Skip to main content

Measurement Standards

Jason Hreha· Updated September 5, 2026

Δ-B #

  • Report baseline %, post %, Δ-B in pp, relative change %, n, denominator, window, assignment, data source.

Time to First Behavior (TTFB) #

  • Definition: elapsed time from exposure/start to first completion of target behavior.
  • Set a target from the behavior’s required timing and domain evidence. Targets such as < 5 minutes for a consumer flow or < 1 day for a B2B pilot are illustrative planning values, not established cross-domain benchmarks.

Behavior Retention #

  • Report the share of a cohort performing the target behavior at decision-relevant intervals. Define the intervals and repeat criteria for the domain, and include cohorting, denominators, and observation windows. D30 and D180 are examples, not universal checkpoints.

Chain Risk Score (CRS) #

  • CRS = min(BR_buyer, BR_champ, BR_user). It identifies the lowest observed role-completion rate; it is not the probability that a whole adoption chain succeeds. Pre-register a domain-specific rollout threshold with a rationale. No universal CRS threshold establishes safety or readiness in regulated settings.

Reporting rules #

  • Label unsourced numbers as examples or link an Evidence Ledger ID.
  • Use behavior language, not “value” proxies. Always include denominator and window.

Instrumentation events #

  • exposure_shown: surface, variant, timestamp
  • behavior_attempt_started: behavior_id, context, timestamp
  • behavior_completed: behavior_id, context, timestamp, success:true
  • behavior_abandoned: behavior_id, context, step, timestamp

TTFB = behavior_completed.timestamp − exposure_shown.timestamp for first completion. Δ-B = completion rate in the window vs baseline, denominator explicit.

Example report #

The following values are illustrative. They are not observed results and do not support an effectiveness claim.

Project: Onboarding flow v2 → target behavior “Upload all required documents in one session”
Segment: New applicants (web), first 14 days
Window: 30 days; Denominator: all exposed

Baseline completion: 28% (n=4,120)
Post completion: 43% (n=4,305)
Δ-B: +15 pp

TTFB (median): 07:40 → 03:10
Retention D30 of behavior: 9% → 17%

Notes: Example calculation only. A published case would require a named calculation or source receipt and a canonical Evidence Ledger ID.

Why temporal design matters #

Measurement outcomes depend on time windows and cadence. Meta-analytic evidence shows that studies with precise temporal hypotheses detect change more reliably. This supports our use of explicit windows, cohorting, and TTFB.

See also