How to scope a Minimum Viable Behaviour (MVB) for pilot testing

Lauren Kelly

A pilot lives or dies on clarity. You are not trimming a feature list. You are proving that one small, observable action is possible and valuable. This can be hard, as behaviour is everywhere. But if you cannot describe that action in under 10 words, it is not minimum.

Pin down the behaviour in one sentence

Use this formula. Keep each part concrete.

Actor + action + context + countable outcome

For example:

Shopper taps "Save to wishlist" on product page once
Visitor enters email in sign‑up box on landing page
Team lead schedules a 10 minute feedback chat this week
Patient records blood pressure in app each morning
Logistics manager scans delivery barcode at depot exit

If any piece feels fuzzy, slice the behaviour smaller and try again.

Build a baseline before you start

You need to know your starting line.

  1. Check existing logs. Analytics, help‑desk tickets, swipe data.

  2. Run a quick tally. Spend one hour counting the action in real life or on screen.

  3. Ask front‑line staff. They often know the current rate off the top of their heads.

  4. Use a short survey. One question. "How often did you do X last week?"


Even a rough number sets a reference point. Zero still counts.

Freeze scope with a written charter

Fill these eight lines and share them with the team.

  • Actor: Store manager

  • Action: Logs daily stock issue in new app

  • Context: At closing time, back office

  • Success metric: 4 logs per week per manager

  • Baseline: 0 logs today

  • Pilot length: 2 weeks

  • Risks: Wi‑Fi dead zone, time pressure

  • Owner: Ops lead

If you struggle with any field, narrow the behaviour again.

Pick one metric, answer one question

More numbers feel safe, yet they blur the lesson.

Stick to a single outcome that proves progress. Here is how to choose it.

How to pick the right metric

  1. Directly matches the action
    Count the behaviour itself, not a hint of it.

    • Good: Number of patients who record blood pressure each morning.

    • Weak: Time spent in the health app.

  2. Binary or simple count
    Aim for yes or no, or a clear tally. Complex ratios invite debate.

  3. Easy to collect inside the pilot window
    If data arrive after the pilot closes, you learn nothing in time to act.

  4. Baseline available
    You must know “before” to judge “after.” If the baseline is unknown, gather it first.

  5. Tied to the pilot question
    Ask, “If this moves, do we learn what we need?” If not, you have the wrong metric.

Things to look out for

Final checklist before you press go
  1. Actor clear

  2. Action observable in one step

  3. Context fixed and realistic

  4. Outcome countable

  5. Baseline captured

  6. Single metric agreed

  7. Risks noted

  8. Owner named with authority

Other guides you may find useful