Eng Intelligence

The Eng Intelligence feature provides users with key metrics and high-level data to gain insights into services, bottlenecks in the pull request lifecycle, incident response, and more. This tool helps users understand what’s going on between entities, teams, and users to help improve productivity.

To view, click Eng Intelligence in the main nav:
Click Eng Intelligence in the left navigation menu

Using Eng Intelligence

Eng Intelligence provides users with a table view of key metrics, detailed below.

By default, each Team entity in Cortex will have its own dedicated row. The View as hierarchy toggle allows you to group teams by the team hierarchy you've created. You can also use the Group by dropdown to view the metrics for groups, users, services, and owners.

Each cell will display a value for the metric listed at the top of each column. These values are recalculated every hour.

For count metrics, like "PRs opened," you’ll see a "0" if there are no data available. For average metrics, like "Avg PR open to close time," you’ll see "N/A" if no data are available to calculate averages.

Clicking on a row will open a side panel with a more detailed view of performance. In the panel, you can see all the metrics available for that row and a historical performance graph for each.

eng intelligence 2

You can adjust the time range for the graphs to be anywhere between 7 days and 6 months. This will update the graph view and maps to the table, so all metrics will reflect the new timeframe.

Show Scorecard view

The Show Scorecard view overlays Eng Intelligence with Scorecard performance by level when grouped by team or service. This view is not available when grouping by group, user, or owner.

eng intelligence 3

When you select a Scorecard from the Show Scorecard dropdown, the icon representing the level achieved by each entity will appear next to its name.

Metrics

You can create custom metrics for Eng Intelligence or you can use the built-in metrics.

Eng intelligence reports on the following built-in metrics:

Avg PR open to close time

Calculates the average time to close pull requests for each PR opened and merged during the selected time range.
Pulls data from GitHub and GitLab.

This metric provides insight into how long it takes to merge something, such as build time, reviews, conversations, fixing linter issues, etc.

If your Average PR open to close time is high, it’s worth investigating to identify the part of the development cycle that contribute the most to this time.

Average PR open to close time is related to other metrics, such as time to review and bottlenecks in average PRs reviewed each week. The key here is to examine the time and quantity of a particular activity.

Note that if some teams are using draft pull requests, their numbers may be higher.

Avg time to first review

Determines average time from first open to first review of a pull request for any PR that has been opened during the selected time range.
Pulls data from GitHub and GitLab.

For a subset of pull requests, this metric can provide insight into potential inefficiencies. For high figures, investigate whether this is due to the software process or roadblocks faced by team members.

Note that if some teams are using draft pull requests, their numbers may be higher.

Avg time to approval

Displays average time from when a pull request was first opened to when it was first approved for any PR opened during the selected time range.
Pulls data from GitHub and GitLab.

Average time to approval can capture review-related bottlenecks in the PR cycle. When this figure is high, there may be opportunities to improve processes and PR sizes.

Note that if some teams are using draft pull requests, their numbers may be higher.

Avg deploys/week

Calculates the average number of deploys per week over the selected time range.
Pulls deploy data added via the Cortex deploys API.

Deploy change failure rate

Displays the number of rollbacks divided by number of deploys in a given time range.
Pulls deploy data added via the Cortex deploys API.

Mean time to resolve incidents

Calculates average amount of time from incident open to resolution in the selected time range.
Pulls data from PagerDuty.

PRs opened

Displays a count of pull requests opened during the selected time range.
Pulls data from GitHub and GitLab.

Pull requests opened is particularly useful as a throughput metric. When reviewing this data, consider the expected minimum activity for a developer.

On an individual level, evaluate how much time a team member spends building features versus supporting others. You can also assess how much time a team is spending shipping code versus other teams.

Note that while this metric provides useful insight, weekly PRs merged may be a more meaningful figure.

Weekly PRs merged

Calculates how many pull requests were opened each week, averaged across the weeks in the selected time range, to determine how many PRs were opened and merged each week.
Pulls data from GitHub and GitLab.

This throughput metric provides insight into how many things make it to the default branch and are closed out.

In theory, this figure should match the trend for Average PR open to close time, since you don’t want too many pull requests kept open.

Average PRs reviewed/week

Calculates the number of pull requests that were reviewed each week, averaged across the selected time frame.
Pulls data from GitHub and GitLab.

This metric helps users understand bottlenecks in the review stage due to load balancing work, education gaps, onboarding, career progression, and domain mastery.

Note that this figure has been deduplicated on a per user basis, so if a user reviews a pull request multiple times, it will only display once within Eng Intelligence.

You may be spending too much time in the review stage if this figure is high, but you have a low number of commits and a low number of merged pull requests. If this is the case, other parts of the PR lifecycle may be at risk.

Average commits per PR

Displays the number of commits required from PR open to close, and averaged across all PRs, for any PR opened and merged during the selected time range.
Pulls data from GitHub and GitLab.

This metric provides insight into activity trends by team members, as greater activity indicates more engagement.

Average commits per PR can be helpful during the onboarding process, so you can gauge how long it takes for a developer to reach the team’s baseline for activity.

Note that if some teams are using draft pull requests, their numbers may be higher.

Avg LOC changed per PR

Displays the average number of lines added plus lines deleted for pull requests that were opened/merged during the selected timeframe.
Pulls data from GitHub and GitLab.

This metric can provide information about pull request size. Ideally, developers should open consumable PRs that are easy to review, and thus are easy to push into production.

This figure can impact other metrics related to the PR cycle.

Incidents opened

Displays sum of incidents opened for that time range; based on the most recently assigned user/team for each incident.
Pulls data from PagerDuty.

Incidents opened/week

Displays sum of of incidents opened, divided by the number of weeks in the selected time range; based on the most recently assigned user/team for each incident.
Pulls data from PagerDuty.