We have specific products which we are testing product images on. and we want to make sure our metrics only include the specific products in a dynamic way. So, instead of looking at 'Purchases', we'd have a dynamic metric of 'Purchases for the products being tested'. Evidently we could make custom metrics for each product (or each test) but this will quickly explode, so ideally I'd like a way to pull in the products which are being tested dynamically.

Hi Joe, Metric Overrides in GrowthBook provide experiment-specific controls for metrics, allowing you to override the metric's default settings for a particular experiment. This can be useful in situations where you want to use different settings for a metric in a specific experiment compared to its usual settings. To configure dynamic metrics in GrowthBook, you can use the "Metric Overrides" feature. This allows you to provide experiment-specific controls for metrics, enabling you to override the metric's defaults. For example, you can override metric windows and risk thresholds. Here's how you can do it: ​*Go to the Experiment Configuration*​: Navigate to the experiment configuration page in the GrowthBook UI. ​*Add Metrics*​: You can add metrics as goal metrics, guardrail metrics, or both. ​*Add Metric Overrides*​: In the metrics section, you can add "Metric Overrides" for the metrics you've added. This will allow you to override the default settings for those metrics for this specific experiment. Further information on this can be found here- https://docs.growthbook.io/app/experiment-configuration#experiment-metrics Would this feature work for your business case?

one way to do this might be on the event tracking side

Hi @flaky-noon-11399: I'm interested in your suggestion of using metric overrrides. How would I select particular products from this interface?

Hi Joe, apologies for the delayed response here. My apologies for the confusion here, it looks like the metrics override feature built into the UI is to override the thresholds and conversion, which may not be applicable for your business case- sorry about that. The best solution here will be the coded method described by Graham. Please, can you review the coded method above?

Thanks Natasha