Analyze every load test from one Metrics Dashboard.
Inspect response times, capacity, page behavior, server resources, and errors across every user level. Diagnose front-end and back-end bottlenecks yourself in six interactive tabs, or, on Pro, generate a written analysis with one click on the Report tab, built from 25+ years of professional performance-engineering methodology.
Explore the metrics behind every load test.
The dashboard has six tabs that each answer a different diagnostic question. Use them in any order. Drill into the tab where the test went wrong.
Overview tab. Test health at every user level.
The Performance Health Matrix shows response time, error rate, CPU, and memory at every tested user level. Green is acceptable, yellow is warning, red is critical. For the test shown, average response time climbs from 907 ms at 50 users to 9,085 ms at 450, and CPU goes from 10 percent to 93 percent over the same range. The first failing user level is visible in seconds without drilling.
Analysis tab. Capacity and the response-time cliff.
The Analysis tab plots response time against concurrent users instead of clock time. Max response time stays flat through the first 250 users in the test shown, then climbs from 25 seconds to over 250 seconds across the next 200 users. That is where capacity ends. Below the chart, the Performance Goals panel turns your duration and error thresholds into three numbers: Minimum Capacity, Maximum Capacity, and Exceeded Capacity. For the test shown, 250 users with all goals passed and 300 users with goals failed.
Pages tab. Which transactions slow down first.
The Pages tab ranks every transaction in the test by cumulative time impact. Click a row to see the response time curve and the component breakdown (DNS, connect, wait, receive) for that page across user levels, plus how its duration changed from baseline to capacity. Real page names like Login.aspx and JSWorkStylesAssessment.aspx with their actual numbers from the run.
Servers tab. What resource saturated.
Server metrics plot on the same x-axis as response time. For the test shown, one server's CPU rises from 10 percent at 50 users to 93 percent at 450, while another stays under 10 percent the whole time. The dashboard flags any resource whose curve correlates above 0.94 with response-time degradation, so the saturated host is found without manual cross-referencing. Server data comes from WPLoadTester's Server Monitoring agent. Read about Server Monitoring
Errors tab. The failure pattern doing the damage.
The Errors tab shows four numbers at the top: average error rate, peak error rate, distinct error types, total errors. Below, every error message is broken out by count and percentage. For the test shown, 1.70 percent average, 8.09 percent peak, nine error types, 926 total errors, with 68.1 percent of failures from a single HTTP 500 on /pdl/secured/disclosure. Click any error to see the exact failed request with the response headers diffed against what was expected.
Report tab. The Pro AI Bottleneck Report.
The Report tab is where the AI Bottleneck Report lives. On a finished Pro test, one click produces a written analysis: performance summary, capacity numbers, slow page groups, saturated resources, dominant error pattern, all in plain English with the actual numbers from your run. The report follows a methodology refined over 25+ years of professional performance engineering (the credibility story below). Pro feature.
Built on 25+ years of professional performance reports.
The AI Bottleneck Report does not invent a report structure on the fly. It uses the methodology Web Performance has refined since 1999 across professional performance engagements. What belongs in a professional load-test report, where to look for each finding, and how to analyze it, all distilled into the same one-click output the Report tab generates for any finished test in Pro. The methodology is the basis, not training data: the AI follows a documented framework and does not learn from your tests.
The report names capacity in user terms. It cites exact page names, durations, and the user level where goals begin to fail. It groups related page behavior instead of listing isolated slow requests. It names the saturated resource and the host it ran on. It identifies the dominant error pattern and what share of total failures it accounts for. This is what an enterprise team needs to act on a load test result.
Same metrics, your report format.
Need a different report layout? An internal template, a release-summary spreadsheet, a compliance audit packet, a Slack post for the team? WPLoadTester 7 ships an MCP server that exposes the same dashboard metrics to any MCP-compatible AI client (Claude Code, Codex, OpenCode, Cursor, Windsurf, Claude Desktop). Point your AI client at the dashboard, ask for the format you want, and the report comes out the other side. The metrics are the same as the ones the Pro AI Bottleneck Report runs on; the output format is whatever your team writes the prompt for.
Desktop for test runners. Portal for the rest of the team.
The dashboard renders in the WPLoadTester desktop app, during a running test and after it finishes. It also runs in the browser through the Web Performance Portal, where every test result your team uploads is visible to anyone in your organization. Same six tabs, same numbers, same Report tab. The desktop app is for the engineer running the test; the Portal is for everyone else.
Try the Metrics Dashboard on your own load test.
The Metrics Dashboard ships in every WPLoadTester 7 install. Try the live demo on a real test, or request the beta to run your own. The AI Bottleneck Report on the Report tab is the Pro-tier addition.