A Score at the End of Every Run
A single 0–100 number, computed from the timings of the transactions you recorded. Not from injected benchmark executables. Click a score chip to open the run summary, drill into the five measures, and generate the built-in PDF report.
Test Runs
Every completed test carries its LoadGen Score, verdict and regression status.
| Score | Load profile | Type | Started | Duration | Users | OK | Failed | Status | ||
|---|---|---|---|---|---|---|---|---|---|---|
| Citrix Enhanced | Citrix Enhanced | 25 Jun 14:23 | 01:00:00 | 11 | 9 | 2 | Completed | |||
| Citrix Enhanced | Citrix Enhanced | 24 Jun 22:10 | 00:41:22 | 11 | 8 | 3 | Partial | |||
| AVD - Marketing | AVD | 23 Jun 09:04 | 00:22:11 | 8 | 5 | 3 | Partial | |||
| Horizon - Finance | Omnissa Horizon | 22 Jun 17:52 | 00:55:03 | 12 | 12 | 0 | Completed | |||
| Citrix Enhanced | Citrix Enhanced | 21 Jun 12:31 | 01:00:00 | 10 | 8 | 2 | Completed |
What the LoadGen Score Is
A Score Across Five Verdict Levels
The score is on a 0–100 scale, also shown as 0–10. Each number carries a word verdict. Partial runs are marked partial. Runs without a baseline are marked unbaselined. Nothing is inferred that the underlying data does not support.
Five Measures, One Score
A Weighted Sum of Five Measures
All five measures are computed from the run itself. Click a measure to drill in. Toggle “omit a measure” to see the score re-normalise in real time. This mirrors what the product does when a component cannot be measured on a given run.
This run's formula
Score = 30.4% Reliability + 30.4% Responsiveness + 13% Consistency + 13% Trend vs baseline + 13% Infrastructure
No hidden weights. When a component cannot be measured (no baseline, no per-machine CPU), it is left out and the remaining weights grow to keep the total 0–100.
Verdict legend
- Excellent90+
- Good75–89
- Fair60–74
- Poor40–59
- Critical<40
Try it
Click any measure to drill down. Try “omit a measure” below to see the score re-normalise in real time.
Omit a measure
Built From Your Applications
Scored From Your Applications, Not Benchmark Executables
| A typical synthetic experience score | The LoadGen Score |
|---|---|
| Weighted average of timers from dedicated benchmark programs injected into the session | Weighted measures computed from timings of your actual applications in the workload you authored |
| Tells you how benchmark executables performed on the machine | Tells you how your business applications performed under load |
| Score inputs are fixed by the vendor | Score inputs are your own transactions and measurements, openable down to the samples |
Understand the Run
Dashboards for the Investigation, Reports for the Hand-Off
Run data is queryable in Analyse. Build dashboards, compare 2 to 5 runs side by side, and chart percentiles, trends, and distributions. An AI assistant lives on the dashboard for on-demand analysis of the visible panels.
Analyse · Dashboard
Real charts over run data. Baseline vs current.
Analyse · Citrix Enhanced dashboard
Updated 3 min ago · owner: LoadGen · 6 panels · 2 comparison series
Baseline-Aware by Design
The Second Run Is Already a Comparison
- ▸ Pick any earlier run of the same profile as the baseline. The previous finished run is suggested automatically.
- ▸ The Trend measure and the run summary's Regressions line are computed against that baseline. “None vs baseline” means the underlying measurements support the claim.
- ▸ Reports with a baseline include improved and regressed columns since the previous run. Release over release, patch over patch.
- ▸ No baseline available? The run is scored on the remaining measures and marked as unbaselined. The Trend measure is dropped from that run's score rather than fabricated, and the weights of the other four measures re-normalise so the number is still on a 0–100 scale.
Screenshots Captured During the Test
Numbers Tell You What Happened. Screenshots Show It.
- Agents capture screenshots during the run, tied to the transaction and sub-transaction executing at that moment.
- Each screenshot carries its timestamp and an optional message (for example, the error text that triggered the capture).
- De-duplicated by content. A repeating error stores once, not five hundred times.
- Surfaces alongside the run's results and feeds into the AI-assisted analysis.
The Six-Step Load Test Flow
From Idea to Decision, in Six Steps
How a load test moves through the platform. Steps 4 through 6 execute automatically after the run finishes.
Author
Platform wizards (Citrix, RDP, AVD, Omnissa) capture the scenario. Version-controlled .lgs workloads. Phases: Idle, Ramp-Up, Sustain, Ramp-Down.
Deploy
The Launcher Appliance provisions the agent fleet from a downloaded VM image, with updates and preflight checks handled by the platform.
Execute
The live cockpit streams concurrency, errors per second, and phase progress in real time. Pause, resume, or stop the run at any point.
Score
The run ends with its LoadGen Score, word verdict, and regressions-vs-baseline line. Computed automatically as the run finishes.
Understand
One-click PDF report, side-by-side comparison of 2 to 5 runs, percentiles and trends, AI analysis over the full run context.
Decide
Baseline, trend, and capacity evidence. Guesswork replaced with numbers you can hand to the release meeting.
Who Reads What
The Same Run, Read Three Ways
Engineer
Drills from the score into measures, samples, and screenshots.
IT Manager
Watches verdict and trend across releases.
Business
Reads one number per release and a PDF that explains it.
Related
Related Reading
See the LoadGen Score on Real Data
Each completed test includes the LoadGen Score and the built-in management-ready PDF report. No extra tooling to configure. No exports to hand-assemble.
