Modular Loyalty Programs: The Crawl, Walk, Run Launch Methodology

A modular loyalty program launches with a small, fully working core, then adds features in planned phases, with each phase designed from the member data the previous phase produced rather than from pre-launch assumptions.

Some loyalty programs are designed for their third year and launched in their first week. The team builds the full vision (tiers, gamification, segmentation, partner rewards, personalized offers) and decides nothing can launch until all of it is ready. Every design decision is then made before a single member has enrolled, so thresholds, earn rates and features rest on predictions about behavior that nobody has yet observed. McKinsey found in 2021 that "around two-thirds of established loyalty programs fail to deliver value, with many actually eroding value." Designing everything before any evidence exists is one way a program ends up there. Brandmovers' Crawl/Walk/Run methodology is a way to avoid it.

Key Takeaways

Why do big-bang loyalty launches struggle?

Big-bang loyalty launches struggle because every design decision is made on assumptions, so the first months of real data often show that some of those decisions were wrong after members have already enrolled on them.

A big-bang launch deploys the full program at once. The team has to set tier thresholds, earn rates, reward catalogs, gamification triggers and segmentation rules before any member behavior exists. Some of those assumptions will be wrong: a threshold turns out to be too high for most members, an earn rate produces no change in behavior, or a partner reward sits unredeemed.

Correcting them after launch is expensive in two ways. There is the direct cost of reworking features already built and deployed. There is also a trust cost: changing a tier threshold, an earn rate or an expected reward after members have enrolled means telling them the program they joined is different, which triggers the notice and fairness issues covered in the terms and conditions guide. The later a wrong assumption is found, the more members it affects.

The alternative is not to wait until every assumption can be proven. It is to launch what the program needs, observe real behavior, and make the later decisions on evidence.

What is a modular loyalty program?

A modular loyalty program is built from independent components that launch in sequence, where the design of each later component is decided only after earlier components have produced member data.

Core earning, enrollment and a basic tier structure form the first module. Segmentation and targeted offers form a second, gamification a third, partner rewards a fourth. Each module works on its own when it is added and adds to the member experience rather than changing what members already rely on.

The distinction that matters is when design decisions are made. In a staged rollout of a pre-designed program, every module is designed before launch and simply switched on in sequence, so later modules still rest on assumptions. In a modular program, the design of module two is not finalized until module one has produced evidence about what it should contain. The modular approach makes its later design decisions before members rely on them, rather than correcting them afterward.

How does the Crawl/Walk/Run methodology work?

The Crawl/Walk/Run methodology launches a minimum working program (Crawl), adds mechanics designed from its data (Walk), and completes the full vision on the combined evidence (Run).

Crawl: what goes in and what stays out

Crawl includes what the program needs to work on day one and to collect the evidence Walk will need: core earning (points for purchases plus at least one other earn action), enrollment with confirmation messages, a basic tier structure, the primary transaction integration, a branded member portal, and one launch promotion to drive enrollment. Brandmovers' implementation benchmark to build and launch a program is 90 to 120 days (disclosed by Brandmovers); the phase timings below start from launch day. Set Crawl tier thresholds as provisional: announce a first qualification period and a review date, so a later adjustment follows the stated rules rather than changing them.

Crawl leaves out the gamification library, advanced segmentation, secondary integrations such as the email platform, customer data platform and paid media, A/B testing infrastructure, and complex partner rewards.

Three questions decide whether a feature belongs in Crawl:

1. Is it needed for the program to work on day one? Would a member who enrolled, bought something and checked their balance hit a broken experience without it?
2. Is it needed to collect the evidence Walk requires? Would leaving it out stop the program learning what it needs to learn?
3. Would its absence damage enrollment or early retention? Would members who joined expecting points for purchases feel something was missing?

A feature that gets "no" to all three waits for Walk or Run. The gamification library usually does: the program works without it, it does not block evidence collection, and members rarely miss it unless competitors have made it standard. Advanced segmentation usually does too, and waiting has an advantage, because Walk segments can then be built on real behavior clusters rather than assumed ones.

Crawl is a complete program, not a trial version. Members can earn, check their balance, progress toward a tier, redeem and take part in a promotion. It is scoped for the evidence it needs to collect.

When to move from one phase to the next

Move between phases when the evidence is in, not when a date arrives. Brandmovers plans Walk for roughly day 60 to day 180 and Run for roughly day 180 to month 12, but these are planning ranges. Triggering Walk on a fixed date, such as "Walk starts at day 90," recreates the big-bang problem on a smaller scale, because Walk gets designed whether or not the data is ready. Small programs, including many B2B channel programs, may not produce enough activity in 90 days for stable rates; set a minimum sample before reading the data, and extend Crawl rather than design Walk on noise. Fund Walk and Run as reserved budget and vendor capacity with a flexible start, so an annual budget cycle does not force a date-based trigger.

Crawl to Walk. Move when four things are clear:

• The active member rate has stabilized, measured against a threshold the program sets in advance. Set it from the business case, for example the active rate the revenue model assumes, so Crawl tests that assumption directly.
• Tier progression shows whether current thresholds move the intended share of members up.
• Earn data shows whether members are accumulating at the planned pace.
• The Crawl window has been read against seasonal norms, so a holiday peak or a slow season is not mistaken for a stable active rate.

These answers shape the Walk decisions on thresholds, earn rates and segmentation. The KPI dashboard guide defines the metrics.

Walk to Run. Move when three things are clear:

• The Walk segments behave differently enough to justify targeting them separately.
• A referral mechanic, if added, produces referrals at a rate that justifies the reward cost.
• Triggered email and text messages drive enough engagement to justify a wider communications program.

These answers shape Run's gamification, partner rewards and personalization.

What does each phase include?

Each phase has a defined scope, a list of deliberate exclusions, and the evidence it must produce for the next phase.

The ranges overlap because Walk design starts once Crawl data is readable, which can be before Crawl formally ends.

How do you protect Crawl scope from creep?

Protect Crawl scope by applying the three scope questions to every addition, because creep comes from three predictable directions.

• Stakeholder advocacy. Someone who shaped the full vision sees a favorite feature excluded and argues it is critical, that competitors have it, or that it will be harder to add later. Run the three questions. Members rarely miss a feature they have not been led to expect, unless competitors have made it standard (see the exceptions below), and holding it back keeps Crawl data cleaner, because fewer features compete to explain what drives engagement.
• Vendor bundling. A platform vendor may switch on gamification, segmentation or extra integrations as part of a standard setup, presented as included at no extra cost. Turn them off for Crawl. The aim is to collect specific evidence, not to use every available feature.
• One-more-connection integration. Each integration opens systems, and it is tempting to connect one more while the work is underway. Each addition is small; together they stretch the launch timeline and add decisions the Crawl data would have informed.

When is a phased launch the wrong choice?

A phased launch is the wrong choice when the program cannot launch small without harming members or breaking an obligation, so check these cases before committing to Crawl.

• Replacing an existing program on a fixed date. If the current platform is being shut down, members already expect certain features, and taking them away at launch is a devaluation. Launch at least at parity, then phase the new features. The sunset and migration guide covers this case.
• Contractual or partner commitments. If partners or a coalition are committed to launch-day participation, their mechanics belong in Crawl.
• Competitive parity. If every direct competitor offers a feature members treat as standard, leaving it out may damage enrollment, which is the third scope question.
• Promising what comes next. Phasing means members see a smaller program first. Do not promise specific future features or dates in member communications, because the Walk design depends on evidence you do not yet have, and changing a promised feature later is its own trust problem.
• The costs of phasing. Members judge the launch version, so a Crawl program that feels thin can dent first impressions and enrollment. Each later phase is also another release, another round of testing and another member announcement; budget for those costs rather than assuming phasing is free.

Brandmovers' work with the Idaho Lottery is an example of the relaunch case, where a legacy program's members already existed: the modernized program on BLOYL™ launched with second-chance promotions and gamification, and site traffic increased over 6X compared to previous years (disclosed by Brandmovers).

How do you measure whether the phased approach is working?

Measure the phased approach by whether each phase answers its questions on time and whether the features added in Walk and Run move the metrics they were designed to move.

• Evidence on schedule. Track whether Crawl produces a reliable active rate, tier progression and earn data within the planning range. If it does not, find out why before designing Walk.
• Feature impact. For each Walk and Run feature, set the metric it is meant to change before launch, and test it against a comparison group where possible. The A/B testing guide covers pilot design.
• Post-launch changes. Record the share of launched mechanics that had to be changed after members were using them. A falling share across phases is one sign the evidence-first approach is working.
• Program economics. By Run, compare member lifetime value and retention against a comparison group, so the full program is judged on what it added, not on what members would have done anyway.

Conclusion

Crawl/Walk/Run is not a smaller ambition; it is a different order of decisions. The full vision still arrives, but its later features are designed on what members actually did rather than on what the team predicted. The hard part is organizational, not technical: holding Crawl scope when stakeholders, vendors and integration work all push to add more. Which of the features in your current roadmap would you design differently if you had 90 days of member data first?

Sources

• McKinsey & Company, "Next in loyalty: Eight levers to turn customers into fans" (October 12, 2021)
• Brandmovers, Idaho Lottery loyalty program case study