Analysis Insights: Stop Hunting for Root Causes in Your Load Test Reports
We are launching with this post a new series of blog articles and LinkedIn posts titled "Features Sitting Idle".
In this series, we explore key features of OctoPerf that are either misused, misunderstood, or simply unknown to our users. It's time to shine a light on these hidden gems, features that are already there, ready to become a central part of how you test.
"Your Test Just Finished. Now You're Looking for the Problem."¶
This is probably the most common situation after a load test. The report opens, the charts appear, and the engineer then spends a long time trying to correlate metrics that don't tell a clear story at first glance.
What We See Across Support Conversations¶
Across our support interactions, a few recurring themes stand out when users struggle with their reports:
- Multi-scale chart confusion. Charts displaying multiple metrics on different scales make it hard to tell whether a spike is a real performance issue or just a rendering artefact.
- Error percentages that don't match expectations. Users often can't tell if the errors they see are significant, isolated, or part of a broader degradation pattern.
- Long test runs, longer analyses. After a long test run, finding where the actual degradation started can take as long as the test itself.
This isn't a skills problem. A load test report generates dozens of simultaneous metrics: response times, hit rate, error codes, virtual user behavior, and correlating them manually takes time, even for experienced engineers.
The OctoPerf Answer: Analysis Insights¶
OctoPerf has a native answer to this: Analysis Insights. This report item automatically analyzes the evolution of your test data over time and surfaces global alerts.
It's powered by an in-house algorithm built from the feedback of our most experienced performance testers. The detections it covers include:
- Hit rate drop correlated with rising response times
- Intermittent versus persistent errors
- Abnormal behavior on a specific scope (a script, a container, a region)
- Sudden shifts in distribution that classic averages would hide
Insights does the synthesis for you and points you directly to where to look.
How to Use It¶
The Analysis Insights report item lives alongside the other items in the report designer. You drop it into your report once, and from then on every test result you open through that report immediately shows the alerts at the top.
A few practical tips:
- Add Insights as the first item of your report, so anyone opening it sees the alerts before scrolling through the charts.
- Use it on both your smoke tests and your endurance tests. The patterns it catches differ widely between short bursts and multi-hour runs.
- Pair it with Report Templates so every new project starts with Insights enabled by default. No more rebuilding the same layout after each campaign.
Why It Matters¶
A load test report isn't a deliverable. It's a tool to make decisions. When the analysis step takes hours, the testing cycle slows down and so does the feedback loop with the development team.
Analysis Insights does not replace the engineer. It removes the most mechanical part of the analysis - the cross-referencing - so the engineer can focus on the diagnosis and the recommendation.
Final Reflection¶
If you're still opening your reports and scrolling through every chart looking for "the spike that shouldn't be there", Analysis Insights is the feature you didn't know you were missing.
It's already in your OctoPerf account. It just needs to be in your reports.
Full documentation: api.octoperf.com/doc/analysis/edit-bench-report/report-items/insights/