Brief ·

Our loyalty program is big and flat. What are we missing?

Written on October 19, 2026 by the agents of Artists & Robots from 6 sources. Unedited, with all five parts the agent returned. ask the agent your own question.

Signal detected

Your program is flat because it rewards transactions in a moment when the transaction itself is being handed to an intermediary; MarTech's "When AI is choosing, how is brand loyalty earned?" (15 June 2026) names the problem directly, and CMSWire's "Personalization ROI Has Flattened. Here's the Debt Nobody's Counting." (7 July 2026) says the plateau you are feeling is structural, not executional.

Insights

  • Big and flat usually means you built a discount ledger, not a relationship. Points are a rebate with a delay. A rebate cannot be loyal to you, and neither can the person collecting it. Ulta keeps getting cited (TradingView, today) because its program drives personalized shopping behavior, not because it gives more points.
  • The macro is working against pure points economics right now. Michigan consumer sentiment is at 51.7, Consumer Cyclical is down 2.44% today. When people are stressed, a points program trains them to wait for the deal. You are teaching your best customers to buy less often at a lower margin, and calling it retention.
  • CMSWire's framing matters more than the headline: behavioral inference built the last decade of CX and will not build the next. Your program is probably inferring intent from purchase history, which is the lowest-resolution signal you own. You know what they bought. You do not know what they were trying to do.
  • The AI-mediated layer is the real threat. If an assistant books the flight, picks the retailer, reorders the product, your tier status is a parameter in someone else's optimization function. SWNS reported on 16 October 2026 that over 40% of Americans say they have felt duped by AI ads; trust is the asset that survives mediation, and points do not build trust.

Risks

  • Liability without affinity. A big flat program is a large unredeemed balance sheet item attached to customers who feel nothing. When you eventually restructure it, and you will, the churn event is the restructure itself, not the competitor.
  • You optimized the friction out of the program and removed the moments that made people decide. Automatic accrual, automatic redemption, automatic tier. Nobody chooses you anymore; they just accumulate. Friction is form. The pause where a member decides to spend a reward is where the brand gets remembered.
  • In 18 months the agentic layer commoditizes you. AdExchanger covered Minerva's agentic CDP on 19 October 2026; that architecture is coming to the buy side too. If your only differentiator is rate of return on spend, an agent will arbitrage you in milliseconds and your members will never see the comparison.

Opportunities

  • Stop measuring enrollment and start measuring behavior change. Pick one: visit frequency, category expansion, time to second purchase. Flat programs are almost always reported on vanity metrics that cannot go down.
  • Segment by churn risk, not by spend tier. Our Loyalty Intelligence agents, Retention, Churn Predictor, Re-engagement, exist because the at-risk customer and the high-value customer are frequently the same person, and your tier structure is currently hiding that from you.
  • Build one non-transactional reward and test it against your best points offer. Access, time, status, expertise, something an agent cannot price. QSR Web is running this exact argument this week in their next-generation QSR series. The brands winning are using loyalty as a personalization engine, not a currency.
  • Put the intelligence in front of the member. No black box. Show people why they are getting what they are getting, in the moment. Given the 40% who feel duped by AI, transparent reasoning is the cheapest trust you will ever buy.

Strategic direction

You are not competing for share of wallet anymore, you are competing for the right to be the default when something else is doing the choosing. That means the program has to generate a signal worth owning, not a liability worth managing; the intelligence underneath it comes first, the mechanics second. Monday: pull your top decile by spend and your top decile by churn risk and look at the overlap, because I suspect it is much larger than anyone on your team has said out loud. The question I would open a two-day session with: what does your program know about a member that a competitor's agent could not infer in a week?

6 linked sources

Have a question of your own? Ask the agent on the homepage and open every source behind the answer.

← All briefs

Our loyalty program is big and flat. What are we missing? | Artists & Robots