Brand Loyalty and Customer Engagement

How to Personalise a Customer Loyalty Programme

Learn how to personalise a loyalty programme using member data, segments, rewards, triggers, consent, testing and measurement.

Δ°lkem Erul Δ°lkem Erul β€’ Published β€’ Updated β€’ 17 min read
How to Personalise a Customer Loyalty Programme

Most loyalty programmes are not short of personalisation ideas. They are short of a structure that makes those ideas measurable, affordable and comprehensible to the member.

This is an implementation guide. It covers assessing what you have, identifying usable member data, designing segments and rewards, building lifecycle triggers, handling tiers and milestones, coordinating channels, managing consent, testing properly, and measuring what the programme actually does.

For why personalisation can affect loyalty at all, and where it backfires, see how personalisation affects brand loyalty. For what companies have built, see personalisation and loyalty examples.

1. Assess what you already have

Before designing anything, answer four questions honestly.

Does the member understand the scheme? If they cannot explain how they earn and what they get, they will not engage with it. They will not try to work it out either. They will simply ignore it, and every additional rule you add makes ignoring more likely.

What proportion of transactions are identified? A programme that captures half your transactions is measuring half your business. Woolworths, as one publicly documented example, reports a scan rate defined as the proportion of transactions and a tag rate defined as the proportion of sales, alongside active membership defined as registered members who scanned in the previous twelve months (1). Whether or not you copy those definitions, having them written down is the point.

What does the programme cost? Reward funding, technology, operational effort and the liability sitting on the balance sheet. More on the last one below.

What do you actually know about members that you could act on? Usually less than the roadmap assumes.

2. Identify usable member data

Three categories, in descending order of reliability.

Declared data. What the member told you: preferences, household composition, category interests. Small, unambiguous and legible to the customer, which is why it holds up.

Transactional data. What they bought, when, how much, through which channel. The most decision-useful data most programmes hold.

Behavioural data. Browsing, dwell time, clicks. Abundant, and less stable in meaning than teams assume, because the same event can represent several intentions. Usable, but it has to be validated against an outcome rather than accepted because the segment looks plausible.

Two structural gaps are worth naming before you build.

The first is identity. If a member buying through a channel they have not used before gets stored as a new person, every downstream number is counting something other than what it claims. Fix that before you buy a decisioning engine.

The second is the marketplace blind spot. A retailer selling the same products through marketplaces has almost no information about who bought there, and that same customer may be buying from competitors too. Programmes built on a partial view tend to drift toward optimising revenue volume, which is how they end up running ad-hoc incentives instead of a coherent scheme.

3. Define segments that map to actions

A segment is only useful if a different decision follows from it. If two segments receive the same treatment, they are one segment.

Build them from things that exist before treatment: tenure, purchase frequency, category breadth, tier, channel mix, realised contribution margin. Not from things that happen afterwards, like whether someone redeemed.

And check the segment definition against the data rather than against the brand’s self-image. A luxury house I worked with was confident its digital target was women aged 40 and over. That was the ideal customer profile and it appeared in every deck. When we examined who actually bought on the website, the answer was women between 27 and 40 and men between 25 and 35. The 40-plus customer was real, she was just the boutique clientele, and the profile had been carried across from the offline heritage channel to a digital business with a different audience.

The customer in the loyalty deck and the customer in the data are frequently different people.

4. Design rewards and benefits

There is no universal ranking of points, cashback and experiential rewards, and anyone who tells you otherwise is generalising from one programme.

The research supports a narrower claim: response varies with involvement, timing and segment. One experimental study found direct rewards preferable in high-involvement situations and immediate rewards favoured in lower-involvement settings (2). Work modelling tier and frequency components found distinct customer segments responding to each rather than one universally superior mechanism (3).

Treat reward type as a design choice to be tested, not a doctrine.

On expiry. Research combining programme data with a laboratory experiment found that finite expiry could increase purchasing, but only where customers had enough flexibility to adapt their buying, with prior programme usage and multi-store shopping identified as relevant sources of that flexibility (4). Expiry is not a single average effect. Test it by measuring acceleration before expiry, points redeemed, points allowed to lapse, subsequent purchasing, and complaints and attrition, separately.

On the incentive that actually works. Across a decade of watching loyalty structures, the businesses that got what they wanted from them were consistently the ones billing members monthly. A subscription creates an incentive no points scheme replicates: the member has already committed money, so buying from you rather than a competitor feels like using something they own. Most businesses cannot adopt that model. It is still the right benchmark for asking what genuine incentive your structure creates.

5. Design lifecycle triggers

A trigger should reflect the member’s actual lifecycle, not a calendar interval.

The useful ones are: enrolment and first-value moment, first repeat purchase, replenishment window based on observed repurchase intervals, approach to a threshold, milestone achievement, benefit expiry, declining activity relative to that member’s own baseline, and service failure.

For the campaign design behind each of these, with signals, decisions, evaluation metrics and risks, see our guide to personalised engagement tactics.

Two rules govern all of them. Suppress on unresolved service issues, because a promotional message during an open complaint does more damage than the campaign is worth. And define the eligibility rule before the trigger fires, not after.

6. Handle tiers and milestones carefully

Tier mechanics are the part of loyalty design most likely to produce an effect opposite to the one intended, and the evidence here is unusually clear.

A field experiment following 95,532 hotel customers before, during and after a promotion found that goal attainment significantly increased subsequent purchasing while goal failure significantly reduced it. Failure had the largest effect among higher-status customers; success had the largest effect among lower-status customers (5).

Read that carefully. Setting a stretching threshold and missing it makes your best members buy less afterwards. The downside is concentrated precisely where you can least afford it.

Demotion is worse. Research found that demoted customers reported lower loyalty than customers who had never held preferred status at all, with a field study using proprietary sales data supporting the effect (6). Taking status away leaves someone worse off than never having given it.

Success can compound. Frequent-flyer data and laboratory evidence indicated that successfully attaining a reward could increase effort on a subsequent attempt, most evident where the goal was sufficiently challenging (7). Which means the measurement window has to extend at least into the next qualification cycle. Stopping at the reward transaction misses the effect entirely.

Practical implications: model the downside among members who fail, not just the uplift among achievers; test grace periods, qualification extensions and partial rather than total benefit withdrawal; and treat demotion as an intervention requiring measurement rather than an administrative event.

One further note. A 2024 study proposed a nonlinear reward gradient, with motivation strengthening after an intermediate point in progress (8). But I could not find a direct replication of the original goal-gradient finding in operational loyalty programmes, so do not build a business case on progress mechanics alone being reliably motivating.

7. Coordinate channels

A member should not receive four uncoordinated versions of the same decision because four teams own four channels.

Every proposed message should pass a contact-policy decision before delivery, with a stated priority order: essential service, then active recovery, then lifecycle support, then general promotion.

Channel timing is worth more attention than it usually gets. In 2017, working with a consumer-electronics retailer, I went through several analytics accounts and found the same pattern repeatedly: desktop traffic peaked during working hours while mobile traffic peaked between five and eight in the evening. We moved email and desktop web push to mornings, and app push and mobile web push to around six. Click-through rates improved dramatically.

The specific windows are market and category dependent, and they will have moved since 2017. The method has not: split send times by the device that receives them, and check your own data rather than inheriting a best-practice send time from a webinar.

8. Set frequency and suppression rules

Frequency caps should operate across channels, not within isolated channel teams. Excessive relational communication has been associated with lower repurchase, with multiple channels intensifying the effect, so contact pressure should be treated as an experimental variable rather than a fixed campaign rule (9). The evidence behind that is discussed in the brand loyalty pillar.

Set the cap across channels rather than per channel. Suppress after purchase, during unresolved service issues, and after non-response thresholds. Measure incremental response per additional contact rather than total attributed conversions, because attributed conversions will keep rising while incremental response falls.

Loyalty data use is not one purpose, and treating it as one is a common compliance shortcut that does not survive examination.

The ICO’s legitimate-interests guidance works through a loyalty-scheme example distinguishing three purposes: calculating points and sending vouchers as the core scheme service; analysing purchase history to provide targeted discounts; and analysing customer behaviour for broader business improvement. It characterises targeted discounts as direct marketing rather than core service, with an accompanying right to object (10).

Map each purpose separately: data used, lawful basis, recipients, retention and individual rights. A single statement that processing is necessary for the loyalty programme will not describe all of them accurately.

On consent, the ICO’s position is that valid consent means genuine choice and control, specific and informed, and easy to withdraw. Offering a loyalty incentive is not automatically invalidating, provided refusal does not create an unfair penalty and genuine choice remains (11). Note that this guidance is marked as under review following the Data (Use and Access) Act 2025.

Large-scale profiling, data matching and certain behaviour or location tracking appear among processing operations that may require a DPIA, with loyalty schemes cited as an example in the tracking section. That does not mean every programme automatically needs one; the risk has to be assessed against the nature, scope and context of your processing (12).

On pricing presentation. European Commission guidance on Article 6a of the Price Indication Directive distinguishes traditional customer-specific loyalty arrangements and genuinely personalised offers, which generally fall outside it, from loyalty promotions announced or made broadly available to members, which can fall within it and its prior-price requirements (13). The practical test is not whether you label something a loyalty price. It is whether the offer is an ongoing customer-specific benefit or effectively a publicly promoted price reduction gated by membership.

The CMA’s 2024 review of UK grocery loyalty pricing found that the overwhelming majority of loyalty promotions represented genuine savings, while a substantial minority of shoppers still considered member-only pricing unfair (14). Operationally, that means fairness perception needs managing separately from pricing integrity.

10. Understand what redemption does to the balance sheet

Marketers routinely treat points as a cost recorded when someone redeems. That is not how the accounting works, and the gap causes real friction with finance.

ℹ️

Finance note: how reward redemption affects revenue recognition

Under IFRS 15, loyalty points are a customer option. Where that option gives a material right the customer would not have received without the original purchase, it is a separate performance obligation, so part of the original transaction price is held as a contract liability rather than recognised immediately as revenue (15). Breakage, the portion expected never to be exercised, is recognised in proportion to the pattern of rights exercised where the entity expects to be entitled to it, and only when exercise becomes remote where it does not (15). The standard’s illustrative example works the arithmetic through in full (16).

The operational consequence is what matters for programme design. A campaign that materially increases expected redemption can change revenue timing and the liability balance even when total customer spend does not move. An expiry or devaluation intended to cut reward cost changes expected breakage, member behaviour and revenue recognition simultaneously.

Tell your finance team before you launch either, not after.

11. Test personalised against standard

Loyalty is a field where the observational evidence is systematically flattering, and three studies explain why.

A grocery study investigating whether member value reflected programme effects or self-selection found that heavier and more frequent customers enrolled earlier, and that behavioural changes were small and eroded within six to nine months (17).

A longitudinal convenience-store study found that heavy buyers were most likely to qualify for and claim rewards but did not change their purchase behaviour, while low and moderate buyers gradually increased purchasing and loyalty (18).

Research on five vendors in a multi-vendor programme found low cardholder responsiveness, no additional impact from joint over individual promotions, and no cross-vendor spillover (19).

None of these prove loyalty programmes fail. They prove that member spend, redemption rates and member-versus-non-member comparisons are not evidence of incremental return, because the members were different people before they joined.

The design that works: randomise an intervention among customers already eligible for the same programme. Assign at member or household level, stratify on pre-treatment attributes, and use a persistent control identifier suppressed across every channel including till and contact centre. Report the intention-to-treat difference. Use contribution margin after reward funding, fulfilment and contact costs as the primary commercial outcome. Report reward liability movement separately from operating margin. And do not use redeemers versus non-redeemers as the causal comparison, because redemption is a post-treatment behaviour.

What testing added, and what it did not

At one of the biggest menswear brands in TΓΌrkiye, campaigns had been reporting conversion uplift in the 7 to 12% range. After we introduced structured A/B testing and gamified mechanics, wheel-of-fortune scenarios and similar, reported uplift reached around 25%. It remains the largest before-and-after I personally saw in ten years.

Two things were happening, and only one of them was the gamification. The mechanics were genuinely more engaging. But the A/B testing meant we were finally comparing variants against each other rather than against last quarter, which changed what the number meant as much as it changed the number itself.

I offer that as an illustration from one brand in one category, not a benchmark. The transferable part is the testing discipline, not the 25%.

12. Measure the programme

Four layers, reported separately.

Adoption. Enrolment, identified transaction share, active membership under a stated definition.

Engagement. Reward views, claims, redemption rate, tier progression, benefit utilisation.

Commercial. Incremental contribution margin against a holdout, incremental purchases, retention within matched cohorts.

Guardrails. Opt-outs, complaints, contact frequency, reward liability, breakage, attrition among members who fail a threshold or are demoted.

For formulas, denominators and interpretation warnings on any individual metric, see our personalisation KPIs reference.

13. Own it operationally

Two organisational failures recur in every programme I worked on.

The first is that leadership altitude changes what looks important. Climbing from account management to running a region, I finally understood why forecasting discipline and CRM hygiene mattered, things that had been invisible to me before. I also started deprioritising process fixes on effort-optimisation grounds, and those fixes were urgent every single day to the people living with them. A loyalty programme has the same structure: the tier rule that is a minor backlog item to a director is the thing a service agent explains forty times a day.

The second is the complaint I heard in nearly every quarterly review across a decade, in some form: our team spends too much time on the tool. I never fully solved it, and part of the reason is that it is not always a straightforward operational statement. People say it to position their own workload. That does not make it false, but it does mean the fix is rarely the one being asked for.

Name three owners for the programme: who owns the scheme rules, who owns the data behind them, and who is accountable for the commercial result. When those collapse into one person, the reporting gets optimistic.

Implementation checklist

  1. A member can explain how they earn and what they get.
  2. Identity is reconciled across channels before decisioning is bought.
  3. Each segment leads to a different action.
  4. Reward type is a tested choice, not a doctrine.
  5. Thresholds are attainable, with a designed path for members who miss them.
  6. Demotion is treated as an intervention and measured.
  7. Triggers reflect the member lifecycle, not the calendar.
  8. Frequency caps operate across channels, not per channel.
  9. Each data purpose is separately mapped to a lawful basis.
  10. Finance knows what a redemption change does to the liability.
  11. A randomised holdout exists among eligible members.
  12. Contribution margin after reward cost is the primary commercial metric.
  13. Three owners are named: rules, data, result.

Frequently Asked Questions

Randomise an intervention among customers already eligible for the same programme and compare assigned groups, using contribution margin after reward funding and contact costs. Comparing members against non-members does not work, because research found that heavier and more frequent customers enrol earlier, with behavioural changes that were small and eroded within six to nine months. A separate longitudinal study found heavy buyers redeemed most while not changing their purchasing at all. Member value is largely evidence about who joined, not about what the programme did.

It depends on whether your members have the flexibility to respond. Research combining programme data with a laboratory experiment found finite expiry could increase purchasing, but only where customers had enough flexibility to adapt their buying, with prior programme usage and multi-store shopping identified as relevant sources. Test it as several outcomes rather than one: acceleration before expiry, points redeemed, points allowed to lapse, subsequent purchasing, and complaints and attrition. Expiry also changes expected breakage, which affects revenue recognition, so involve finance before changing the policy.

They buy less afterwards, and the effect concentrates among your best members. A field experiment following 95,532 hotel customers found that goal attainment significantly increased subsequent purchasing while goal failure significantly reduced it, with failure having the largest effect among higher-status customers. Demotion is worse still: research found demoted customers reported lower loyalty than customers who had never held preferred status. Design attainable thresholds, test grace periods and qualification extensions, and treat demotion as an intervention that requires measurement.

Because points are not simply a cost recorded at redemption. Under IFRS 15, where points give a material right the customer would not have received otherwise, they are a separate performance obligation, so part of the original transaction price is held as a contract liability rather than recognised as revenue. Breakage, the portion expected never to be redeemed, is recognised in proportion to the pattern of rights exercised where the entity expects to be entitled to it. A campaign that raises expected redemption can therefore change revenue timing and the liability balance even if total spend does not move.

Less often than most programmes do, with a cap that operates across channels rather than per channel. Research covering personalised relational communications and repurchase over three years found an inverted relationship, where communication helped up to an ideal level and customers reacted negatively beyond it, with multiple channels intensifying the effect. No universal frequency applies across categories. Measure incremental response per additional contact against a holdout, because attributed conversions keep rising while incremental response falls.

References

  1. Woolworths Group. 2024 Annual Report. 28 August 2024. https://www.woolworthsgroup.com.au/content/dam/wwg/investors/reports/f24/f24/Woolworths%20Group%202024%20Annual%20Report.pdf
  2. Yi, Y. and Jeon, H. Effects of loyalty programs on value perception, program loyalty, and brand loyalty. Journal of the Academy of Marketing Science, June 2003. https://link.springer.com/article/10.1177/0092070303031003002
  3. Kopalle, P. K. et al. The Joint Sales Impact of Frequency Reward and Customer Tier Components of Loyalty Programs. Marketing Science, 24 January 2012. https://pubsonline.informs.org/doi/10.1287/mksc.1110.0687
  4. Breugelmans, E. and Liu, Y. The effect of loyalty program expiration policy on consumer behavior. Marketing Letters, 21 September 2017. https://link.springer.com/article/10.1007/s11002-017-9438-1
  5. Wang, Y., Lewis, M., Cryder, C. and Sprigg, J. Enduring Effects of Goal Achievement and Failure Within Customer Loyalty Programs: A Large-Scale Field Experiment. Marketing Science, 35(4), 565-575, 2016. https://pubsonline.informs.org/doi/10.1287/mksc.2015.0966
  6. Wagner, T., Hennig-Thurau, T. and Rudolph, T. Does Customer Demotion Jeopardize Loyalty? Journal of Marketing, 1 May 2009. https://journals.sagepub.com/doi/10.1509/jmkg.73.3.069
  7. Drèze, X. and Nunes, J. C. Recurring Goals and Learning: The Impact of Successful Reward Attainment on Purchase Behavior. Journal of Marketing Research, 1 April 2011. https://journals.sagepub.com/doi/10.1509/jmkr.48.2.268
  8. Ko, W. L. and Song, T. H. Nonlinear Reward Gradient Behavior in Customer Reward and Loyalty Programs. Journal of Hospitality and Tourism Research, 9 February 2024. https://journals.sagepub.com/doi/10.1177/10963480231226083
  9. Godfrey, A., Seiders, K. and Voss, G. B. Enough Is Enough! The Fine Line in Executing Multichannel Relational Communication. Journal of Marketing, 1 July 2011. https://journals.sagepub.com/doi/10.1509/jmkg.75.4.94
  10. Information Commissioner’s Office. When can we rely on legitimate interests? Accessed 27 July 2026. https://ico.org.uk/for-organisations/uk-gdpr-guidance-and-resources/lawful-basis/legitimate-interests/when-can-we-rely-on-legitimate-interests/
  11. Information Commissioner’s Office. What is valid consent? Accessed 27 July 2026. https://ico.org.uk/for-organisations/uk-gdpr-guidance-and-resources/lawful-basis/consent/what-is-valid-consent/
  12. Information Commissioner’s Office. Examples of processing likely to result in high risk. Accessed 27 July 2026. https://ico.org.uk/for-organisations/uk-gdpr-guidance-and-resources/accountability-and-governance/data-protection-impact-assessments-dpias/examples-of-processing-likely-to-result-in-high-risk/
  13. European Commission. Guidance on the interpretation and application of Article 6a of Directive 98/6/EC. 29 December 2021. https://eur-lex.europa.eu/legal-content/EN/TXT/HTML/?uri=CELEX%3A52021XC1229%2806%29
  14. Competition and Markets Authority. Review of loyalty pricing in the groceries sector: executive summary. 27 November 2024. https://www.gov.uk/government/publications/review-of-loyalty-pricing-in-the-groceries-sector/executive-summary
  15. IFRS Foundation. IFRS 15 Revenue from Contracts with Customers. 2025 issued edition. https://www.ifrs.org/content/dam/ifrs/publications/html-standards/english/2025/issued/ifrs15.html
  16. IFRS Foundation. IFRS 15 Revenue from Contracts with Customers: Illustrative Examples. 2025 issued edition. https://www.ifrs.org/content/dam/ifrs/publications/html-standards/english/2025/issued/ifrs15-ie.html
  17. Leenheer, J. et al. Grocery retail loyalty program effects: self-selection or purchase behavior change? Journal of the Academy of Marketing Science, 20 November 2008. https://link.springer.com/article/10.1007/s11747-008-0123-z
  18. Liu, Y. The Long-Term Impact of Loyalty Programs on Consumer Purchase Behavior and Loyalty. Journal of Marketing, October 2007. https://journals.sagepub.com/doi/10.1509/jmkg.71.4.019
  19. Dorotic, M., Fok, D., Verhoef, P. C. and Bijmolt, T. H. A. Do vendors benefit from promotions in a multi-vendor loyalty program? Marketing Letters, 22(4), 341-356, 2011. https://link.springer.com/article/10.1007/s11002-010-9128-8
Δ°lkem Erul

Written by

Δ°lkem Erul

Contributor

I have over nine years of experience in digital marketing, account management, and B2C loyalty. I've helped global brands grow, and now, as a co-founder of Herm.io, I work on smarter, safer shopping experiences for consumers.

More from Δ°lkem

Related Articles