything.

Hi everyone! I have some problems with my subscription, is there a way that someone could help me?

@future-teacher-7046 @better-magician-65629 @proud-baker-47556 Hello Everyone, All of sudden our CDN requests increased to millions, without any deployment? What could be possible reason? We have checked, there is no multiple initialisations Our current implementation is we are using growthbook as provider to openfeature. BackgroundSync : true This spike started on 27 Oct, and from IOS only, Investigating but didn't find anything. Its urgent as billing is increasing

Hello all, I am curious as to why the displaying of results is tied to the number of conversions, rather than the number of total samples (exposures) in the experiment? And I wonder if it's possible to modify this somehow, so that results are not shown below some specified threshold of exposures irregardless of the number of conversions?

Hey all, We currently using custom S3 endpoint to store images, and apparently GrowthBook doesn't support it, so we basically can't upload any images in GrowthBook. I found an issue created some time ago: https://github.com/growthbook/growthbook/issues/2655, which also mentioning end of support (September 8, 2025) for AWS SDK, but there are no progress there. Maybe someone have plans, solutions, workarounds or something to make it possible?

Hey, Is it possible to assign a specific user to a particular experiment and variation using an override rule? We’re using CDN caching, so there’s a possibility that a user (say, U1) might see a cached experience. We want to ensure that when the same user later views a non-cached page, the GrowthBook SDK consistently assigns them to the same experiment and variation as before. Thanks!

Hey, anyone here integrating GB with Amplitude? I dont pass converted currency in my Amplitude events but rely on their multicurrency derived properties features. Unfortunately they don't offer a way to export this to my warehouse but just wondering how people get past this issue without passing a converted value myself

hey growthbook team what is the possibility of a forced value not working on a feature-flag. we recently noticed some of the users present in saved groups were not evaluated as per the saved group but default value was getting assigned to them?

I’m running a global Visual Editor test that’s tracking properly in GA4. However, another global test using Feature Flags is only showing 7,200 users instead of the expected 24,000 exposures. Does anyone know what might be causing this discrepancy?

Hey GrowthBook team! I have a question about the evalFeature method on the JS SDK: https://github.com/growthbook/growthbook/tree/main/packages/sdk-js#evalfeature • Not mentioned in the docs but it seems to return an

on

and an

off

property with a boolean value. What is the condition when

on

is returned as true?