https://www.growthbook.io/ logo
Join Slack
Powered by
# ask-questions
  • f

    fierce-tent-11841

    07/27/2025, 5:09 PM
    Hello, is it possible to have multiple growthbook proxy instances? We're in a situation where we self-host 1 growthbook that is shared with both staging and production, but we have separate proxy instance for each staging and production because staging and prod services live in separate VPC. But I notice
    PROXY_HOST_INTERNAL/PROXY_HOST_PUBLIC
    only support one proxy address.
    g
    • 2
    • 2
  • b

    bitter-airport-72278

    07/28/2025, 8:08 AM
    Hi folks, my organization is currently evaluating feature flagging and experiment analysis tools, and GrowthBook is one of the candidates. We're doing some preliminary cost estimates - who would be the best person to reach out to for information on enterprise pricing? I understand you offer seat-based pricing, but couldn't find concrete numbers on the website. Thanks in advance!
    s
    • 2
    • 1
  • m

    melodic-megabyte-84210

    07/28/2025, 9:24 AM
    Subject: Compliance question.. currently the cookie gbuuid is set before cookie consent, if the cookie is set after consent, then the experiment won't trigger for us, has anyone experienced something like this or have a workaround on this?
    s
    • 2
    • 1
  • b

    brash-football-99337

    07/28/2025, 12:24 PM
    Hi! We have started working with Growthbook recently and added to some of our products. We are running into some issues around the measurement protocol. Our workaround conflicts with channel attribution (leading to an increase in unassigned). We are a bit stuck regarding next steps so any help is appreciated
    👀 1
    s
    • 2
    • 1
  • p

    polite-hairdresser-26298

    07/29/2025, 8:23 AM
    A question on multi exposure and user_id. Can some implementation lead to Growthbook triggering multi-exposure in case visitor logs in and Growthbook realizes it is the same user later triggering multi-exposure in such cases. We did not have multi-exposure before but started getting it some time ago after making some changes in the code.
    s
    • 2
    • 5
  • b

    billions-house-96196

    07/29/2025, 10:47 PM
    Is there any timezone syntax in the date {{Handlebars}} helper? I don't know how you would do it but something like
    {{date startDateISO "yyyyMMdd" "America/Denver"}}
    Context: data source query I'm just now realizing that with GA4, the table_suffix used in the where clause by the query is in local timezone.
  • g

    gentle-house-39268

    07/30/2025, 1:07 PM
    Hi We have started to explore growthbook network split based on attributes but we are getting experiment values outside the rules and condition defined in the experiment. Greatly appreciate if someone can guide with steps to debug. Sharing the experiment config below We are getting variation value as 0 for hubIds outside 5 (outside the condition defined)
    s
    • 2
    • 3
  • n

    nutritious-garden-91325

    07/30/2025, 6:49 PM
    Hi all, me and my team are trying to setup sticky bucketing using the Java/Kotlin SDK but are having a hard time doing it. Can anyone assist? This is our current implementation (we are testing it locally trying to observe what happens to the same user id when using stickybucketing vs. when not).
    Copy code
    val trackingCallback = object : TrackingCallback {
        override fun <ValueType : Any?> onTrack(
            experiment: Experiment<ValueType?>,
            experimentResult: ExperimentResult<ValueType?>,
        ) {
            println("[TrackingCallback] Experiment is: $experiment")
            println("[TrackingCallback] ExperimentResult is: ${experimentResult.toJson()}")
        }
    }
    
    val stickyBucketService = object : StickyBucketService {
        override fun getAssignments(
            attributeName: String?,
            attributeValue: String?,
        ): StickyAssignmentsDocument? {
            println("[StickyBucketService] Getting assignments")
            return StickyAssignmentsDocument(
                "id",
                "b0c7f2f4-7575-4c31-88b2-823d2c88484f",
                mapOf("test__0" to "0"),
            )
        }
    
        override fun saveAssignments(doc: StickyAssignmentsDocument?) {
            println("[StickyBucketService] Saving assignments $doc")
        }
    
        override fun getAllAssignments(attributes: Map<String?, String?>?): Map<String?, StickyAssignmentsDocument?>? {
            println("[StickyBucketService] Getting all assignments")
            return null
        }
    }
    
    fun main() {
        val featuresRepository = GBFeaturesRepository
            .builder()
            .apiHost("MY_HOST")
            .clientKey("MY_CLIENT_KEY")
            .refreshStrategy(FeatureRefreshStrategy.STALE_WHILE_REVALIDATE)
            .build()
            .apply {
                initialize()
            }
        val ctx = GBContext
            .builder()
            .featuresJson(featuresRepository.featuresJson)
            .attributesJson(
                """
                    {
                        "id": "b0c7f2f4-7575-4c31-88b2-823d2c88484f"
                    }
                """.trimIndent(),
            )
            .trackingCallback(trackingCallback)
            .enabled(true)
            .isQaMode(false)
            .stickyBucketService(stickyBucketService)
            .build()
        val gb = GrowthBook(ctx)
        val result = gb.getFeatureValue("test", "error")
        println("Result is: $result")
    }
    looking 1
    s
    f
    a
    • 4
    • 9
  • n

    nutritious-garden-91325

    07/30/2025, 6:50 PM
    This is the output we are receiving. You can see that the stickyBucketUsed is returning false:
    Copy code
    [TrackingCallback] ExperimentResult is: {"value":"ON","variationId":1,"inExperiment":true,"hashAttribute":"id","hashValue":"b0c7f2f4-7575-4c31-88b2-823d2c88484f","featureId":"test","hashUsed":true,"key":"1","name":"Variation 1","bucket":0.993,"stickyBucketUsed":false}
  • a

    adorable-balloon-99556

    07/30/2025, 10:22 PM
    Hello, everyone! I am having a hard time trying to get my Kotlin backend work properly. I am following instructions from the "SDK Connections" page at GrowthBook as well as from the Kotlin SDK README. I have tried a bunch of different approaches but I can't get the SDK working. I tried
    GBNetworkDispatcherKtor
    rather than
    GBNetworkDispatcherOkHttp
    , I tried wrap the code within
    runBlocking
    , tried to call
    refreshCache
    manually, but the features can't get loaded. Here's the dependencies:
    Copy code
    // build.gradle.kts
    dependencies {
      implementation("io.growthbook.sdk:GrowthBook:5.0.1")
      implementation("io.growthbook.sdk:NetworkDispatcherOkHttp:1.0.2")
    }
    The program can't load the features although I can fetch them properly through the host url. Also, the status is "Connected" in the SDK Connections page.
    gb.getFeatures()
    is always empty. Could anybody please help? The code is in the snippet below
    GrowthBook.kt.cpp
  • m

    most-beard-6772

    07/31/2025, 1:12 PM
    Hi Guys! Need your help. I started running A/B testing experiment last July 31. When I clicked the update button - No data appeared. But when I checked DB, it appeared. How can I fix it?
    s
    • 2
    • 3
  • e

    elegant-solstice-68604

    07/31/2025, 8:41 PM
    Quick question about Filters on Growthbook fact tables. The docs on filters read:
    If a metric uses this Filter, GrowthBook will add it to the WHERE clause automatically. If there are multiple Filters, they will be ANDed together.
    However, inspecting the generated SQL query reveals filters are inserted as a column in the SELECT statement as
    SELECT CASE WHEN [[filter]] ELSE NULL
    . I believe these are functionally different with respect to certain table operations involving indexes. Curious if anyone knows why this design decision was made and/or if the documentation needs to be updated to describe this discrepancy?
    👀 1
    s
    • 2
    • 2
  • m

    magnificent-furniture-52657

    08/01/2025, 10:00 AM
    Hello! Posting here for the first time! Has anyone ever started receiving default feature flag values from experiments after they changed the variants weights to 0%? I had 4 variants 25% each and i reduced 2 of the variants to 0% and left the other 2 variants to run 50% each.
    f
    • 2
    • 4
  • g

    gifted-eye-87311

    08/01/2025, 12:19 PM
    Hey team, we are using a snowflake as our data source. Are there any plans to support key/value auth, as we are aiming to deprecate user/password.
    f
    • 2
    • 2
  • p

    plain-cat-20969

    08/02/2025, 8:50 AM
    Hi all! I'm new to Growthbook and running into what is hopefully a very easy issue to solve-- I set up a ratio metric that shows accurate numerator and denominator in the metric analysis, but is off when I use it in an experiment. Any advice?
    f
    • 2
    • 3
  • s

    straight-honey-15210

    08/04/2025, 12:57 AM
    Good morning, I cannot find a way to schedule the automatic ending date and time for a test. But if anyone knows a way, please let me know.
    f
    • 2
    • 2
  • t

    thousands-activity-913

    08/05/2025, 7:28 AM
    Hey folks! We've implemented GrowthBook as suggested in the respective guidelines (i.e., it is COMPLETELY self-hosted). Please advise how exactly should I declare Data Safety statement for Google Play Market to properly pass the review and get the app released
  • t

    thousands-activity-913

    08/05/2025, 7:30 AM
    @fresh-football-47124 can you help please? It is really important for us to properly pass the review
    f
    • 2
    • 15
  • b

    busy-psychiatrist-5181

    08/05/2025, 10:06 AM
    Hi! I’d like to change the email address associated with my GrowthBook account. Is that something I can do myself, or should I share the current and new email here so someone from the team can help? Thanks in advance!
    s
    • 2
    • 5
  • a

    acceptable-king-82846

    08/05/2025, 4:33 PM
    hi all! I’m attempting to implement sticky bucketing on both swift and kotlin SDKs. The function signatures for the three functions in the interface/protocol are not async and do not accept callbacks. This presents problems because our implementation requires fetching the assignments from our API. We’ve already implemented this with the React SDK where the functions are async. Any recommendations?
    a
    • 2
    • 16
  • w

    witty-rocket-75751

    08/07/2025, 3:02 PM
    Hi growthbook team were there any changes in the deterministic hashing logic used in the gb_sdk_flutter between these 2 versions:
    2.1.0+0
    3.9.10
    @freezing-postman-69602 can you help here.
    f
    • 2
    • 4
  • b

    bulky-oil-75691

    08/07/2025, 3:12 PM
    Hey folks, I'm trying to migrate our metrics to the fact metrics approach. In particular, I'm working on those metrics that are defined as ratio metric such as retention rate: the denominator shouldn't include all the users exposed to the experiments but only those that are eligible to retain (those that at some moment converted to pay customers). While this was pretty easy to do in the old metrics (
    Fig.1
    ), I'm struggling to replicate the same with the fact metrics approach (
    Fig.2
    ), were apparently a user doesn't have to belong to the denominator to be eligible for the numerator -indeed you can end up having results which are bigger than 1 (
    Fig.3
    ). Is this expected or am I doing something wrong? Thanks fyi @stocky-zoo-87795
    s
    h
    • 3
    • 4
  • q

    quick-student-63473

    08/07/2025, 7:28 PM
    Hey all, I have a question about namespaces. I have an interesting requirement where I have 3 different experiments being run, A / B / C A needs to be completely exclusive from the other two experiments, but B and C can have some overlap and both can be actively running at the same time My question is: If I add them all to a single namespace and allot A to the first 50% of the traffic, then allot the B & C to the same final 50%, will it allow that functionality to work? The UI seems to allow me to have the two experiments occupy the space in the traffic allocation without issue. But I don't have an easy way to test if this will have the result I'm hoping for.
    f
    • 2
    • 1
  • w

    wooden-farmer-44624

    08/07/2025, 9:12 PM
    Quick question here team, any thoughts on building out personalization functionality in the near term?
    f
    • 2
    • 2
  • s

    stocky-zoo-87795

    08/08/2025, 8:58 AM
    Hey team, we are currently migrating our fact tables to the
    metrics.yml
    and I noticed some potential bug on this integration. 1. I had written an entire factMetric in my file, which later I removed from the .yml 2. Even though the metric didn't exist into the file, I still could see it into my UI 3. For being an official fact metric, I could not edit not delete it from the UI. So I ran a CURL DELETE on this specific metric. 4. Even though I had a positive response from the API, the metric still lives into our self-hosted environment. How can we clean that up? Any advices?
    f
    • 2
    • 4
  • p

    plain-cat-20969

    08/08/2025, 8:45 PM
    Hello all! We are interested in using the metric correlations and metric effect features but are a tiny team of three. Can we access those enterprise features somehow?
    f
    • 2
    • 1
  • f

    flaky-lion-33569

    08/12/2025, 5:16 AM
    Hey Growthbook Team, For the AI integration, do we already or plan to support Gemini models? Right now, documentation only talks about OPENAI.
    s
    • 2
    • 1
  • n

    narrow-afternoon-3154

    08/13/2025, 7:24 AM
    Hi Team do you have a good example of how to integrate GrowthBook with NestJS?
    s
    • 2
    • 1
  • q

    quick-ability-87647

    08/13/2025, 11:47 PM
    hi team, can someone explain what the difference between
    RUN_VISUAL_EXPERIMENTS
    @
    edge
    vs
    browser
    is?
    s
    • 2
    • 1
  • a

    adorable-bear-66287

    08/14/2025, 7:49 AM
    Heyo! We have two deployments of growthbook running (test and prod). We have pro for the deployment in prod but not in test. This limits our development as we can't do eg. bandit testing in our test env. How can we get pro for our test env as well?
    s
    • 2
    • 1