MongoDB Growth Experiment: Performance Advisor Impact Clarity
While the goal of MongoDB's Performance Advisor is to provide users with guidance through the index creation process, an analysis showed that between December 2024 and March 2025, only 22.7% of Performance Advisor visitors with index suggestions applied an index.

This low conversion rate sparked questions around why users were viewing our suggested indexes yet not acting upon them. In partnership with the Growth team, we designed an experiment to increase index creation within Performance Advisor.

Experiment results showed a 9.4% increase in the Performance Advisor index creation rate post changes.
Final flow surfacing expected impact of indexes to users
What is Performance Advisor?
Performance Advisor is a tool for MongoDB's paid-tier customers that provides index suggestions to improve their query performance. Proper index usage is crucial to ensure a customer's queries remain performant. While large enterprise companies have teams of engineers dedicated to monitoring query efficiency, self-serve customers often struggle in this regard, having to evaluate metrics such as query frequency, scanned bytes, and documents examined to make a decision about which index to use. To assist users with this process, Performance Advisor was built.
Performance Advisor in Atlas
Problem and Hypothesis
The key problem with Performance Advisor in the initial state of the project was that we presented only the raw query metric data to users with each index suggestion, placing the burden on the user to evaluate the impact of the suggested index; however, many users simply do not possess the necessary technical knowledge to make a decision. Our hypothesis for this experiment was that by directly surfacing the estimated impact of each index, we would see an increase in index creation within Performance Advisor.
Users struggle to translate query metrics into actionable insights about indexes
Surfacing Expected Impact
To give users a clearer picture of the effect a suggested index would have on their queries, we partnered with the Telemetry team. Internally, we rank each suggested index's impact based on the number of queries affected and the total amount of data read, and we added a banner at the top of each index card to surface this information. I also took this opportunity to update the visual polish of the card by adding a heading section to create visual separation in the card.
Educating Users
While our goal was to simplify the process for users to understand the impact of a suggested index, we also wanted to encourage users to learn more about indexes in general. To do so, we added in-line definitions for each index metric to guide users towards how they should be evaluating each metric.
Providing Positive Feedback
To further emphasize the impact of creating an index, we directed users to Namespace insights at the end of the index creation flow. Namespace insights is another tool in Atlas that allows users to monitor their query latency across MongoDB collections. Redirecting users to Namespace Insights after they create a suggested index allows them to see the real-time change in their query latency, acting as positive reinforcement for index creation.