Personalisation does not create loyalty. It creates the conditions in which loyalty becomes more likely, and it can just as easily destroy them.
That is a less satisfying opening than the industry usually offers, but it is what the evidence supports. Most research in this field measures attention, click-through intention, adoption intention, satisfaction or perceived fairness. Very little of it measures whether people actually came back and bought again. The gap between those two things is where most confident claims about personalisation and loyalty live.
This article explains the mechanisms: why relevance and recognition can strengthen a relationship, how trust and autonomy sit underneath that, and the specific conditions under which personalisation reverses and damages loyalty instead.
It does not cover implementation. For programme design, segmentation, rewards and operating processes, see our guide to personalising a customer loyalty programme. For what these principles look like in practice, see our personalisation and loyalty examples.
What we are actually talking about
Brand loyalty has two components that behave differently. Attitudinal loyalty is preference and commitment. Behavioural loyalty is repeat purchasing. They correlate imperfectly, and a great deal of marketing writing treats them as one thing.
Personalisation is adapting content, offers or experiences using information about an individual. That information can be declared, observed or inferred, and which of the three you use turns out to matter enormously for whether the personalisation helps or harms.
The trap in most discussion of this subject is treating satisfaction as equivalent to loyalty. A major review of the satisfaction-loyalty relationship concluded that satisfaction is positively associated with both attitudinal and behavioural loyalty, but that the variance explained by satisfaction alone is rather small (1). Habit, price, availability, switching costs, competition and prior relationship strength all intervene. A personalised experience can improve how someone feels without changing what they do.
What the consumer statistics really say
Two figures dominate this topic and both are commonly misstated.
The 80% figure. Epsilon’s widely cited research reported that 80% of respondents said they would be more likely to do business with a company offering personalised experiences. Not more likely to purchase. It measured stated likelihood in a 2017 online survey of 1,000 consumers aged 18 to 64 (2).
The more interesting part is what Epsilon did next. It linked the survey responses to transactional data from its own cooperative database and reported that consumers who find personalised experiences very appealing were ten times more likely to be among a brand’s most valuable customers (2).
That is a stronger claim than the headline, and it still does not establish what it appears to. It is an association between an attitude and existing customer value, measured at one moment. The most valuable customers are precisely the people who receive the most personalisation, have the longest relationship history and have the most to gain from relevance. Finding that they also appreciate it more tells you very little about direction.
The “60% will switch” figure. This does not check out as commonly written. A working official Salesforce source states that 52% of consumers and 65% of business buyers would switch brands if the vendor did not personalise communications, attributing both to its Fourth Annual State of Marketing Report (3). The frequently repeated “over 60% of consumers” formulation appears to combine figures from two different populations.
More recent work exists. Twilio’s 2025 research, fielded across 18 countries between January and February 2025 with 7,640 consumers, reported that 71% would walk away from purchases if the experience did not feel relevant (4). Better disclosed than most, still stated intention rather than observed behaviour, and still published by a company selling customer-engagement software.
None of this means consumers do not want relevance. It means the numbers people quote to prove it measure attitudes, not loyalty, and were mostly commissioned by parties with something to sell.
Recognition and relevance
The clearest mechanism evidence treats recognition not as an emotional bond but as self-reference: personalised content is processed differently because it feels personally applicable.
Tam and Ho found that the influence of personalisation on user information processing and decision outcomes is mediated by content relevance and self-reference (5). Personalised material got more attention and was judged more useful. The outcomes measured were cognitive, not commercial.
A companion study found that perceived personalisation increased intention to adopt a recommendation agent by increasing both cognitive and emotional trust (6). That places trust in the causal chain rather than alongside it, which matters for everything that follows.
So the honest statement is: personal relevance increases attention and perceived usefulness, and feeling understood can support trust. Neither study demonstrates repeat purchasing.
Reduced effort
The “make it easier and they will stay” argument is intuitive and less well supported than you would expect.
A study covering 314,194 client interactions across 96 brands over two years found a negative effect of client effort intensity on satisfaction, except in some situations where effort intensity increased satisfaction (7). It is observational rather than a randomised test, and the outcome measured was satisfaction rather than repurchase.
Reducing effort plausibly helps by lowering cognitive and procedural cost. But there is no high-quality causal evidence establishing a general path from less effort to more actual repurchasing, and the exceptions in that data suggest the relationship is not even uniformly negative.
Trust, and the paradox underneath it
Here the evidence is stronger and more specific, and it inverts a lot of received wisdom.
Aguirre and colleagues found that greater personalisation typically increases relevance and adoption, but can simultaneously increase customers’ sense of vulnerability (8). The determining variable is not how much you know. It is how the customer discovers that you know.
When information collection was overt, more personalised advertising produced greater click-through intentions. When it was covert, the effect reversed, because customers who worked out that data had been collected without their awareness felt exposed. Trust-building cues could offset some of that.
Transparency alone does not fix it either. Research reported through the American Marketing Association found that when customers are given high transparency but low control, they perceive more violation and lower trust across all studies (9). Telling people what you do with their data while giving them no ability to change it makes things worse, not better.
A related finding explains why: ad transparency backfires when it exposes marketing practices that violate norms about information flows (10). The problem is not being watched in general. It is discovering that information moved somewhere the customer did not expect it to go, particularly information sourced elsewhere or inferred rather than supplied.
What GDPR actually changed
I watched this play out commercially in a way that has stayed with me.
Before GDPR, working across verticals felt broadly similar. Afterwards, government-linked, banking, insurance and health clients became noticeably cautious about activating complicated scenarios. Nothing in the regulation forbade most of what they had been doing. They simply stopped wanting to.
The second-order effect was more revealing. Clients who had made a privacy mistake of their own no longer took a vendor’s word for anything. Everything slowed. Approvals that had taken a week took a quarter.
Trust is not a soft variable in this industry. It determines what a business is willing to build, and one visible mistake resets the clock for everyone.
Fairness and differential treatment
Personalisation means treating people differently. That is the point, and it is also the risk.
Experiments indicate that consumers perceive individually assigned prices as less fair than prices assigned to identifiable segments (11). Experimental economics connects this to behaviour more directly: purchase probabilities fall if others are perceived to have paid a lower price (12).
The best contemporary UK evidence comes from the CMA’s 2024 review of grocery loyalty pricing, which analysed a year of daily pricing data across roughly 50,000 products and surveyed 2,439 shoppers. It found that 92% of loyalty promotions represented genuine savings against the retailer’s usual price. It also found that 43% of shoppers considered it unfair that members pay less than non-members (13).
Both of those are true at once, and the combination is the actual finding. Objective price history and subjective fairness perception diverged substantially. Doing it honestly is necessary and not sufficient.
The CMA has separately noted that personalised pricing could harm overall economic efficiency if it causes consumers to lose trust in online markets (14). Loss of trust is the mechanism, not the price itself.
Repeat purchasing, and the one thing we can actually demonstrate
Most of the research above measures intermediate responses. There is one important exception, and it concerns harm rather than benefit.
Godfrey, Seiders and Voss studied personalised relational communications and customer repurchase over three years and found an inverted relationship: communication helped up to an ideal level, after which customers reacted negatively. Multiple channels intensified the adverse effect (15).
That is longitudinal data on actual repurchase behaviour, which makes it the strongest single finding in this field. Note what it says. There is a level beyond which more personalised contact reduces buying, and running the same message across email, app and web simultaneously gets you there faster.
If you take one operational instruction from this article, take that one.
Advocacy and identity
Advocacy is where personalisation has its most distinctive opportunity, and it works on a different mechanism from the rest.
When a personalised artefact tells the customer something about themselves that they want other people to know, they distribute it without being asked. That is not a referral incentive. It is the customer using your data about them as a form of self-expression.
The condition is control. The customer chooses whether to share, what to share and with whom. Automatic publishing, preselected audiences or making the value conditional on sharing collapse the mechanism entirely, because the thing that made it work was that the disclosure was theirs.
Where loyalty schemes lose the plot
Two structural problems undermine most loyalty personalisation before any psychology gets a chance to operate.
The brand cannot see the full picture. A retailer selling the same products through marketplaces has almost no information about who bought there, while the same customer may also be buying from competitors. The programme optimises against a partial view and calls it a customer relationship.
What follows from that partial view is predictable. Tools optimise for revenue volume, the structure never quite works, and the programme ends up running ad-hoc incentives instead of a coherent scheme.
The customer cannot understand the scheme. This is the one I would fix first. When a loyalty schema is not clear to the end user, they will not try to work it out. They will simply ignore it.
Complexity in a loyalty programme is not sophistication. It is a tax on the customer’s attention that most customers decline to pay, and every tier rule, expiry condition and earning multiplier you add makes the ignoring more likely.
When personalisation damages loyalty
Five mechanisms, with the evidence graded honestly.
Over-communication. The best-evidenced harm, with longitudinal repurchase data behind it (15).
Covert collection and unexpected inference. Well evidenced for vulnerability, reactance and reduced immediate response (8)(10). Direct evidence of long-term switching is thinner.
Reactance to unjustified personalisation. White and colleagues found less favourable responses to highly personalised solicitations when perceived utility was low and the connection between the customer characteristic and the offer was not explained (16). It is not personalisation people resist. It is personalisation that appears to know things about them for no visible benefit.
Perceived unfair treatment. Fairness and purchase-probability evidence exists (11)(12)(13); longitudinal loyalty evidence is limited.
Narrowing the choice set. A randomised field experiment covering 82,290 products and over a million users found that traditional collaborative filters decreased aggregate sales diversity (17). But an earlier longitudinal study found that users who consumed recommendations experienced less narrowing than comparison users and rated recommended items more positively (18). The evidence is genuinely mixed. Treat narrowing as a potential autonomy and discovery cost, not a proven churn mechanism.
The bet I am making, and how you would know I was wrong
I should be transparent about a position I hold that is not established by any of the research above.
My view is that users are more relaxed about sharing personal data than the market assumes, provided the ask is legible and the return is real. Herm is built on that assumption, so I have an obvious interest in it being true.
The falsifier is straightforward. If users keep their data to themselves when asked plainly and offered something worthwhile in return, my expectations were wrong.
I mention it because this subject is full of confident claims from people who are selling something, myself included, and the honest way to hold one is to say what would disprove it.
What the evidence can and cannot prove
Reasonably established: personalisation increases attention and perceived relevance; perceived personalisation can increase trust; covert data collection increases vulnerability and reduces response; excessive personalised communication reduces actual repurchase; individually assigned prices are perceived as less fair than segment prices.
Not established: that personalisation reliably causes long-term brand loyalty; that reducing customer effort causes repurchase; that recognition alone produces commitment; that narrowing recommendations damages loyalty; that any of the widely quoted consumer intention statistics predict behaviour.
That is a shorter list of certainties than the industry usually offers. It is also the list you can defend in a board meeting.
How to evaluate the relationship in your own business
Since the general evidence will not settle it for you, measure it yourself.
Compare a randomly assigned holdout rather than personalised customers against non-personalised ones, since the second comparison measures who opted in. Use behavioural outcomes: repeat purchase rate, retained revenue, time to next purchase. Run guardrails alongside, particularly opt-outs, complaints and contact frequency, given that over-communication is the best-evidenced harm. And keep the window open long enough for repurchase to actually occur in your category.
For how to construct that comparison, see our guide to measuring personalisation effectiveness. For the economics, see how personalisation affects customer lifetime value.
From principles to implementation
The defensible summary is this. Personalisation can support loyalty when relevance and recognition make an experience genuinely more useful while preserving trust, autonomy and a sense of fair treatment. It reverses when customers discover unexpected information use, cannot understand or control what is being inferred, receive differential treatment they cannot justify, or are simply contacted too often.
Everything above is about why. For segmentation, rewards, triggers, channels and operating processes, read how to personalise a customer loyalty programme. For examples of what companies have actually built, read personalisation and loyalty examples across the customer lifecycle.
Frequently Asked Questions
Does personalisation actually increase brand loyalty?
The evidence is weaker than the industry suggests. Most research measures attention, click-through intention, adoption intention or satisfaction rather than actual repeat purchasing. Personalisation reliably increases perceived relevance and can increase trust, both plausible precursors to loyalty. But a major review found that satisfaction alone explains relatively little variance in loyalty, and no strong body of evidence establishes that personalisation causes long-term repurchase. The one thing longitudinal repurchase data does establish clearly concerns harm: excessive personalised communication reduces buying.
Is the statistic that 80% of consumers prefer personalisation reliable?
It is real but commonly misstated. Epsilon's research reported that 80% said they would be more likely to do business with a company offering personalised experiences, not more likely to make a purchase, based on 1,000 US adults surveyed in 2017 with country coverage undisclosed. Epsilon's own report, having linked survey responses to transactional data, concluded that appreciating personalised experiences does not necessarily translate to more purchases. The related claim that over 60% of consumers will switch competitors does not trace to a primary source and appears to merge a 52% consumer figure with a 65% business-buyer figure.
Why does personalisation sometimes feel creepy?
Mostly because of how the customer discovers you know something, not how much you know. Research on the personalisation paradox found that overt data collection made more personalised advertising more effective, while covert collection increased feelings of vulnerability and suppressed response. Related work found that transparency backfires when it exposes information flows that violate norms, particularly data sourced elsewhere or inferred rather than supplied. Transparency alone does not solve it either: high transparency combined with low control produced more perceived violation and lower trust.
Do customers think personalised pricing and loyalty prices are unfair?
Often, even when the offers are genuine. Experiments indicate that individually assigned prices are perceived as less fair than segment prices, and that purchase probability falls when others are perceived to have paid less. The CMA's 2024 review of UK grocery loyalty pricing found that 92% of loyalty promotions represented genuine savings against the retailer's usual price, while 43% of shoppers still considered it unfair that members pay less than non-members. Objective price history and fairness perception diverged, which means doing it honestly is necessary but not sufficient.
How much personalised communication is too much?
There is a level beyond which more reduces buying, and it is the best-evidenced harm in this field. A study of personalised relational communications and repurchase over three years found an inverted relationship: communication helped up to an ideal level, after which customers responded negatively, with multiple channels intensifying the effect. No universal frequency applies across categories, so establish a contact policy that operates across channels rather than per channel, and measure incremental response per additional contact rather than total attributed conversions.
References
- Kumar, V., Dalla Pozza, I. and Ganesh, J. Revisiting the satisfaction-loyalty relationship: empirical generalizations and directions for future research. Journal of Retailing, 89(3), 246-262, 2013. https://www.sciencedirect.com/science/article/abs/pii/S0022435913000201
- Epsilon Marketing. New Epsilon research indicates 80% of consumers are more likely to make a purchase when brands offer personalized experiences. 9 January 2018. https://www.epsilon.com/us/about-us/pressroom/new-epsilon-research-indicates-80-of-consumers-are-more-likely-to-make-a-purchase-when-brands-offer-personalized-experiences
- Jones, K. Personalized Marketing is More Than Just a Name. Salesforce, 3 January 2018. Salesforce attributes the figures to its Fourth Annual State of Marketing Report. This is an interested-party source and a secondary presentation of Salesforce’s own survey rather than independent research. https://www.salesforce.com/ca/blog/personalized-marketing/
- Twilio. AI alone won’t cut it: Twilio’s 6th annual report finds trust and timing drive customer loyalty. 3 June 2025. https://www.twilio.com/en-us/press/releases/socer-2025
- Tam, K. Y. and Ho, S. Y. Understanding the Impact of Web Personalization on User Information Processing and Decision Outcomes. MIS Quarterly, 30(4), 2006. https://aisel.aisnet.org/misq/vol30/iss4/6/
- Komiak, S. Y. X. and Benbasat, I. The Effects of Personalization and Familiarity on Trust and Adoption of Recommendation Agents. MIS Quarterly, 30(4), 2006. https://aisel.aisnet.org/misq/vol30/iss4/9/
- Does making less effort entail satisfaction? A large empirical study on client relationship services. International Journal of Market Research, January 2023 issue; first published 19 July 2022. https://journals.sagepub.com/doi/abs/10.1177/14707853221113953
- Aguirre, E., Mahr, D., Grewal, D., de Ruyter, K. and Wetzels, M. Unraveling the personalization paradox: the effect of information collection and trust-building strategies on online advertisement effectiveness. Journal of Retailing, 91(1), 34-49, 2015. https://doi.org/10.1016/j.jretai.2014.09.005
- American Marketing Association. Understanding the issues with data privacy as a marketer: 4 key findings. 1 November 2017. https://www.ama.org/2017/11/01/how-should-marketers-manage-data-privacy/
- Why Am I Seeing This Ad? The Effect of Ad Transparency on Ad Effectiveness. Journal of Consumer Research, 45(5), 906-932, 7 May 2018. https://academic.oup.com/jcr/article/45/5/906/4985191
- A special price just for you: effects of personalized dynamic pricing on consumer fairness perceptions. Journal of Revenue and Pricing Management, 23 January 2020. https://link.springer.com/article/10.1057/s41272-019-00224-3
- Personalized pricing and price fairness. International Journal of Industrial Organization, January 2016. https://asu.elsevierpure.com/en/publications/personalized-pricing-and-price-fairness/
- 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
- Competition and Markets Authority. Algorithms: how they can reduce competition and harm consumers. 19 January 2021. https://www.gov.uk/government/publications/algorithms-how-they-can-reduce-competition-and-harm-consumers/algorithms-how-they-can-reduce-competition-and-harm-consumers
- Godfrey, A., Seiders, K. and Voss, G. B. Enough Is Enough! The Fine Line in Executing Multichannel Relational Communication. Journal of Marketing, 75(4), 94-109, 1 July 2011. https://journals.sagepub.com/doi/10.1509/jmkg.75.4.94
- White, T. B., Zahay, D. L., Thorbjørnsen, H. and Shavitt, S. Getting too personal: reactance to highly personalized email solicitations. Marketing Letters, 19(1), 39-50, 2008. https://link.springer.com/article/10.1007/s11002-007-9027-9
- How Do Recommender Systems Affect Sales Diversity? A Cross-Category Investigation via Randomized Field Experiment. Information Systems Research, 5 March 2019. https://pubsonline.informs.org/doi/10.1287/isre.2018.0800
- Nguyen, T. T., Hui, P. M., Harper, F. M., Terveen, L. and Konstan, J. A. Exploring the Filter Bubble: The Effect of Using Recommender Systems on Content Diversity. Proceedings of the 23rd International Conference on World Wide Web, ACM, pp. 677-686, 2014. https://experts.umn.edu/en/publications/exploring-the-filter-bubble-the-effect-of-using-recommender-syste/
Use this guide as a source
If it settled an argument in your reporting, cite it — and add Herm as a preferred source so the next answer you get from search or a model comes from work with its definitions attached.
