The short answer

Measure adoption across three layers: outcome metrics (is the technology improving the specific things it was supposed to improve?), depth metrics (how embedded is the technology in how people actually work, not just whether they've logged in?), and sustainability metrics (is adoption holding three and six months after go-live, or quietly declining?). Define all three before the programme starts — not after go-live.

Why the standard metrics don't work

The reason most transformation programmes default to activity metrics — logins, completions, activations — is that they're easy to collect and they look like progress. They're generated automatically by the technology itself. They can be pulled into a report with no additional effort. And in the early stages of a rollout, they tend to go up, which creates a sense of momentum.

The problem is what they actually measure. A login tells you someone opened the system. It doesn't tell you whether they used it to do anything differently. A training completion tells you someone sat through a session. It doesn't tell you whether they changed their behaviour as a result. A licence activation tells you IT provisioned access. It tells you nothing about adoption.

Measuring logins is like measuring whether people entered the gym. It tells you nothing about whether they got fit.

When programmes measure activity rather than adoption, they create a dangerous illusion. The dashboard shows green. The board is reassured. And the actual adoption problem — the one that will determine whether the transformation delivers its business case — remains invisible until the post-implementation review asks why the expected benefits haven't materialised.

Metrics that mislead
  • Licence activations
  • Training completion rates
  • Login frequency
  • Session length
  • Number of features accessed
  • "Users trained" headcount
Metrics that tell you something
  • Task completion time (before vs. after)
  • Workflow penetration by team
  • Employee confidence scores
  • Error or rework rates
  • Reversion rate at 90 days
  • Outcome attainment vs. business case

The three layers of meaningful measurement

Layer 1: Outcome metrics

These are the metrics that justify the transformation in the first place. They should be defined in the business case — and if they weren't, define them now, before go-live, so you have a baseline to measure against.

Outcome

Task-level time and quality

Time on task (before vs. after)
How long does it take to complete a specific, defined task that the technology is supposed to accelerate? Measure it before go-live. Measure it again at 30, 60, and 90 days. This is the most direct evidence of whether the technology is delivering its promise.
How to measure: time-track a sample of the target task in the two weeks before go-live and the equivalent periods after.
Error and rework rates
For transformations that are supposed to improve accuracy — in reporting, in data entry, in document production — error rates before and after are among the most compelling evidence of real impact. They are also often the hardest to collect, which is why they get skipped.
How to measure: audit a random sample of outputs before and after, using a consistent quality rubric.
Business case outcome attainment
What did the business case say the transformation would deliver — cost savings, revenue improvement, capacity release, customer satisfaction gains? These should be tracked explicitly, with the same rigour as financial performance. If the business case was built on assumptions rather than measurable commitments, this is the moment to convert them.
How to measure: map each business case assumption to a measurable metric and track it quarterly from go-live.

Layer 2: Depth metrics

These measure not just whether people are using the technology but how deeply it has penetrated their actual workflow. High depth means the technology is genuinely embedded. Low depth — even with high login counts — means it's peripheral.

Depth

Workflow penetration and confidence

Workflow penetration rate
Of the specific tasks the technology was supposed to support, what percentage are actually being completed using the technology — not just logged in alongside? This is different from login rate. A team where 80% of members log in weekly but only 20% complete the target task using the tool has 80% login penetration and 20% workflow penetration. The latter is what matters.
How to measure: define the target task precisely, then track completion in-system vs. outside-system for a sample period.
Employee confidence scores
Ask people directly, at 30 and 90 days post-go-live, how confident they feel using the technology for their specific role. Use a simple 1–5 scale for two or three specific use cases — not for the technology in general. Confidence scores are leading indicators of sustained adoption: low confidence at 30 days predicts high reversion at 90 days.
How to measure: a two-minute pulse survey, role-specific, asking about specific tasks rather than the tool in general.
Manager adoption rate
Track adoption separately for managers and individual contributors. Manager adoption is a leading indicator of team adoption — and a lagging manager adoption rate is one of the most reliable signals that team adoption will stall. If managers aren't embedded users by 60 days, the teams they manage are unlikely to sustain adoption past 90.
How to measure: workflow penetration rate, segmented by management level.

Layer 3: Sustainability metrics

These are the metrics most programmes forget to collect — and the ones that most accurately predict whether a transformation will hold. Adoption that peaks at go-live and erodes over six months is not adoption. It is a temporarily successful launch.

Sustainability

Retention and trend

Reversion rate at 90 days
Of users who were active in the first 30 days after go-live, what percentage have reverted to pre-transformation ways of working by day 90? A reversion rate above 20% is a serious warning sign. Above 40% is a programme-level failure that needs immediate intervention.
How to measure: cohort analysis of active users at day 30 vs. day 90, using system data.
Adoption trend (not snapshot)
Plot adoption metrics monthly for the first six months after go-live. A healthy transformation shows a rising or stable trend. A stalling transformation shows a peak at launch, a dip in months two and three, and a recovery only if active intervention occurs. The shape of the trend tells you more than any single snapshot.
How to measure: track your chosen adoption metric monthly and chart the trend line explicitly.

Define your metrics before go-live — not after

There is one discipline that separates organisations that genuinely measure transformation from those that post-rationalise it: defining metrics — including baselines — before the technology goes live.

If you don't know how long the target task took before the transformation, you cannot measure whether it's faster after. If you don't survey employee confidence before go-live, you have no reference point for the 30-day score. If you don't define what workflow penetration means for each team, you have no way of knowing whether the logins are translating into genuine use.

Pre-go-live measurement is not administratively convenient. It requires extra effort at exactly the point when the programme team is most stretched. It is also the only thing that makes post-go-live measurement meaningful — and the only thing that gives you credible evidence to present to a board that wants to know whether the transformation delivered what was promised.

Key takeaways

Stop measuring logins, activations, and training completions as primary adoption metrics — they measure activity, not adoption. Measure across three layers: outcomes (task time, error rates, business case attainment), depth (workflow penetration, confidence scores, manager adoption), and sustainability (reversion rate at 90 days, monthly trend). Define and baseline all metrics before go-live, not after.

Want a measurement framework built into your transformation?

Our AI Adoption Roadmap includes a pre-built metrics framework — outcome baselines, adoption depth scoring, and sustainability checkpoints — tailored to your programme.

Get your AI adoption roadmap
← The operating model question All insights →