Potential issue: I've an experiment that times-out after 3 hours, when refreshing data Potential solution: Is it possible to turn off the health tab on 1 experiment? More details When we have an experiment that has complex SQL metrics (like new users who visited checkout) then the data refresh can take a really long time, and I wondered if it's pulling tons of data for the health tab as it once timed out on that query - GB said the query was a 47gb. If this is the case, it would be useful to be able to disable the Health tab/feature on an experiment. (As always I love using GB and loving the updates 🙂)

Hi, I love the new approval feature for feature flags- Are there any plans to create something similar for starting/stopping Experiments?

Visualisation of “gained confidence” I think to mitigate the early peeking problem, but at the same time allow for early stopping of tests it would be great to have a visualisation of the uplift over time as more data comes in. This would allow to see when test results stabilises, but at the same time also highlight (likely) false-positives due to fluctuations in the data

We have multiple Projects and my team is always forgetting to assign a Project when creating feature flags, which can cause frustration as it's not obvious why your feature isn't showing up where you would expect. This could be mitigated by adding the Project assignment field to the Create Feature modal. 🙏

Make it possible to edit Visual Editor code changes directly in the experiment config interface:

DBT Integration

Hi there. Do you plan to allow for DBT integration so that model documentation, for example, could also be shown on metrics calculations? We are transferring some models to DBT for version control, and it would be nice to have this possibility, to avoid too much effort on metrics maintenance.

Automatic refresh of (key) dimensions Would be nice to select some relevant dimensions for a test that are always automatically refreshed. Currently they always require a manual run, which makes it more cumbersome to get the full picture of an experiment.

Hi All, I opened a Github issue about this, but I wanted to check in here what people think. In my organization we run a lot of experiments where we need to filter for activation with specific tags. Right now it seems the way to set up metrics based on these filters is to set up an activation metric for each individual experiment. Firstly, I'm not sure if I'm just missing something and there is a good way to accomplish this? If there isn't a good way to do this currently, I could imagine that parametrized metrics would be a good way to achieve this. This would allow you to create custom variables that are accessible in queries so you could include something like

WHERE tag = {{ customVariable }}

I think this could be implemented relatively easily and I would be interested in working on it if it would be interesting.

Features can work across projects (unless I'm misunderstanding you). You just need to make sure that you set up an SDK connection for all projects and define the features under that.

Hey, I create a lot of custom reports, and it would be great to have: • Duplicate existing report to modify with all the metrics, change dimensions etc • Create templates that could be reused I prefer using custom reports as I can link them in docs for teams to look at the numbers, not just screenshots.

Not sure if it already works, otherwise consider it a suggestion - it would be great to have the option to be able to create saved groups using emails that contain wildcards. For example - my email address is lee@company.com. when testing features / our app we often create new user accounts and use a syntax like lee+test1@company.com. It would be great to be able to setup a saved group with a wildcard so i don't have to add lots of users to our internal test user saved group (e.g. lee+*@company.com to capture all test accounts i might create automatically).

Hi! We have a core set of metrics that we add to every single test. It would be great if we could define this as a pack (in the desired order) and add them with one click instead of selecting each one every time

Give user control over how activation metrics are used. Basically with a lookback window (could happen before experiment exposure) or only after (current config).

Hi. A suggestion to make power analysis easier to use would be remove the need to enter a mean and std deviation for each metric and estimated user per week by using historical data. The user would simply select a metric and select an ‘activation metric’ to represent the population going into the experiment. The analysis would then calculate required inputs using historical data.

Really need to identify experiments with temporary rollouts in the experiments list. I would say include them in the "Running" list but with some visual identifier.

Hi, I've been playing around with coderefs recently in our GB instance and I have a couple of things I think might make the feature more useful: 1. in our codebase right now we have a couple of long-lived branches (main, develop) I'd like to be able to "pin" these to the top of the list when showing code refs as these would generally be the most important references to any features in our apps and 2. provide a way to remove old code references once the branch they were submitted from has been deleted. I'm not sure if this exists already but I don't see anything about it in the github workflow documentation

"Hi, do you know how I can intercept the feature request in Android in order to map it? Thanks!"

Hi it would be great if we could add I.e click tracking on an element in the visual editor. The output could then be some event push on the datalayer

Hey, I want to ask about the interaction effects feature. I heard in a video that there are plans to develop this, and it might already be in progress. Any update would be great 🐿️

Hello, We do a lot of A/B tests where we add new functionality to a page, the new functionality almost always gets new tracking just so we can see the usage and how new features perform. Of course we can’t compare these numbers to the control as the functionality doesn’t exist there. So we would like to have a 4th type of metric in the reporting called “Informative metrics” which then only shows the metrics from the new feature and where we can also connect other existing metrics too like transaction rate etc. so we can compare these metrics with the baseline like for example do users that use the new functionality perform better compared to the default segment. Basically the goal is to get more holistic view that can help us iterate and optimise new features. Hope it makes sense if not I would be happy to jump on a call and explain more in detail. cc @kind-father-85802

Love the new metric correlations feature! Helps me understand the wisdom (or lack of) in using different funnel metrics. I have found that a single outlier can really destroy the visualization's utility though. Requesting Pan/Zoom controls to help remedy this.

How about a dimensional breakdown for other experiment variations? I'm thinking about how to build one into a data model. Not simple though.

Could there be an alternative to _`PROXY_HOST_PUBLIC` for internal requests?_ _For the SDK (client/server -> proxy),

PROXY_HOST_PUBLIC

is used_ _For GB -> proxy,

PROXY_HOST_PRIVATE

could be used_

Hi growthbook team. I’ve just created a small feature request which would be very helpfull for the userbase: Problem Statement Currently, when creating an experiment using the Visual Editor, the URL targeting set for visual editing automatically becomes a mandatory runtime targeting condition. This creates an unnecessary limitation where visual experiments can only run on the exact URLs (or URL patterns) specified during visual editing, even when users want to apply the same visual changes across multiple page types using attribute-based targeting. Current Behavior 1. User creates a visual experiment on URL:

<https://www.sanitairwinkel.nl/douche/douchekoppen/>

2. This URL is automatically added as a targeting condition for the experiment 3. The experiment will only run on pages matching this URL pattern 4. Even if users add additional attribute targeting (e.g.,

pageType: "category"

), the URL restriction still applies 5. Users are forced to use complex regex patterns or multiple wildcards to target similar pages Expected Behavior 1. The Visual Editor URL should be used only for: ◦ Loading the page in the visual editor interface ◦ Applying visual changes during experiment setup 2. Runtime targeting should be completely controlled by: ◦ The targeting conditions defined in the “Targeting” section ◦ Custom attributes and rules set by the user ◦ No automatic URL restrictions unless explicitly added by the user Would love to upvotes on the isssue ❤️ https://github.com/growthbook/growthbook/issues/4380

Hi team, I’ve updated the issue with more information hope you find this feature ready to be worked on. https://github.com/growthbook/growthbook/issues/4200

Hi GrowthBook team, I have a small feature request that would significantly improve our development workflow. We currently use three environments. We start developing tests in dev, run QA in staging, and finally launch experiments in production. At the moment, it’s quite tedious to ensure that a test is running only in the intended environments, especially when multiple feature flags are involved. To move a test between environments, we need to manually disable/enable the test rules one by one. It would be extremely helpful to have a single toggle in the test configuration to control in which environments the test is active. This would simplify moving a test through the dev → staging → production pipeline. As an alternative, if GrowthBook DevTools allowed activating tests that are still in draft (not running yet), we could run QA without having to start the test prematurely. That way, we could validate everything in staging before officially launching in production. Thanks for considering this request!

MCP Request (Claude Summarized based on limitations I hit when wanting to interact with metrics)

Enable drag-n-drop re-order of the dimension column list in the data source query modal OR enable reorder in the experiment results dimension drop-down which is really where it matters.

Sorry I put this one in the wrong channel so here its again but then in the right channel. @strong-mouse-55694 I hope you can find time for this one… might also be usefull to make it available in the mcp: https://github.com/growthbook/growthbook/issues/4599

Hi GrowthBook team, I would like to request a feature which is quite important for our self hosted instance. 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. So, if you would have this somehow prioritized would awesome🙂 Thanks in advance