Case Studies
Selected work.
Case Study 01
GameStop: Revenue From Relevancy
Loyalty ecosystem • First-party media network • CFO-grade attribution
Context
At its peak, GameStop was a $9.5 billion retailer with 6,600 stores worldwide, a 65-million-member loyalty program, and Game Informer — the 3rd largest consumer magazine in the United States with 8 million monthly readers. It was the dominant specialty retailer in gaming globally.
By 2018, the ground had shifted. Physical game sales were declining 3–5% per year as digital downloads accelerated. The trade-in and pre-owned model — the engine that powered loyalty economics — was eroding in real time. By 2020, 91% of gaming industry revenue was digital. The company’s largest net loss in history ($673M) landed in the first year of this engagement.
The operating environment: 5,800 stores across 14 countries, multiple CEO transitions, a global pandemic, and a January 2021 short squeeze that put the company on every front page in the world.
The question was whether a precision revenue system could be built and proven inside a business under structural pressure. The answer was yes.
What Was Built
The data foundation.
Before CDPs existed as a category, we assembled a unified customer identity layer across six data sources: web clickstream, in-store transactions, loyalty card, email, mobile app, and customer profile. We developed 79 customer attributes per member — spanning purchase behavior, category affinity, channel preference, economic motivators, and loyalty engagement. 5.5 million user profiles were enriched with cross-channel behavioral context within the first 90 days of deployment.
We engaged Pointilist to validate the infrastructure across 2.1 billion customer events and 3TB of data. Identity resolution reached the top end of retail industry benchmarks. The proof-of-concept validated six cross-channel lifecycle use cases with quantified revenue impact — including a $3–9M annual opportunity from email optimization alone, and a finding that email subscribers were 7.5x more likely to repurchase than non-subscribers (70% repurchase rate vs. under 10%).
The media network.
We built one of retail’s earliest first-party media networks — before the retail media category had a name. The asset stack: 65M loyalty member behavioral data, Game Informer’s editorial reach and subscriber base, and GSTV’s in-store broadcast network across 3,800 locations. Publisher co-op campaigns ran for EA, Activision, Take-Two, Ubisoft, Warner Bros., Sony, Microsoft, and Nintendo.
We also negotiated and integrated platform partner data from Xbox, PlayStation, and Nintendo — gameplay behavior, friends lists, titles owned, achievement data — to build a proprietary Gamer Graph enabling social-graph-level audience targeting that no competitor or publisher had access to.
The attribution system.
The measurement infrastructure was built to CFO standard. Weekly attribution reports tracked marketing investment directly to store-level P&L — email-attributed store revenue measured at $10.7M per week. The model used holdout groups and matched cohorts, was presented to and approved by the CFO, and ran on a weekly operating cadence. This wasn’t a marketing dashboard. It was a business operating tool.
The international architecture.
The loyalty system operated across 14 countries. GDPR compliance was designed market-by-market: annual list purge protocols in France, non-monetary reward structures in Germany and Italy where points-for-discounts were restricted. A unified global data model was maintained throughout.
Results
The Insight
The loyalty program was the asset — all of it. And only because it was built around the customer, not the transaction.
The program, the media network, the unified data, the behavioral signal, the customer relationship, the balance-sheet treatment — six facets of one thing. What made any of it compound was that the organizing discipline was customer lifetime value, not transaction volume, points liability, or accounting treatment. Without CLV as the anchor, a 65-million-member program is a very expensive coupon schedule with a database attached.
With CLV as the anchor, behavioral signals, purchase history, and platform identity unify into a single profile that is used to serve the customer — not just to count the transaction. Loyalty stops being a marketing program and starts being a revenue system. The CFO sees it in the P&L. Media partners pay for access to the audience. The customer experiences relevance, not promotion.
The broader lesson: structural disruption accelerates the advantage of precision. When physical game sales decline 3–5% a year, growth comes from knowing customers well enough to build what they’ll actually come back for.
That system — the architecture, the attribution, the operating model — is what this work is about.
Ran as a 34-person growth, CRM, and loyalty organization from 2018 through 2021 — $40M operating budget, $200M annual media investment, through a global pandemic and a short squeeze.
Case Study 02
Dick’s Sporting Goods × TaylorMade
Cross-brand identity bridge • Salesforce-based loyalty • Dual-KPI design
Built an identity bridge between two major brands using a single QR scan. A TaylorMade product at Dick’s sent the customer into a TaylorMade-controlled brand experience — including a window into the Kingdom, TaylorMade’s premier custom driver facility in Southern California.
Dick’s got a premium brand experience for their customer without building it. TaylorMade got a direct customer relationship and a path to custom driver sales. Same customer, same scan, two entirely different definitions of success — both achieved.
3x
member engagement
250%
click-through lift
10x
custom product orders
Case Study 03
Canada Post: B2B Churn Prediction
Predictive modeling • Retention strategy • Intuition-to-evidence
Built a scored, testable churn prediction model on event-level data — transaction history, service interactions, engagement patterns, and account behavior signals. The model surfaced a completely different set of churn predictors than the sales team expected. Some of the variables the team was most confident about had weak predictive power. Others nobody was tracking turned out to be among the strongest signals.
The output was a productized churn scoring feed that flagged at-risk accounts with enough lead time for targeted save campaigns. Instead of blanket retention offers, the team could focus resources on the accounts most worth saving, with the interventions most likely to work. The lesson: intuition scales poorly. Scored models do not.
Next step
Build something like this.
If you’re building the customer revenue system at your organization — loyalty, identity, retail media, attribution — let’s talk. Advisory engagements and leadership conversations both start here.