Loyalty is infrastructure, not a feature

Ask ten SaaS companies where their loyalty program lives and most will point at a single app: a plugin in the storefront, a tab in the mobile app, a widget near checkout. Ask where their payments live and nobody points at one app. Payments live everywhere, because payments are infrastructure. Loyalty, for almost everyone, is still a feature. That's the whole problem.

What "feature" loyalty looks like

Feature loyalty has a home. It's a points balance that lives inside one storefront, one app, one vendor's dashboard. It knows what happened at checkout and nothing else - not the support ticket from last week, not the in-store visit, not the referral that actually brought the customer in. It can't answer a question asked from outside its own app, because it was never built to be asked.

That's fine right up until you need it to do something it wasn't built for: power a marketing send, flag a churn risk to support, feed a lifetime-value model finance actually trusts, or hand a coherent customer history to whatever you plug in next. Feature loyalty breaks at exactly the moment it becomes useful, because usefulness means other systems need to reach it, and it was built to be reached by exactly one.

Infrastructure doesn't ask permission

Nobody thinks of Stripe as a "payments feature" of their app. It's the layer every other system - finance, support, fraud, reporting - reaches into without asking the checkout flow's permission first. Twilio isn't a texting feature; it's the messaging layer underneath a dozen products that never talk to each other otherwise. That's what infrastructure means in practice: it's boring to look at, it's invisible to the end customer, and everything else in the stack quietly depends on it.

Loyalty infrastructure means the same thing applied to the customer relationship: one durable record of what a customer did and what it meant, that a marketing tool, a support console, a finance model, and whatever you build next quarter can all query without you rebuilding it four times.

What that looks like, concretely

It starts with an event ledger, not a points table - every purchase, visit, referral, and redemption recorded as it happens, append-only, so the history is reconstructable rather than overwritten. On top of that sits a behavior layer that reads the ledger in real time and can act on it: adjust a tier, trigger a reward, surface a churn signal, the moment the underlying event occurs, not in a nightly batch (we go deeper on why that timing matters in a follow-up post).

Then it needs a way out: APIs and webhooks so the CRM, the support tool, the finance model, and the loyalty program agree on one customer history instead of reconciling five different ones. And it needs to carry enough context - margin, not just spend - that the number it hands back is one finance actually trusts (more on that in a follow-up post on margin visibility).

None of that is visible to the customer. That's the point. Infrastructure earns its keep by being reliable and reachable, not by being seen.

Why this matters more now, not less

The case for infrastructure over features used to be about avoiding vendor lock-in and rebuild costs. It still is, but there's a sharper reason now: AI agents doing personalization, support, and lifecycle work need structured signal they can actually trust - not a scraped points balance from a widget that was never meant to be read by anything but its own front end. An agent deciding whether to offer a discount, flag a risk, or recognize a loyal customer is only as good as the data it's handed, and a bolted-on feature was never built to hand anything to anyone (we cover the risk side of this in a follow-up post).

That's the bet we built Loom Loyalty on: Weave as the event-native ledger, Patterns as the behavior engine reading it live, and an API layer that assumes from day one that something else - a tool you already use, or one you haven't picked yet - is going to need to ask it a question. Loyalty that's infrastructure survives your next platform migration. Loyalty that's a feature doesn't outlive the app it's bolted to.

If you're deciding whether to bolt on another loyalty widget or build the layer underneath it, let's talk fit.