Which Promotion Should You Give Which Customer?
Imagine sending a €10 voucher to five customers. The first customer was already about to buy, the voucher reduces margin but creates no incremental behavior. The second customer already has store credit, so the new voucher creates unnecessary reward overlap. The third customer is returning an incompatible tech accessory, a discount doesn't solve the underlying problem. The fourth customer is close to canceling a subscription box because the next shipment is arriving too early, a pause would be more useful than money off. The fifth customer is genuinely disengaging and may respond to a controlled incentive.
The voucher is identical in every case. The commercial outcome isn't. Personalized promotions for specialty ecommerce should begin with a decision, not with a code.
First Decide Which Behavior You Want
A promotion should have one primary objective: converting a first purchase, creating a second purchase, reactivating a lapsed customer, increasing adoption of a relevant category, preventing avoidable subscription cancellation, encouraging a referral, moving a customer toward a tier, or supporting a slow commercial period.
"Generate more sales" is too broad to actually work from. Without a defined behavior, the business can't determine who should qualify, how much value is appropriate, which products should be included, when the promotion should end, or whether the campaign actually created incremental value.
Check Whether the Customer Needs an Incentive at All
The first decision should always be whether an extra incentive is even necessary. Strong purchase signals include a recently booked appointment, repeated product-page visits, an active waitlist request, a replenishment pattern, an upcoming subscription order, or a customer already progressing toward an existing reward.
These signals don't guarantee a purchase. They do suggest the brand should test whether information, convenience, or visible loyalty value is sufficient on its own. "No incentive" has to remain a possible outcome. Otherwise personalization becomes a mechanism for distributing more discounts rather than making genuinely better decisions.
Check What the Customer Already Has
Before generating new value, check the existing loyalty state. The customer may already have points, an earned reward, a voucher, gift-card balance, store credit, tier benefits, or membership privileges.
Leat Customer Profiles provide a real-time view of points, credit, tier level, earned rewards, and redeemed rewards. Suppose a customer is 50 points away from a reward, a progress reminder may be sufficient on its own. Suppose another customer has an active €10 voucher already, issuing a second one may create unnecessary stacking, confusion, or margin loss. Existing loyalty value should be treated as available incentive inventory, not ignored in favor of manufacturing something new.
Identify the Actual Friction
A promotion works only when price or perceived value is genuinely part of the problem. Specialty ecommerce has many other forms of friction that a discount simply can't fix.
Compatibility friction shows up when an electronics customer is unsure whether an accessory works with their device, the correct action is verified compatibility information, not 15% off. Skill friction appears when a customer likes a craft kit but worries it may be too difficult, the correct action may be a tutorial, workshop, or clearer difficulty indicator. Timing friction happens when a subscription-box customer has too much product or will be away for the next delivery, the correct action may be a skip, pause, or date change. Trust friction shows up when a gift buyer is unsure whether the recipient can exchange the product, the correct action is a clear gifting and return experience. Availability friction appears when a collector wants a sold-out item, the correct action may be a waitlist or future-access invitation.
Discounts shouldn't be used to paper over operational problems that need a different fix entirely.
Match the Incentive to the Desired Behavior
When an incentive is genuinely appropriate, use the smallest form of value likely to work. A reminder helps when timing or awareness is the main issue. Visible progress helps when the customer is already close to a reward or tier. Recognition suits anniversaries, community participation, or valued membership behavior. Early access works well for collectibles, new games, stationery launches, or limited hobby products. Bonus points encourage a selected category or action without immediately lowering the checkout price. An extra stamp suits a clearly defined, repeatable behavior. Conditional shipping helps when delivery cost is genuine friction and a minimum-spend condition remains commercially sensible. A voucher fits when a specific financial nudge is justified. Gift-card or store credit works for controlled recovery, referrals, or selected customer journeys. And no incentive is right whenever the expected behavior is already likely to occur on its own.
Apply Customer Eligibility Before Value Is Issued
A relevant promotion still needs operational boundaries. Check customer segment, current loyalty state, recent purchases, recent returns, subscription status, product eligibility, location, channel, minimum spend, validity, frequency cap, cooldown period, previous campaign exposure, and campaign budget.
Leat Vouchers can be managed centrally with defined rules and validity, and current guidance also describes single- and multi-use structures, total campaign-volume limits, and dynamic activation or expiration periods. This allows a promotion to be specific without becoming uncontrolled.
Protect Margin at Product and Category Level
Specialty assortments often contain very different economics. An electronics store may sell low-margin devices, higher-margin accessories, warranty-related services, cables, replacement components, and promotional bundles. A hobby retailer may sell core tools, consumable materials, limited kits, workshops, and licensed collectibles. The same voucher shouldn't automatically apply to all of them.
Leat's Orders API can associate order lines with products and categories, enabling product-level filtering and loyalty logic in a connected implementation. Possible controls include excluding constrained products, allowing bonus points only on selected categories, preventing voucher use on subscription renewals, applying benefits to accessories rather than core electronics, or limiting a reward to a defined collection. A promotion strategy should reflect contribution margin, inventory, and customer relevance, not a single blanket rule applied everywhere.
Let AI Estimate; Let Rules Control
AI can help identify patterns that are difficult to monitor manually, which customers are becoming less active, which reward is losing effectiveness, which members are close to a meaningful threshold, or which campaign may influence a target behavior. That prediction shouldn't directly issue unlimited discounts on its own.
Leat AI currently uses points, transactions, and redemption history and can determine whether a customer receives a point boost, voucher, or gift-card credit within merchant-set margin and approval limits, with actions above the configured threshold waiting for approval. A responsible sequence moves from prediction, to customer eligibility, to product eligibility, to an existing-value check, to a margin check, to an approval check, and only then to action. AI estimates what may work. Rules determine what's actually allowed.
Five Customers, Five Decisions
Consider a niche ecommerce brand selling desk accessories, stationery, tech organizers, and a monthly productivity box.
The likely full-price buyer has viewed a new desk organizer several times, added it to a wish list, and regularly buys launches at full price. The decision is a product-availability reminder, with no incentive attached.
The near-reward member is 100 points away from a reward and has recently viewed refills. The decision is showing loyalty progress, with no new value added.
The overstocked subscriber begins canceling a monthly productivity box and selects "I have too many unused items." The decision is offering a pause or skip through the subscription platform, with no incentive unless the customer remains at genuine risk after the timing issue is resolved.
The category explorer has repeatedly purchased notebooks but never tried compatible refill inserts, which are relevant and commercially eligible. The decision is introducing the refill category, with controlled bonus points rather than a percentage discount.
The service-recovery case received a damaged gift item and had a delayed support experience. The decision is human-led service recovery, with approved store credit subject to an appropriate value and validity period.
The brand hasn't created five different promotional campaigns for the sake of complexity. It's matched five different actions to five genuinely different customer situations.
This is an illustrative industry setup. It does not describe the configuration or results of a specific Leat customer.
Control Promotion Stacking
Promotion stacking becomes genuinely dangerous when different systems issue value independently. A customer might combine a public sale, a loyalty voucher, a subscription discount, a free-shipping code, store credit, and a gift-card balance, some of these combinations may be intended, others may reduce the order below an acceptable margin without anyone noticing.
Define which benefits can combine, which benefit applies first, whether points are earned on discounted value, whether sale items qualify, and what happens after a return. The customer-facing rules should stay understandable throughout. Margin control shouldn't create a checkout experience where nobody can explain why a reward failed to apply.
Measure Incremental Value Rather Than Redemption
A high redemption rate can make an inefficient campaign look successful when it isn't. Measure incremental conversion (how many customers purchased because of the promotion rather than merely using it), full-price conversion rate (how many eligible customers purchased without an extra incentive), incentive cost per incremental order (how much reward value was required to create additional behavior), margin after incentives (what contribution remains after discounts, points, credit, and shipping benefits), suppression rate (how often did the system correctly select no incentive), existing-value utilization (did customers use rewards they already held), promotion overlap rate (how often would multiple benefits have been issued without controls), and post-promotion retention (did customers return again, or only use the one-off offer).
The Best Promotion Strategy Gives Fewer Unnecessary Promotions
The objective of personalization isn't inventing a different discount for every customer. It's understanding what each customer actually needs next.
Some need information. Some need flexibility. Some need existing progress made visible. Some need a controlled reward. Some need a service problem resolved. And some need nothing at all. When customer data, AI estimates, and business rules work together, promotions become deliberate commercial decisions rather than automatic margin reductions handed out by default.
Frequently Asked Questions
How do you decide which promotion to give a customer?
Start with the desired behavior, then evaluate purchase likelihood, existing rewards, customer friction, product eligibility, margin, and recent campaign exposure before choosing anything.
Should every lapsed customer receive a win-back voucher?
No. Some customers have naturally long purchase cycles, unresolved service issues, or existing loyalty value already sitting in their account. The intervention should match the actual reason for inactivity, not a default assumption.
Can AI automatically choose loyalty incentives?
AI can support prediction and selection, but eligibility, product restrictions, margin limits, and approval thresholds should determine what can actually be issued in every case.
