Skip to content
Gamify It.
Connect

How we know what worked

The measurement playbook

How to tell whether a change worked: what to measure, whether your traffic can answer the question, and how to read a result honestly.

Verified October 24, 2026

Most product changes move metrics less than teams expect, and many early wins fade. So every move Gamify It suggests comes with a way to read it, sized to the traffic you actually have.

When traffic is too low for a reliable A/B test, we say so, recommend bigger swings and label the read as directional. A result that is inconclusive is still a result.

Measurement

The playbook, step by step

The measurement entries in our library, grouped by the question they answer.

  1. 01

    Pick what to measure

    One metric that reflects delivered value, the signals that move it, and the guardrails that keep it honest.

  2. 02

    Know what your traffic can tell you

    Before testing, check what your data supports and how many weeks a readable result would take.

  3. 03

    Run tests you can trust

    Decide the rules before the data arrives, expect most ideas to lose, and check that wins last.

Limits

What we don’t claim

These caveats travel with the frameworks everywhere they appear, on this site and in every answer Gamify It gives in your AI assistant.

  • The Hook Model is a practical design tool, not a validated or complete theory of behavior.Hooked, Introduction
  • Octalysis is a practitioner framework built from observing games, not a validated scientific model.Actionable Gamification Ch.15, pp.399-400
  • Left Brain and Right Brain are a metaphor for extrinsic and intrinsic motivation, not a claim about neuroscience.Actionable Gamification Ch.13, pp.342-343
  • The evidence is correlational until it is tested in your product.
  • Habit metrics need the longer habit horizon (weeks to months) before they can be read.
  • The threshold is a heuristic, not a measured cut-off.
  • Bartle himself doubted his player types outside the settings he studied.

Go deeper

More from the knowledge base

  • Framework · Behavioral science

    Realistic effect sizes for behavioral levers

    what to expect from a single change

    Single behavioral tweaks usually move a metric by low single digits. Across 126 nudge-unit trials (23 million people) the average lift was 1.4 percentage points, about 8% relative, versus 8.7 points in journals; and most program effects fade once the program ends. Plan bundled changes and durability checks.

  • Framework · Behavioral science

    Claims never used as mechanisms

    replication-crisis do-not-use list

    Several popular behavioral claims failed replication or were retracted: willpower as a depleting resource, subtle priming, the sign-at-the-top honesty nudge, the Zeigarnik memory effect, 21-day habits, and choice overload or decoys as universal laws. They are never presented as working mechanisms; the nearest valid lever is offered instead.

  • Framework · Behavioral scienceEvidence grade B. Moderate evidence: solid studies with some limits.

    Streaks and streak repair

    streak design with repair

    People repeat a logged behavior more when the record shows an intact streak than a broken one, even with the same past behavior; the drop after a break shrinks when the streak can be repaired. Match the streak unit to the natural cadence and offer freezes or repair windows. Moderate evidence (B).

Sources

Source IDs: OCT-M03F26INF-K-03HK-T13HK-MEA04HK-MEA11F15HK-MEA03HK-TRG37F07F04F12F13F03GT#68F22CD8CASE-M01CASE-M25

Book content is paraphrased, never quoted at length. Last verified October 24, 2026.

Your product

Find the stage your product is stuck in

Gamify It reads your product, picks the one move that matters at its stage, and sizes the test to your traffic. Connecting is free; a Pro subscription starts it.

Add Gamify It to your assistantTake the 2-minute check-up