25 results found
-
Support ODP Audience Configuration per Environment in Feature Experimentation
Recently, I have been working on a project where we integrate ODP Audiences with Feature Experimentation and I noticed one limitation that makes testing and environment setup harder than expected.
In ODP, we have separate environments, for example int/preprod/prod, each of them has its own URL and public/private keys.
In Feature Experimentation, we usually work with one project that contains multiple environments (added in settings), Feature flags are then configured and evaluated separately per environment inside the same project.
The problem is that the ODP Audiences integration is configured only once for the whole Feature Experimentation project. That means I…
10 votes -
A method to list all available experiments
We are currently planning implementation of Optimizely as replacement of our current experimentation solution. In order for developers - and also business - to verify behavior we currently show a list of all active experiments and allow users to force a variation. In the current SDK there's no way to list all experiments. I prefer not to parse the datafile myself, so I would like to see a method that lists all experiments, including all variations, so we can migrate with the same functionality as we currently have.
Since the SDK already parses and interprets the datafile I feel this…
3 votes -
Dealing with group AI Policies blocking out Opal
Customers with corporate AI policies restricting usage to approved vendors are blocking Opal adoption entirely or partially. While each scenario has its unique challenges, CS teams currently lack a structured playbook to navigate this. including security and compliance talking points, and escalation paths to help customers gain internal IT/legal approval to whitelist Opal.
1 vote -
Improved visibility on credit consumption
A recurring theme in user feedback concerns the predictability of credit consumption. Two related issues come up most often:
Lack of upfront visibility — users are unable to anticipate how many credits a given task will consume before running it.
Inconsistent consumption — when users have an expectation, actual usage varies run-to-run for comparable tasks
The underlying ask is for greater predictability and traceability: a clearer sense of expected cost before execution, and a transparent breakdown of what was actually consumed afterward. This would meaningfully increase customer confidence and reduce hesitancy around running larger or repeated tasks.
1 vote -
Allow MCP server to be able to create metrics based on event tags and rename them.
Currently, I'm not able to use MCP server to create metrics on A/B tests. It can only create the metrics without event tag filtering or customisation, and it also cannot do custom naming for the metrics. If I need to create the same metric as total conversion and unique conversions, I'm not able to do that.
1 vote -
Multi-flag experiment capability in Feature Experimentation (cross-flag MVT / interleaved testing)
Customer: Inspire Brands
Request: Multi-flag experiment capability in Feature Experimentation (cross-flag MVT / interleaved testing)
Problem:
Inspire Brands uses Feature Experimentation to manage their testing program across their QSR digital ordering funnel. They deploy individual feature flags for distinct features on the same page — for example, their menu page has separate flags controlling a search feature, a product carousel, and filters. Over time, they accumulate multiple independent flags on a single page or funnel area and need to understand which combination of those flags produces the optimal user experience.
Today, the only way to test combinations of features across…1 vote -
Allow MCP server to query MAU data
I've got an IT support ticket to get my Claude connected to the remote server. In the meantime, I'm using the locally installed server to audit our MAU consumption. The MCP server is struggling to get users for each a/b test rule, which feels like a miss that might be easily fixed in a future release? I was able to create a workaround with a script that queries the Results API, but this took a while and was quite clunky.
2 votes -
Default Experiment Metric Auto‑Apply
Feature: Default Experiment Metric Auto‑Apply
Automatically apply a predefined default metric to every new experiment created in Optimizely, ensuring that all tests are consistently configured with a core measurement from the outset.
Today, metrics must be manually added during experiment setup, which introduces variability and increases the risk of inconsistent or incomplete measurement across teams. This feature removes that dependency by embedding a standard metric directly into the experiment creation flow.
Improves experiment quality and consistency
By ensuring every experiment starts with a core metric, teams align on how success is measured. This directly addresses known challenges around inconsistent metric…
1 vote -
Filter by Product Owner & Product Area Field
User story: As an engineer/product manager I want to filter experiments by product owner and product area so I can quickly see all experiments I own without manual searching
Problem:
Even when ownership or context is known at an individual experiment level, this information is not surfaced in programme‑level reporting. As a result, Program Overview and Trend Reports provide activity and outcome data without sufficient context around who is responsible or which area the work relates to. This makes programme reporting less actionable for leadership, experimentation governance, and portfolio decision‑making.Solution:
Extend Program Overview and Trend Reports to include Product…
2 votes -
Product Owner Field
Use case:
As an engineer/product manager, I want to add product owner name to an experiment so that ownership is clearProblem:
Today, experiments in Optimizely do not clearly identify who owns them from a product perspective. Ownership is often inferred from context, naming conventions, or tribal knowledge, which makes it difficult to quickly understand accountability, especially in programme‑level reporting, cross‑team reviews, or leadership updates. This lack of explicit ownership creates friction when reviewing experiment outcomes, following up on long‑running tests, or assessing portfolio health.Solution:
Introduce a dedicated Product Owner field at the experiment level that can be populated…2 votes -
navigation
In feature experimentation it would be good to be able to add custom fields to each experiment. For instance location or team - this way teams can easily filter to their experiments to reduce noise
3 votes -
Support multiple User IDs
Currently, Optimizely (FX) only supports one User ID.
However, a user might use different IDs, depending eg. on log-in status or similar. The same way, experiments might require randomization and sending events on any of those IDs.
This will be a key feature for us going forward.
1 vote -
Make custom fields usable on Flag overview page
On the flag overview page it would be very helpful if content of a custom field would be
- filterable
- searchable
- sortable
This way the custom fields would be even more helpful to organise a decentralised experimentation program.1 vote -
Service-Level API Key Management
Currently, generating an API key requires associating it with an individual user account, which introduces an anti-pattern for teams and production workflows. This approach can create challenges around ownership, security, and maintainability, especially when users leave or roles change.
We would like to request support for service-level or organization-level API keys that are not tied to a specific personal user account. This would enable more robust and scalable integrations, better access control, and improved operational reliability
1 vote -
Allow access to FX instance without project assignment
Enable a way to invite a new user to an instance of Feature Experimentation without having to assign them to at least one existing project. Use case is for a large org with many product teams. Say an admin wants to extend Optimizely access to a new product team, but wants to do so without giving them access to any existing project - there is no way to do this today.
1 vote -
Customizable MAB interval
Make the calculation interval of MABs customizable (currently 60 mins by default). The specific need is around an MAB in which we would expect a delay between the initial decision and when we would expect the primary metric to be triggered (up to several hours). This creates a scenario in which the MAB is diverting traffic based on user behavior that has not yet occurred.
1 vote -
Attribute creation within audience builder
Within the FX audience builder, if i search for an attribute that does not exist I should be able to create and add a new attribute without having to exist audience builder, go to attributes tab, and then back to audience builder.
1 vote -
logic
Current limitation:
When setting up logic for a template, users can only apply the AND condition. This restricts flexibility because it prevents combining AND and OR conditions within the same template.Why this matters:
Most systems allow both AND and OR options so that users can build more complex logic in a single template. Without this capability, users are forced to create multiple templates for different workflow scenarios instead of managing everything under one template using combined AND/OR logic.2 votes -
Product Area Field
User story: As an engineer/product manager, I want to add product area to an experiment so that I and other stakeholders can review how many experiments there are in our domain
Problem:
Experiments are currently difficult to group or analyse by product domain (e.g. Search, PDP, Checkout). As a result, programme reporting lacks clarity on where experimentation effort is concentrated, which areas are under‑ or over‑represented, and how different product areas are performing over time. This limits the ability to assess experimentation coverage, prioritisation, and strategic alignment.Solution:
Add a Product Area field to experiments, using a controlled list of…1 vote -
New Booleans in the Decision Object
It would be easier to have some Boolean values in the decision object, which would simplify recognizing the exact status of a visitor (e.g., whether they are in the "off" variation or the "everyone else" category, or in the concluded/deployed "on" variation). "Enabled" and "variationKey" are not enough to distinguish the above cases.
Currently, they need to check string values and match specific values like "rollout-default" or "deployedtdl" which is not ideal. Booleans would be easier.
2 votes
.png){kind=link}
- Don't see your idea?