How to Build a Retention and Loyalty Operating System
Apr 2026 · 16 min
After years building teams across Amazon, HelloFresh, MakeSpace, and Mejuri, I've come to one conviction that shapes how I operate: the value is always in the operating framework, not in any single tactic or campaign. And the hardest part is rarely the technology - it's the change management required to break down siloed ways of working.
That belief is the foundation of Fullstack OS. Teams that compound build systems that cut across functions instead of running a calendar of tactics that work in isolation and never add up to much. Tactics need a framework in order to compound, and the job of the leadership team is to install that operating system while keeping the tactical work sharp.
Why Retention & Loyalty Need an Operating System
Retention sits at the intersection of product, brand, CRM, CX, analytics, and stores. Every function has a seat at the retention table, and most bring their own tool, KPI, and calendar. When those functions run as one system, retention becomes one of the most compounding investments on the P&L.
Retention is also where the word loyalty tends to appear - sometimes as a synonym, occasionally as a program name, and rarely as a distinct thing. That distinction is worth drawing because it changes how you measure, fund, and operate.
Retention
Decay Reduction
Do customers come back? Retention is measurable, tactical, and responsive. A well-timed reminder, a relevant cross-sell, a service follow-up can all improve it quickly.
- Tracked via Repeat Rate
- Levers: Relevance, Timing, Incentives
- Usually owned by Retention or Growth
Loyalty
Preference & Advocacy
Do customers pick you first, and do they bring others? Loyalty takes longer to build and it shows up in the choices customers make when no one is prompting them.
- Tracked via NPS, Referrals
- Levers: Experience, Community, Product
- Owned across Brand, Product, and CX
You can't build loyalty without first retaining customers long enough for preferences to form. And you can't sustain retention indefinitelywithout the preferences that loyalty creates. They're two distinct jobs within one system, often run in siloes by different teams.
That's why I think of it as a Retention & Loyalty Operating System (RLOS) - one cross-functional model with two distinct jobs, a shared north star, and a single-threaded owner.
1. Start with the Metric: EGR
If retention and loyalty are two jobs of one system, the system needs a single number that captures both. The metric I prefer is Earned Growth Rate (EGR) — the share of revenue growth that comes from existing customers and from new customers earned through them.
North Star
EGR = NRR + ENC
Earned Growth Rate = Net Revenue Retention (what existing customers are worth to you) + Earned New Customers (the new customers who showed up because of them).
NRR
Net Revenue Retention
How much revenue your existing customers generate over a defined period. This is the retention side of the equation - revenue that returns without needing to be reacquired.
ENC
Earned New Customers
New customers acquired through existing ones: referrals, word-of-mouth (HDYHAU at checkout), and advocacy driven by delight — NPS promoters, five-star reviews, UGC.
EGR is the compounding outcome. It's not the only number you need — cohort curves, predicted CLV, latency curves by category, tier migration rates, benefit usage all matter. But those are diagnostic. EGR is the outcome they roll up to, and it's the number I'd put on the exec dashboard and review weekly.
This framing also allows brands to tie loyalty directly to revenue, while adding referrals and word-of-mouth to the list of core KPIs they track. Operationalizing HDYHAU at checkout, at events, and in stores can turn word-of-mouth from something "untrackable" into something you can actually measure and grow.
2. The State Machine
Below the metric sits the spine of the system: a lifecycle state machine.
Every customer sits in exactly one state at any given time. Each state has a goal, a trigger, a channel mix, and a measurement. When the customer's behavior changes, the state changes - and the journey should change with them. The calendar should not run the journeys - the signals should.
At Mejuri, we implemented this with windows and thresholds tuned to jewelry economics and customer shopping behavior. In an apparel business, those numbers would look entirely different. The shape of the machine is universal; the thresholds are category-specific, and they should be tuned annually.
Where Loyalty Lives in the State Machine
Loyalty shows up as a membership layer that packages access, experiences, and community on top of the customer’s lifecycle state. I am not going to mention discounts here because competitors can match your 15% off but they cannot easily match a bring-a-friend event, a private preview, or a 1:1 relationship with a stylist or advisor.
A simple three-tier structure that works for most consumer brands:
Entry — gated on first purchase or subscription. Benefits tend to be high perceived value, low cost to serve: early access, a birthday moment, a shipping perk.
Mid — gated on annual spend. Starts feeling like membership rather than a punch card: free services, free product consultations, invite-only events.
Top — for the highest-value customers. Limited editions, co-creation councils, 1:1 clientelling. At this tier, you're running a relationship business.
The KPIs that matter here include: incremental NRR by tier, signups, tier distribution, tier migration rates, and benefit usage. Incremental NRR is the one I would watch most closely - if a tier isn't delivering meaningfully higher NRR than non-members (controlled for self-selection), the layer isn't pulling its weight yet, and the program needs iteration.
3. The Retention & Loyalty Stack
Science without Art produces cold, generic messaging. Art without Science produces emotionally resonant content that lands with the wrong customer at the wrong moment. The goal is to keep these three layers moving together so they can iterate and compound over time.
Art
Where retention programs usually earn their emotional resonance. Why a customer bought, what would delight them next, and what to say.
Science
Before you personalize or optimize anything, you need to know which customers are active, which is churning, and which are worth the next dollar of investment.
Technology
The stack that makes the first two layers executable at scale.
4. Tactics
A system doesn't replace tactics. It gives them structure.
Quarterly commercial targets are real, and a reactivation push ahead of a slow month is a valid move. The difference is that the tactical work sits inside a measurement framework that can tell you what actually worked, who it worked on, and whether it translated into durable behavior. Tactical work becomes faster to execute, sharper in targeting, and more honest in its measurement.
Tactics that run inside the system create a feedback loop that makes the next one smarter.
Where Are You in the Journey?
A quick self-diagnostic. Check the ones that are true today. There's no right score — the point is to see where your system is most and least developed.
0/6
Most teams check 2–4 boxes. That's a normal starting point.