Design a Loyalty and Rewards Program

Full method. Works on any model.

You are a senior loyalty and behavioral design strategist. You design rewards programs
using established behavioral science (goal gradient, endowed progress, variable
reinforcement, the habit loop) and you insist on validating every mechanic with the
client's own data rather than treating research as gospel.

Your task: design a loyalty and rewards program for the business described below.

CONTEXT
→ Business and product: {{BUSINESS_AND_PRODUCT}}
→ Primary customer segments and their visit or purchase frequency: {{SEGMENTS_AND_FREQUENCY}}
→ Current program if any, and what is failing: {{CURRENT_PROGRAM}}
→ Business goal for the program: {{GOAL}}  (e.g. increase visit frequency, lift retention, grow basket size)
→ Touchpoints you control: {{TOUCHPOINTS}}  (app, email, receipt, point of sale, SMS, staff)
→ Margin and cost constraints on rewards: {{REWARD_BUDGET}}

If any of these are missing or vague, especially segment frequency, margin constraints,
and the business goal, ask me up to five clarifying questions and wait for my answers
before designing anything. Do not guess the frequency of a typical customer; it decides
whether rewards are achievable.

METHOD (work through every step, show your reasoning briefly, then produce the blueprint)

1. Choose the program structure before any mechanics. Lay out two or three candidate
   structures that fit this business (for example a visit threshold card, a points program,
   status tiers, a paid membership, or a hybrid). For each, give one line on how it serves
   {{GOAL}}, what it costs against {{REWARD_BUDGET}}, and its main behavioral risk. Recommend
   one, say why it beats the others, and build everything below on that structure.

2. Separate the two goals. Rewards pull people into your environment; loyalty is built by
   pairing that entry with a consistently great experience over time. State plainly which
   parts of your design are the incentive and which parts protect the experience, because a
   rewards card alone produces repeat visits driven by convenience, not loyalty that survives
   a price increase.

3. Set the core reward so it passes both mandatory tests:
   → Desirable: worth wanting for this segment (an upgrade, a genuinely good free item), not a
     token discount.
   → Achievable: reachable in a modest number of actions for a TYPICAL customer, not just your
     heaviest users. Model attainability against the median frequency in {{SEGMENTS_AND_FREQUENCY}}.
     If a reward is desirable but unreachable for a segment (the classic airline miles trap for
     infrequent flyers), it has zero motivational value for that segment. Segment reward tiers
     by frequency.

4. Tune reward frequency. For fixed rewards, frequency is the single lever that decides whether
   they work. Too infrequent and there is no reinforcement, so no effect. Too frequent (every
   second or third visit) and the reward stops feeling meaningful and gets expensive. Pick a
   threshold and justify it against segment frequency and {{REWARD_BUDGET}}.

5. Add variable reward on top of the fixed structure. A surprise and delight element (a random
   free upgrade, an unexpected bonus) is most effective layered onto a predictable core reward,
   because unpredictability sustains engagement that a fixed schedule alone lets go stale. Tie
   the variable reward to what the customer actually came for.

6. Engineer the goal gradient. People accelerate effort as they perceive themselves nearing a
   goal, so make the goal and the closing distance visible everywhere:
   → State the goal explicitly (what reward, at what threshold).
   → Show live progress as visually as possible (progress bar, points to go, "you are almost
     there"). Never rely on the customer to track it themselves.
   → Proactively remind when they are close, via the channels in {{TOUCHPOINTS}}, rather than
     hoping they open the app.

7. Apply the endowed progress effect. Never start a customer at zero, which is one of the least
   motivating states. Grant free or prefilled starting progress so the meter reads, for example,
   "you are already 20 percent there." It is fine to enlarge the total goal as long as the
   customer perceives a start above zero and the same real effort remains.

8. Protect PERCEIVED fairness, not just legal fairness. Never wipe out earned progress. Demoting
   a customer and erasing accumulated progress so a member near the threshold restarts at zero
   generates strong backlash even when defensible, because loyal customers feel punished for
   loyalty. Write an explicit rule for how status and progress are handled on downgrade.

9. Design the communication loop. Integrate the program into every touchpoint so progress updates
   during any interaction, and send a message after each purchase that shows updated progress,
   thanks the customer, and asks for feedback.

10. Add counterfactual reflection where natural. Prompting members, positively and never with guilt
    or threat, to reflect on the worse alternatives they would face without your product increases
    loyalty. Suggest one honest place to use it.

11. Set the recovery rule for service failures. When you get something wrong, a single brief,
    direct apology neutralizes the customer's urge to retaliate. No excuses, no long explanations,
    because those undercut it. Include this as a program policy.

OUTPUT FORMAT
1. Structure decision: the candidate structures you considered and why the recommended one wins
   for {{GOAL}} within {{REWARD_BUDGET}}.
2. Program summary (one paragraph: the incentive and the experience it pairs with).
3. Reward structure: fixed core reward, threshold, variable layer, and an attainability check
   for each segment (a small table).
4. Goal gradient and endowed progress mechanics: exact starting progress, progress display, and
   reminder triggers.
5. Communication cadence: message per touchpoint, with timing.
6. Fairness and recovery policies: downgrade handling and the apology rule.
7. Risks and what to test first, with the specific metric each test moves.

SELF CHECK before finishing
→ Verify the core reward is achievable for the TYPICAL customer of each named segment, not only
  heavy users.
→ Verify the meter never starts at zero and progress is visible at every touchpoint.
→ Confirm no mechanic erases earned progress.
→ Failure modes to avoid: rewarding spend in a way that makes rewards unreachable for regular
  customers with small baskets; opaque status tiers that hide distance to reward; bolting on
  points and badges that do not connect to why the customer is there.
→ Flag every assumption you made about frequency, margin, or behavior, and mark it as something to
  validate with the client's own data before launch.
→ If any check fails, revise the design before presenting it. Do not ship the blueprint with a
  known failed check.

For the most capable models. Goal and quality bar up front.

You are a senior loyalty and behavioral design strategist grounded in goal gradient,
endowed progress, variable reinforcement, and the habit loop.

Goal: design a complete, shippable loyalty and rewards program for the business below.
Deliverable: a program blueprint a team can build from, led by a one line verdict on the
single program structure you recommend, with the reasoning and full design after it.

CONTEXT
→ Business and product: {{BUSINESS_AND_PRODUCT}}
→ Segments and visit or purchase frequency: {{SEGMENTS_AND_FREQUENCY}}
→ Current program and what is failing: {{CURRENT_PROGRAM}}
→ Business goal: {{GOAL}}
→ Touchpoints you control: {{TOUCHPOINTS}}
→ Margin and cost constraints: {{REWARD_BUDGET}}

Non negotiable principles:
→ Choose one program structure first (weighing two or three candidates against {{GOAL}} and
  {{REWARD_BUDGET}}), then build everything on it.
→ Separate incentive from experience: rewards pull people in, a consistently great experience
  keeps them; label which parts do which.
→ The core reward must be both desirable and achievable for the TYPICAL customer of each
  segment, modeled against median frequency, not just heavy users. Segment tiers by frequency.
→ Tune fixed reward frequency so it reinforces without becoming meaningless or unaffordable,
  then layer a variable surprise reward on top of the predictable core.
→ Engineer the goal gradient: state the goal, show live progress visually at every touchpoint,
  and remind proactively when a member is close.
→ Never start progress at zero (grant endowed starting progress) and never erase earned progress
  on downgrade; protect perceived fairness, not just legal fairness.
→ Build the communication loop into every touchpoint, and set a recovery policy of one brief
  direct apology on service failure.

Quality bar (excellent output satisfies all of these):
→ The structure decision names the candidates and argues why the winner beats them for {{GOAL}}
  within {{REWARD_BUDGET}}.
→ A per segment table shows the core reward is desirable and achievable at each frequency.
→ Starting progress is above zero, progress is visible everywhere, and no mechanic erases earned
  progress; downgrade and apology policies are written out.
→ A test plan names the specific metric each experiment moves, so nothing ships as unvalidated
  theory.

Boundaries:
→ Do not invent data, frequencies, or statistics; flag every assumption as something to validate
  against the client's own data before launch.
→ Do not pad with generic loyalty advice or bolt on points and badges that ignore why the
  customer is there.
→ If segment frequency, margin constraints, or the business goal is missing, ask one focused
  question before designing rather than guessing.

Lead with the recommended structure and program verdict; put the supporting design below it.

Five lines. Speed over rigor.

Design a loyalty program for {{BUSINESS_AND_PRODUCT}} serving {{SEGMENTS_AND_FREQUENCY}}, aimed at {{GOAL}} within {{REWARD_BUDGET}}.
Pick one structure, set a core reward that is desirable AND achievable for the typical customer of each segment.
Never start progress at zero, show progress at every touchpoint in {{TOUCHPOINTS}}, and never erase earned progress.
Quality bar: the reward must be reachable for a median frequency customer, not just heavy users.
If segment frequency or budget is unclear, ask one question first instead of guessing.