Gamification

Design game elements into a non-game context so they serve a real behavioral objective — not "add points and badges". The evidence says gamification works modestly and contextually: small effect sizes that shrink further under methodological rigor, and benefits that decay with novelty. So every mechanic must earn its place by mapping to a motivational need.

Two reference files, loaded on demand:

• mechanics.md — per-mechanic catalog: evidence, when to use, when it backfires.
• frameworks.md — the named frameworks (Octalysis, 6D, HEXAD, Hooked, RECIPE) and how they fit together.

Motivation core

Four load-bearing facts. Every design decision below leans on at least one.

1. SDT — three needs drive intrinsic motivation (Deci & Ryan; PENS): competence (mastery, granular feedback, tuned difficulty), autonomy (meaningful choice, no coercion), relatedness (teams, community). Mechanics are only delivery vehicles for these needs.
2. Overjustification (Deci, Koestner & Ryan meta-analysis): expected tangible rewards for an already-interesting activity reduce intrinsic motivation. Informational feedback (progress, mastery signals, praise) and unexpected rewards do not. Never bolt rewards onto what users already love.
3. Flow channel (Csikszentmihalyi): challenge must track skill — too far above → anxiety, too far below → boredom. Onboarding starts below current skill and ramps; let users pick their own challenge level where possible.
4. B = MAP (Fogg): a behavior fires only when Motivation, Ability, and a Prompt converge. If the missing piece is ability, simplify the action instead of gamifying it.

Design process

Follow the steps in order. Mechanics are step 5, not step 1 — mechanics-first copying of points-badges-leaderboards is the failure mode behind Gartner's famous 80%-failure prediction (a practitioner prediction, not a study, but its rationale matches the empirical failure literature).

1. Define the objective

Name the business or behavioral outcome and a measurable target that is not an engagement vanity metric (DAU, session length). "Users complete their first project within a week", not "users open the app more".

Done when: the objective is written down with a metric a stakeholder outside the feature would recognize as valuable.

2. Delineate target behaviors

List the specific, observable user actions that produce the objective. Run each through B=MAP: is the blocker motivation, ability, or prompt? Behaviors blocked on ability get simplified, not gamified; behaviors blocked on prompt get a trigger, not a reward.

Done when: every behavior has a named blocker (M, A, or P) and only the motivation-blocked ones proceed to gamification.

3. Describe the players

Segment the audience (HEXAD types in frameworks.md — most users respond to meaning, autonomy, and social connection, not competition) and assess baseline intrinsic interest. High existing interest triggers the overjustification rule: informational feedback only, no tangible rewards.

Done when: you can state which segments exist, what each responds to, and whether the tangible-reward path is open or closed.

4. Design activity loops

Two loops, both explicit:

• Engagement loop (short): action → feedback → next action. Feedback is immediate and informational.
• Progression loop (long): onboarding → mastery → endgame. Keep the difficulty inside the flow channel across the whole arc.

Plan past the honeymoon: novelty decay is a recurring finding across the literature, and a post-reward reset (effort dips right after a goal is reached) is documented — so design what a month-three user does, not just a day-one user.

Done when: both loops are sketched, and the post-novelty / post-reward phases each have a designed answer.

5. Choose mechanics from motivation, not fashion

For each candidate mechanic, write the SDT need (or Octalysis drive) it serves; a mechanic with no mapping is decoration and gets cut. Cross-cutting defaults that survive the literature:

• Informational feedback over controlling rewards.
• Cooperation over global competition.
• White-hat drives (epic meaning, accomplishment, creative empowerment) as the foundation; black-hat drives (scarcity, unpredictability, loss) only as short accents, never the engine.

The per-mechanic rules — evidence, use-when, backfires-when, and the non-negotiables like relative leaderboards, streak forgiveness, and goal chaining — live in mechanics.md; apply them from there.

Done when: every kept mechanic has a written need→mechanic mapping, passes its mechanics.md entry, and every cut is deliberate.

6. Ethics gate

Run before build, not after launch:

• Alignment test: does the designer's incentive point the same way as the user's genuine goal? (Streaks that help you learn: aligned. Spend mechanics dressed as play: not.)
• Manipulation Matrix (Eyal): build habit loops only where you would use the product yourself and it materially improves users' lives.
• Black-hat dosage: retention running primarily on FOMO/loss/scarcity fails this gate.
• Long-term change test (Nicholson): if the goal is durable behavior change, avoid rewards entirely — design for reflection, choice, and personal meaning instead (RECIPE, in frameworks.md).
• Regulatory floor: the EU DSA bans manipulative interface design (dark patterns); loot-box-like mechanics and anything touching minors face escalating regulatory scrutiny.

Done when: each check has an explicit pass, or the design changed.

7. Deploy, measure, iterate

A/B against the step-1 objective. Watch specifically for: metric gaming (quantity over quality), bottom-of-leaderboard churn, streak-cliff abandonment, and the novelty-decay curve. Expect small effects and compound them; a flat result against the real objective beats a big win on a vanity metric.

Done when: the objective metric moved, or the losing mechanic was removed rather than tuned harder.

Anti-patterns

The process already blocks most documented failures: rewarding the already-interested (step 3), novelty-only design (step 4), global leaderboards and bare streaks (step 5 via mechanics.md), black-hat engines (step 6). Three more to reject on sight: