Skip to main content
Every report carries a quality score graded A → E with a 0–100 value and a GO / NO-GO verdict — in the report header, the dashboard widget and the PDF / HTML exports. It is computed by a single deterministic engine from the data already on screen: no extra API call, no AI. The badge summarises the metrics; it never replaces them.

​
How the grade is built

The 0–100 composite weighs four components. These are the weights of the Standard profile, selected by default; Strict, Ramp-up and Custom profiles can use different values: Standard profile grades, before caps are applied: A ≥ 90 · B 75–89 · C 60–74 · D 40–59 · E < 40. Other profiles’ thresholds are listed below. When traceability is absent, coverage is dropped and the remaining weights are renormalised (a flag says so). Fewer than 5 executed tests raises a low confidence flag.

​
Transparency

Click Show details to expand the breakdown — each component (value, points, weight), the caps applied and any flags. The score is never an opaque number. Under the grade, one line names the scale in force and the GO threshold actually applied, and carries the low confidence reserve when fewer than five tests ran — so a verdict can be defended without opening Settings. On screen only: the exports gain no new line.

​
Non-negotiable invariants

  • ≥ 1 open blocking bug among the bugs of the report (open state and blocking severity according to the applied scale) caps the grade at C and forces NO-GO.
  • An untested in-scope requirement forces NO-GO.
  • No executed test shows not evaluated — never an E.
  • GO also requires the composite to reach the scale’s GO threshold and coverage to reach its minimum.
  • A pass rate below the scale’s floor caps the grade at C and forces NO-GO. The other components cannot compensate for it — an average compensates, a floor does not.
  • An execution rate below that same floor does the same: satisfying the pass rate by running less no longer buys a GO. A case marked Not Applicable is not a skipped case.
  • A failed test with no linked defect caps the grade at C and forces NO-GO (on by default since v2.64.0).

​
Scale profiles

In Settings, choose a project-level scale so everyone scores a report identically: The values below belong to the three predefined profiles, before customisation. The grade is determined by the highest threshold reached, then caps are applied; below the D threshold it is E. An open High bug therefore does not, on its own, trigger the blocking-bug gate in Ramp-up. The verdict’s other rules still apply. Each scale’s GO threshold doubles as its pass-rate floor and its execution-rate floor — 0.85 Strict · 0.75 Standard · 0.65 Ramp-up — so there is no separate number to set. Tuning is bounded (GO threshold 50–95, a minimum coverage, component weights forced to sum to 100, A > B > C > D thresholds); any off-preset value becomes a Custom scale. The rules above apply using the selected scale’s values. When interpreting a saved report, consult the scale captured in that report: changing the project setting does not rewrite its results.
If an older report has no saved scale, TestPulse uses the Standard default. It does not apply the current project profile to that report. A report with a saved scale continues to use that scale.

​
Risk-based weighting (optional)

An optional, orthogonal toggle shows a weighted confidence beside the grade. It weights execution and coverage by the importance of what is missing — test-case priority, story risk and story points — using transparent additive tiers (never a multiplication). The flat A–E score is never altered; a dimension whose data is absent or uniform auto-disables and says so.

Decision support

Watch points, failures without a defect, override.

Coverage depth

Is it well covered?