Customer Loyalty Platform
Powered by BLOYL™. From straightforward programs through to complex multi-brand ecosystems
AI-Driven Next Best Action in Loyalty: Beyond Next Best Offer
|
How this guide was prepared. Last updated October 2026. It draws on Brandmovers' experience designing loyalty programs and member communications. It also draws on McKinsey's personalization research and US consent and privacy rules, each checked at its source. |
Next Best Action (NBA) in loyalty decides what to do for each member right now, choosing the offer, the channel, the timing and the reward type together, while Next Best Offer (NBO) chooses only which offer to show.
Next Best Offer models are now common in loyalty: they predict which offer from a set catalog a member is most likely to redeem. But the right offer can still fail if it arrives through a channel the member ignores, at a time they never engage, or as a reward type that does not motivate them. A discount sent to a member who has repeatedly ignored discounts is not really personalization. NBA widens the decision to four dimensions. This guide covers the difference between NBO and NBA, the four dimensions, the data each needs, how to prioritize actions, privacy and consent, how to measure results and how to start without a data science team.
Key Takeaways
|
What is the difference between Next Best Offer and Next Best Action?
Next Best Offer picks the most relevant offer from a set catalog, while Next Best Action decides the whole interaction: which action to take, through which channel, at what time and with which reward type, for a specific behavioral goal.
An NBO model answers a narrow question: of the offers available, which is this member most likely to redeem? The channel, timing and message are usually set by the campaign manager, the same for everyone in the campaign.
An NBA system answers a wider one: given this member's purchase history, engagement, channel habits, reward preferences and position in the lifecycle, what is the most valuable action to take now?
|
Dimension |
Next Best Offer |
Next Best Action |
|---|---|---|
|
What it decides |
Which offer to show from a set catalog |
The offer, channel, timing and reward type for a target behavior |
|
Optimization goal |
Usually redemption probability |
Incremental behavior change, such as retention, frequency or tier progress, net of reward cost |
|
Data inputs |
Purchase history, profile, past redemptions |
NBO inputs plus channel engagement, engagement times, reward-type response and churn risk |
|
Output |
"Show this member Offer A, not Offer B" |
"Send this member an in-app message on Thursday evening with an early-access reward, because they are at risk of lapsing, respond in the app rather than email and rarely redeem discounts" |
|
Where it fails |
Right offer, wrong channel, time or reward type |
Fragmented data, or decision rules never checked against outcomes |
Personalization done well is worth the effort, though these figures cover personalization broadly, not NBA or loyalty programs specifically. McKinsey's personalization research (2021) found that "personalization most often drives 10 to 15 percent revenue lift (with company-specific lift spanning 5 to 25 percent, driven by sector and ability to execute)," and that 71% of consumers expect personalized interactions while 76% get frustrated when they do not get them.
What are the four dimensions of Next Best Action?
The four dimensions are offer selection, channel selection, timing and reward type, and each needs its own signals.
Offer selection
Offer selection is what NBO already does: models trained on past responses predict which offers, values and categories each member will respond to. NBA adds weight to the individual member's own history. If a member's redemptions are all experiences and they have never responded to a price promotion, a discount that performs well across the program is still the wrong offer for them.
Channel selection
Many programs send the same message in the same channel to everyone in a segment. A member who uses the mobile app or wallet pass but has not opened a loyalty email in months gets the same email as a member who reads every one. Channel selection uses simple behavioral signals: each member's open and click rates by channel, push and in-app response, text response where consent exists, and how recently they used each channel. The obstacle is usually data, not modeling: if email data sits in the email platform and app data in a separate analytics tool, the decision cannot be made with full information.
Timing
Members have their own engagement windows. A member who reliably opens loyalty messages on weekday mornings is telling the program when to reach them. Send-time optimization learns these patterns from each member's engagement history and queues messages for that window instead of one campaign time for everyone. Many email and messaging platforms offer it; the requirement is enough history per member to see a pattern. Test it against your usual send schedule before rolling it out.
Reward type
Programs often offer every member the same reward type, usually points or discounts, whether or not it motivates them. A member who has let points accumulate for a long time while buying at a steady rate may be disengaged from points, or may be saving for a larger reward; check redemption history before switching them to early access or an experience. A member who redeems immediately and buys more during point multiplier events is motivated by financial rewards. Useful categories to distinguish are financial (points, cash back, discounts), experiential (early access, events), recognition (status and milestones) and service (free shipping, priority support).
What data does Next Best Action need?
NBA needs a unified member profile that brings together purchases, offer responses, engagement by channel, engagement times and reward-type history, because each dimension depends on signals that usually sit in different systems.
|
Dimension |
Signals needed |
|---|---|
|
Offer selection |
Purchases by category and product, past offer responses, browsing and app searches, segment membership |
|
Channel selection |
Open and click rates by channel, recency of use per channel, consent and opt-in status by channel |
|
Timing |
Engagement timestamps by channel, day-of-week patterns, engagement falling below the member's own baseline |
|
Reward type |
Redemptions by reward type, purchases after each reward-type event, stated preferences, whether the member stockpiles or redeems quickly |
A customer data platform, or a loyalty platform that takes in data from other systems, can assemble these into one profile. Stated preferences collected directly from members fill gaps that behavior leaves open.
How should a program decide which action to take first?
Use a priority queue that ranks possible actions by urgency, with frequency caps so each member receives only the most important action in a given period.
A member can qualify for several actions at once, such as rising churn risk, nearness to a tier threshold and a high likelihood of redeeming. Scores that estimate these likelihoods, often called propensity scores, feed the queue, and the program's rules decide the order. This is where AI does its work in NBA: models estimate each member's churn risk, redemption likelihood and best send time, and the rules turn those estimates into one action. Many programs put churn prevention first, but rank by the expected incremental value of each action: some at-risk members will not respond to any offer, and routine win-back rewards can teach members to lapse. Frequency caps stop NBA from sending so many messages that members stop paying attention. The gamification guide covers the same notification discipline for challenges.
|
Example priority logic (calibrate the thresholds to your own data) |
Action |
|---|---|
|
1. Churn risk above your chosen threshold |
Re-engagement offer in the member's most responsive channel at their usual engagement time, with the reward type they respond to |
|
2. Likely to redeem, with enough points for a reward |
"Your reward is ready" message featuring the reward they are most likely to choose |
|
3. Close to a tier threshold with recent activity |
Progress message and "almost there" nudge in the app or by push |
|
4. Likely to try a new category |
Introductory offer with a first-purchase bonus in their preferred reward type |
|
Frequency cap |
For example, one NBA message per member per week across all teams sending program messages, with an exception for churn prevention |
As a hypothetical example: a member has enough points for a reward, has not opened a loyalty email in 90 days, opens the app most Sunday evenings and has repeatedly ignored discounts, and their churn score has crossed the threshold. The queue ranks churn prevention first, so they receive one in-app message on Sunday evening offering early access to a sale, with their ready reward mentioned in the same message, and the frequency cap holds every other action until the following week.
The guide to re-engaging dormant loyalty members covers the churn-prevention step in more depth.
What privacy and consent rules apply to Next Best Action?
NBA can only use channels the member has agreed to, and data collected in exchange for rewards can trigger additional notice requirements.
- Text messages. Under FCC rules, marketing texts sent with automated technology generally require the recipient's prior express written consent (see the FCC's guidance on robocalls and texts), and under 47 CFR 64.1200 a revocation must be honored "within a reasonable time not to exceed ten business days." Channel selection should only choose text for members with valid consent.
- Push and in-app messages. Members control notification permissions on their devices, so treat opt-in status as part of channel selection.
- Data for rewards. For businesses covered by California's CCPA, a program that offers benefits in exchange for personal information is likely a financial incentive under Civil Code section 1798.125, which requires notice and opt-in consent and limits price or service differences to those reasonably related to the value of the data. Colorado's privacy rules also address bona fide loyalty programs.
- Automated decisions. California's regulations on automated decision-making technology became applicable at the start of 2026, according to the IAPP, so check whether your use of member data is covered.
- Sensitive inferences. Avoid actions that reveal or rely on sensitive inferences, such as health conditions inferred from purchases, and follow your privacy notice on how member data is used. Review whether offer values vary in ways that act as personalized pricing or track protected characteristics; New York's GBL 349-a requires a disclosure when a price is set by an algorithm using a consumer's personal data.
This is general information, not legal advice.
How do you measure whether Next Best Action is working?
Compare members who receive NBA actions with a control group that receives your standard communications over the same period.
- Purchase frequency and retention for members who received churn-prevention actions versus the control group.
- Redemption rate after "reward ready" prompts versus the control group.
- Open and click rates for NBA-timed messages versus your standard schedule.
- Tier progression after proximity nudges versus the control group.
- Unsubscribes and opt-outs, to check that frequency caps are working.
A before-and-after comparison cannot separate NBA's effect from seasonality or other campaigns. Keep a small share of actions for testing alternatives, and hold out a persistent control group, so the model does not keep confirming its own early choices. The loyalty KPI dashboard guide covers the underlying metrics.
How can a program start with Next Best Action without a data science team?
Start with the use case that needs the least new data, usually churn prevention scored by a predictive model, or send-time optimization, and add dimensions as the data allows.
- Churn prevention first. Use the transaction data the loyalty platform already holds, such as declining purchase frequency and unused points, to trigger re-engagement.
- Send-time optimization next. Most email and messaging platforms already record when each member engages.
- Reward type selection. Add it once you have enough history of each member's responses to different reward types.
- Channel selection last. It needs engagement data from all channels in one place.
New members and smaller programs rarely have enough individual history. Use segment-level rules and stated preferences until each member's own data builds up.
Platforms differ in how much of this they provide out of the box, so ask vendors which dimensions their tools cover, what data they need, how they handle members with little history, and for incremental lift measured against a holdout group rather than before-and-after results. The agentic AI guide covers where AI decisioning is heading. BLOYL™, Brandmovers' B2C enterprise loyalty platform, includes a rules engine that sets earning and redemption rules by customer segment, purchase channel, time window and behavioral action, A/B testing against a control group, predictive churn analytics and data flows to connected CRM and customer data platforms.
Frequently Asked Questions
-
Next Best Action is a decision approach that chooses, for each member, the offer, channel, timing and reward type most likely to produce a target behavior, such as a repeat purchase, re-engagement or tier progress. It combines several signals about the member into one recommended action.
-
Next Best Offer chooses which offer to show from a set catalog. Next Best Action also chooses the channel, the timing and the reward type. A program can pick the right offer and still fail if it sends it in a channel the member ignores or at a time they never engage.
-
A unified member profile that combines purchases and offer responses, engagement by channel, engagement times and redemptions by reward type, plus consent status for each channel. Many programs hold much of this data already, but in separate systems that need to be connected first.
-
Yes, if they start small. Churn prevention can use transaction data the loyalty platform already holds, and send-time optimization is built into many messaging platforms. Reward type and channel selection can follow as data allows. Ask vendors which of these their tools support and how results are tested.
-
Compare members who receive NBA actions with a control group that receives standard communications over the same period, tracking purchase frequency, retention, redemption, engagement and opt-outs. A before-and-after comparison cannot separate NBA's effect from seasonality or other campaigns.
Conclusion
Personalizing only the offer solves one of four problems. NBA adds channel, timing and reward type, and each depends on data many programs already collect but keep in separate systems. Unify member profiles, rank actions with a priority queue and frequency caps, respect consent for every channel, start with the use case that needs the least new data, and prove each step against a control group. Programs that personalize the full interaction, not only the offer, give members more reason to feel the program is working for them, which is what the loyalty perception gap is about.
|
Building Next Best Action into your loyalty program? Brandmovers designs and runs loyalty programs on BLOYL, including the rules, testing and data connections that NBA depends on. Request a demo to talk through your program's data and personalization plans with the Brandmovers team. |
Sources
- McKinsey & Company, "The value of getting personalization right, or wrong, is multiplying" (November 12, 2021)
- Federal Communications Commission, "Stop Unwanted Robocalls and Texts"
- 47 CFR 64.1200, Delivery restrictions (revocation of consent)
- New York General Business Law section 349-a (personalized algorithmic pricing)
- IAPP, "New year, new rules: US state privacy requirements coming online as 2026 begins" (January 5, 2026)
- Colorado Secretary of State, 4 CCR 904-3, Colorado Privacy Act Rules
- California Civil Code section 1798.125 (financial incentives)
Barry Gallagher
Barry Gallagher is a loyalty and digital marketing strategist at Brandmovers, where he leads content strategy across B2C and B2B loyalty programs. He writes on program design, engagement mechanics, and the data signals that separate high-performing loyalty programs from the rest.