A targeted loyalty offer goes to the members a program picks, and to no one else. The program decides who gets it and what it is worth. It also decides who pays. Then it checks the result against members it left out on purpose.
What are targeted offers?
Targeted offers are promotions a loyalty program sends to chosen members in place of one deal for the whole base. The choice rests on member data such as purchase history or tier, or on a score from a model. Some programs pick a group and send it one offer. Others choose a different offer for each member, which is what most people mean by personalized offers.
Large programs describe this on their own pages. Chase targets Chase Offers using each customer's purchase history (Chase). Kroger's retail media arm says its data science informs the personalized offers each customer gets, and when they arrive (Kroger Precision Marketing). Marriott ties one points offer to members with a paid stay in the last 12 months. The offer goes out by personalized email (Marriott Bonvoy terms).
How are targeted offers selected?
Programs use three methods, often in layers.
- Rules. A rule picks members by facts on the record. The Marriott offer above is a rule: a paid stay in the last 12 months. Rules are easy to explain and easy to audit. They cannot rank one member above another inside the rule.
- Propensity models. A model gives each member a score for how likely they are to act, such as buy a new category or stop buying. Kroger's Targeted Digital Coupons go to households, and its 84.51° data science informs who gets which one (Kroger).
- Next best offer. For each member, the engine ranks every offer the member qualifies for and picks one. It works inside limits on budget and contact. An offer engine holds the offers and the rules that bind them.
The common mistake is to score who will buy. The better score is who will change. Eva Ascarza tested both using field data from a wireless provider and a membership group. In simulations on both data sets, targeting by expected response cut churn more than targeting the members at highest risk (American Marketing Association). Her study looked at retention offers, and the logic carries to any offer. A member who buys the product every week does not need the coupon.
Who funds targeted offers?
Three parties pay for targeted offers. Who pays sets the terms, and it decides what the program has to record.
| Who pays | Why they pay | Example from the company's own page | What the program tracks |
|---|---|---|---|
| The program itself | More visits and spend from its own members | Marriott sends a points offer to members with a paid stay in the last 12 months (Marriott) | Offer cost against the lift it bought |
| A partner brand | To win new or lapsed customers among the program's members | Chase charges brands only when a customer views a campaign and makes a purchase (Chase) | Each qualifying purchase, for billing |
| A supplier | Trial and repeat purchase of its products in the retailer's stores | Kroger Precision Marketing sells Targeted Digital Coupons to brands for product trials and repeat purchases (Kroger Precision Marketing) | Each redemption, by funder |
Supplier money takes its own path. Kroger's coupon rules send coupons from manufacturers to Inmar for setup. Kroger's own coupons are set up by Kroger (Kroger coupon guidelines). When a partner or supplier pays, each redemption becomes a line on a bill. The program needs a record of each one, at the agreed rate.
What privacy rules apply to targeted offers?
Targeting runs on personal data, so privacy law sets the edges. Four rules shape how loyalty teams target in the EU and the US.
In the EU, a person can object at any time to direct marketing, and that right covers profiling tied to it. Once they object, the data can no longer be used for it. The program must flag this right no later than its first message (GDPR Article 21).
Under GDPR Article 22, people have the right not to face a decision made only by machine if it has a legal or similar serious effect. EU data protection regulators say most targeted ads do not reach that bar. They add that charging people different prices could, if high prices shut someone out of goods or services (EU guidance).
California's CCPA lets a business offer financial incentives for personal data, and it names loyalty and club card programs as allowed. The business must give notice and get prior opt-in consent that describes the material terms. The member can revoke that consent at any time (Cal. Civ. Code 1798.125).
New York requires a notice when an algorithm sets a price using a consumer's personal data. The notice must read: THIS PRICE WAS SET BY AN ALGORITHM USING YOUR PERSONAL DATA (N.Y. Gen. Bus. Law 349-a). Programs that let a model set a member's price in New York should check offers against it.
How do you measure a targeted offer?
The number to measure is the sales the offer added. Redemptions overstate it, because members who would have bought anyway redeem too. The fix is a holdout. Pick eligible members at random, leave them out, and compare the two groups over the same weeks.
Shortcuts mislead. A study of 15 large ad tests at Facebook compared random-group results with methods that skip the random group. Those methods often got a different answer, even with rich data on who people were and what they did (Gordon et al.). Kroger says it compares exposed and unexposed groups to find true impact (Kroger). Chase gives brands reports on the lift (Chase).
For illustration, a grocery program picks 200,000 members for a $3 coupon on a coffee brand. It holds back another 20,000 at random. Over four weeks, 6% of the holdout buy the coffee. In the offer group, 10% buy it. Every buyer redeems the coupon.
That is 20,000 redemptions, costing $60,000. Without the offer, 6% of the 200,000, or 12,000 members, would have bought anyway. So the offer added 8,000 buyers, at $7.50 each. A report based on redemptions shows $3 per buyer. It hides that 60% of the money paid for habit.
How does offer fatigue affect targeted offers?
Every offer teaches members something about price. One study used more than eight years of shopper panel data for a packaged good. Its results fit the view that consumers grow more price and promotion sensitive over time as promotions rise (Mela, Gupta and Lehmann, Journal of Marketing Research). Targeting slows this, because fewer members see each deal. A member who sees a deal every week has less reason to pay full price.
Three controls keep fatigue in check. Cap how many offers each member gets in a period. Keep a long-running holdout, so the program sees whether response falls over months. Drop members whose gap over the holdout has closed. The glossary entry on offer fatigue covers the warning signs.
How GRAVTY handles targeted offers
GRAVTY is the Loyalty Juggernaut platform. It runs 500M+ members in production. Its rules engine, GRAVTY Visual Rules, is patented and needs no code. The loyalty team sets who gets an offer itself, segment by segment. Offer Studio builds the targeted offers those rules send. Score models can be tuned for each offer.
Agentic AI Compass tests offers and runs a simulation before an offer goes live. Each piece of advice it gives can be explained and audited. When a partner pays for an offer, the settlement ledger records each partner transaction, with its rate and the way the money moves. The guide to loyalty personalization covers the data under it.
Frequently asked questions
What is the difference between targeted offers and personalized offers?
A targeted offer goes to a chosen group of members, while a personalized offer is chosen for one member. Both sit on the same scale. A personalized offer is often picked by a model that ranks every offer the member qualifies for.
Are targeted offers the same as dynamic pricing?
No, a targeted offer is a reward or discount for chosen members, while dynamic pricing changes the price as conditions change. The two meet when a model sets a price from a member's data. New York calls that personalized algorithmic pricing and requires a notice (N.Y. Gen. Bus. Law 349-a).
Which offer does a member see when several apply?
The offer engine decides, using priority rules or a next-best-offer ranking. Rules can make offers stackable, so the member gets both. They can also make them exclusive, so the higher priority wins. Give partner-funded offers a rule of their own, since each partner pays only for its own offer.
Do members have to consent to targeted offers?
In California, yes: a business needs prior opt-in consent before it enters a consumer in a financial incentive program (Cal. Civ. Code 1798.125). In the EU, the member can object at any time to direct marketing and the profiling tied to it (GDPR Article 21).
Should every targeted offer have a holdout?
Yes, every offer the program funds at scale should run against a random holdout. Without one, redemptions count members who would have bought anyway as wins. Partner-funded offers need it too, because the partner pays for each purchase.