Behavioural Change Toolkit · Plan

Measurement Plan — Leading and Lagging

If you can’t show movement, you’re guessing — and when budget gets tight, guesses get cut.

The four types you need

Type Answers Example Frequency
Activity Did we do what we said? Sessions run, champions recruited Weekly
Leading — behaviour Are people doing the thing? % of calls logged same-day Weekly
Lagging — outcome Did it deliver the benefit? Forecast accuracy, cycle time Monthly/quarterly
Health / counter What are we breaking? Team sentiment, error rate, attrition, overtime Fortnightly

⚠️ Activity metrics are not progress. “94% training completion” measures attendance. It is the most commonly reported change metric and the least informative. Report it if you must, but never as the headline.

⭐ The behaviour metric is the one that matters. It’s the earliest honest signal you’ll get, and it’s the one most programmes don’t have.

The plan

# Metric Type Source Baseline Target By when Owner Cadence
1                
2                
3                
4                
5                

Four to six metrics maximum. More than that and nobody looks at any of them.

⚠️ Baselines must be captured before you start. Non-negotiable. Without a baseline you cannot demonstrate change and you will end up arguing about impressions with someone who has a different impression.

Designing a behaviour metric

Fill in

Behaviour: _______________ The countable event: _______________ Where it’s recorded: _______________ Denominator (% of what?): _______________ Can it be gamed? How? _______________ What we’ll also watch to catch gaming: _______________

Every behaviour metric can be gamed. Ask the question openly and design for it — a metric with a known gaming route and an unwatched counter-measure will be gamed, and the first you’ll hear of it is when the outcome doesn’t follow the behaviour numbers.

Prefer system-generated data. Self-reported adoption is optimistic by a wide margin, and asking people to report on their own compliance also signals distrust.

Counter-measures — pick at least two

The things that could quietly get worse while your headline number improves:

  • Quality / error / rework rate
  • Time taken per task ⭐ (always include this — “adoption up, everything takes 20% longer” is a real and common outcome)
  • Team sentiment / short pulse
  • Overtime or workload
  • Customer-facing impact
  • Attrition, especially of experienced people
  • Volume of workarounds

Counter-measures protect you

If you report only your success metric, someone else will eventually produce the deterioration you didn’t mention — and it will land as concealment. Reporting your own counter-measures, including when they’re moving the wrong way, is the single most effective way to be trusted with the good numbers.

Reporting to leaders

One page. Every time. Same format.

┌──────────────────────────────────────────────────────┐
│  [Change name] — Week 6                              │
│                                                      │
│  BEHAVIOUR      34% ▲ from 11%   (target 70% by wk12)│
│  OUTCOME        too early                            │
│  HEALTH         sentiment 3.2/5 ▼ from 3.6           │
│                                                      │
│  What changed this week:                             │
│  • Cut form from 12 fields to 4 (raised by champions)│
│                                                      │
│  What I need from you:                               │
│  • Decision on legacy switch-off date — by Friday    │
│                                                      │
│  What I'm worried about:                             │
│  • Northern region has no champion; adoption 9%      │
└──────────────────────────────────────────────────────┘

The last two boxes are the point. A report with no ask and no worry reads as either complacent or evasive — and it invites no decision, which means it changes nothing.

Rules

  1. Report trends, not snapshots. A single low number is a stick; a rising line is a story.
  2. Never publish low compliance widely. It normalises the behaviour you’re trying to change. Social Norms and Influence
  3. Say what you don’t know. “Too early to tell” is a legitimate and credible answer.
  4. Attribute changes to the people who prompted them. It costs nothing and buys everything.
  5. Set the review dates now — including the Drift Check — 30 60 90 dates, before anyone disbands.

→ Change on a Page · Pilot Design and Experiment Card · Drift Check — 30 60 90

← Back to the Behavioural Change Toolkit

Last updated: 11 August 2026