Retailers have spent years trying to make customers more loyal. Points, tiers, birthday rewards, member pricing, early access, free delivery and ever more elaborate earn-and-burn mechanics have become standard parts of the commercial toolkit. Many of these programmes are well designed, and some are genuinely valuable.
Yet the underlying customer relationship often remains surprisingly weak. A customer can accumulate points with a retailer for years and still switch the moment another business offers a better product, easier experience or more relevant service. The programme may be working exactly as designed while loyalty itself remains largely transactional.
That distinction matters because the industry has tended to treat loyalty as something that can be manufactured through mechanics. I think the more durable version is much simpler: customers stay close to businesses that continue to understand what matters to them, remove unnecessary effort and make the relationship feel increasingly useful over time.
The problem, in other words, is not that retailers lack loyalty programmes. It is that too many of those programmes reward transactions without becoming meaningfully more relevant to the customer.
Loyalty mechanics are not the same as loyalty
A loyalty programme can change behaviour without creating much attachment. Give customers enough points, discounts or exclusive access and they may consolidate spend because the economics make sense. That is useful, but it should not be confused with a stronger relationship.
The distinction becomes obvious when the incentive disappears. If a customer immediately defects when another retailer offers a better promotion, the programme may have created a switching cost rather than loyalty. There is nothing inherently wrong with that, but leaders should be clear about what they are buying.
True loyalty is usually built through accumulated evidence. The retailer consistently has the right product, recognises the customer across interactions, resolves problems well, uses information intelligently and does not make the customer repeat things the business should already know. Over time, those moments create confidence that returning is easier and more rewarding than starting again elsewhere.
This is why some customers remain attached to brands with fairly ordinary loyalty mechanics while others churn despite rich rewards. The formal programme is only one layer of the relationship. Product, service, convenience, trust and recognition usually carry far more emotional weight.
The commercial implication is important. If the loyalty strategy begins and ends with points economics, the business is optimising a mechanism rather than the relationship itself.
Recognition without relevance quickly becomes noise
Most retailers have become much better at recognising customers. They can connect purchases, browsing behaviour, campaign engagement, loyalty status, category preference and increasingly store interactions into a richer profile than was possible a decade ago.
Recognition, however, is only useful if it changes the experience in a way the customer values. Knowing that someone bought a pair of running shoes is not the same as understanding whether they are training for a marathon, replacing an old pair, buying a gift or simply responding to a promotion.
Personalisation often becomes superficial at precisely this point. The business has more data, so it sends more targeted messages. The customer receives something that is technically personalised but not particularly useful.
The problem is not lack of intelligence. It is often the absence of a clear decision about what relevance should mean at a particular moment.
A retailer might know that a customer has bought from the same beauty category several times. Relevance could mean reminding them when replenishment is likely, recognising a preference for a particular formulation, surfacing a new product that genuinely fits, or ensuring an associate can continue the conversation in store without starting from zero. Those actions are different from simply inserting the customer’s name into another campaign.
The best personalisation is often less visible than marketers expect. It reduces irrelevant choice, prevents unnecessary communication, anticipates a need or removes a piece of friction. The customer may never describe that as personalisation, but they experience the relationship as easier.
That is a higher standard than targeting, and it changes what the customer-data strategy is for. More attributes in a profile do not create value unless they improve what the business chooses to show, say, recommend, prioritise or suppress.
Loyalty becomes more powerful when it shapes the whole experience
One reason loyalty programmes underperform is structural. They frequently sit inside CRM or marketing while the moments that determine customer loyalty sit everywhere else.
A customer may have top-tier status and still face an unhelpful return experience. They may receive beautifully personalised communication and then discover that store staff cannot see the same history. They may be recognised online and treated as a stranger by customer service.
From the organisation’s perspective, these are separate processes. From the customer’s perspective, they are one relationship.
At that point, loyalty becomes an operating-model issue rather than a marketing programme. The value of customer knowledge depends on whether merchandising, service, stores, digital, supply chain and commercial teams can use it coherently.
That does not require every function to become part of the loyalty team. It requires the organisation to agree which customer signals matter and how those signals should influence decisions outside marketing.
For example, a high-value customer experiencing repeated fulfilment failures may deserve a different service response before receiving another promotional offer. A customer showing strong full-price behaviour may not need the same discount cadence as someone who shops almost entirely during markdown. A store associate serving a loyal online customer should not need to ask questions the business already knows the answer to.
These are small examples, but together they define whether the programme feels like a relationship or a database. The challenge is that cross-functional relevance is harder to manage than campaign performance.
It cuts across budgets, ownership and KPIs. Yet that is precisely where the most valuable loyalty opportunities usually sit.
The economics should reward future value, not just recent spend
Most loyalty models are built around historical behaviour. Customers earn status because they spent a certain amount, visited a certain number of times or accumulated enough points. That makes sense because past behaviour is observable and easy to administer.
The limitation is that historical spend is not the same as future value. A customer who spent heavily during one promotional period may be less valuable than a customer with lower current spend but strong frequency, full-price behaviour and growing engagement across categories.
A long-standing customer with declining activity may need a different intervention from a newly acquired customer showing unusually strong repeat behaviour. Customer economics therefore belongs inside loyalty design.
The useful question is not simply how much the customer spent. It is how the relationship is developing: frequency, margin, retention, category breadth, service cost, discount dependency and the probability that the customer will remain valuable over time.
That does not mean reducing every customer to a lifetime-value score. Models are imperfect, future value is uncertain and customers do not behave according to spreadsheets. It does mean the business should avoid rewarding behaviour that looks attractive in the short term while quietly weakening the economics of the relationship.
Heavy promotional dependency is a good example. A programme can increase transaction frequency by repeatedly offering incentives and still train customers to wait for the next one. The engagement dashboard improves while margin and willingness to pay gradually deteriorate.
A more mature loyalty model knows when to reward, when to recognise, when to serve and when to do nothing. Restraint is part of relevance too.
AI raises the standard for relevance rather than solving it
AI will make it much easier to personalise at scale. Models can interpret broader customer histories, predict likely needs, generate content, recommend products and increasingly decide which action should happen next.
That capability is valuable, but it does not eliminate the strategic problem. In fact, it raises the standard because businesses will be able to produce far more personalised activity than customers ever asked to receive.
The danger is obvious. A retailer can become extremely efficient at sending the wrong thing to the right person.
The better use of AI is not simply to increase message precision. It is to improve judgment around relevance: whether to communicate, which need matters most, which channel fits the context, when a service intervention is more important than an offer, and when the business should leave the customer alone.
That requires data, but it also requires clear commercial rules. The model needs to understand what the company is optimising for. If the objective is only short-term conversion, personalisation will drift toward whatever produces the next transaction, even when that damages the longer relationship.
This is why loyalty, data and AI increasingly belong in the same commercial conversation. The technology can interpret more signals than any team could manually, but leadership still has to decide what a good customer relationship looks like and which outcomes matter.
A business that gets that wrong will simply scale irrelevance more efficiently. The technology may be sophisticated; the customer experience will still feel tone-deaf.
The strongest loyalty is often almost invisible
The irony is that the most effective loyalty experiences do not always feel like loyalty programmes. They feel like the retailer remembered, the recommendation made sense, the service team already understood the issue, and the associate knew enough to be helpful without becoming intrusive.
The product was available when expected and the benefit appeared at the right moment rather than buried inside another campaign. None of these moments is dramatic on its own, but together they make the relationship feel easier and more considered.
Customers gradually learn which businesses make their lives easier and which ones consume unnecessary effort. They remember who handles a problem well, who repeatedly sends irrelevant offers and who seems to understand the difference between recognising a customer and exploiting their data.
That accumulated judgment is difficult to manufacture with points alone. For leaders, this changes the ambition of loyalty.
The programme still matters, and the economics of rewards still need to work, but the larger goal is to make the relationship progressively more useful as the business learns more about the customer. That is what relevance looks like at scale.
Retail does not need to abandon loyalty mechanics. It needs to stop asking those mechanics to do the entire job.
The strongest customer relationships are built when recognition improves the experience, data improves decisions and loyalty becomes something the whole enterprise helps deliver. At that point, the points balance becomes useful, but secondary.
Customers rarely stay because a programme tells them they are loyal. They stay because continuing the relationship still feels worth it.