Case 02 · Nike México
Nike i3 Labs
Nike investing in its partners' digital performance.
- Role
- UX & CRO Designer
- Client
- Nike México
- Timeline
- 2019 — 2021
- Partners
- Innovasport · Invictus · Coppel · Liverpool
The challenge
One optimization loop, four very different storefronts.
Mexican retail partners optimized through structured CRO experiments
Users tested in a single Innovasport experiment alone
Order conversion rate lift — Innovasport, at 99% confidence
Estimated incremental annual revenue from one experiment
Context
Nike i3 Labs was a CRO investment program in which Nike México funded its domestic retail partners to run structured UX experiments aimed at improving conversion rate and average order value on their e-commerce platforms.
The partners — Innovasport, Invictus, Coppel and Liverpool — operated independent storefronts selling Nike product. Nike's strategic insight was that lifting digital performance across the partner ecosystem would multiply sell-through and brand revenue at scale.
My team ran the experience optimization work: mapping funnel friction, forming data-grounded hypotheses, designing variants, coordinating releases and measuring impact. Every experiment was reported back with full methodology and business impact documentation.
My contribution
- Identified conversion friction points in each partner's e-commerce funnel
- Designed A/B variant concepts and the UX execution for every experiment
- Documented experiment results and reported findings to Nike and each partner
- Worked within the global accounts team supporting Nike's investment program
How it happened
The work, phase by phase.
A repeatable six-step optimization loop
The same loop ran on every storefront, which is what made results comparable and the program defensible to both Nike and its partners.
- Observe — funnel data, micro and macro conversion per step
- Hypothesize — grounded in behavioral data, analytics and heuristics
- Design — variant concept, success metrics, effort vs. impact
- Release — A/B experiment in production with statistical rigor
- Measure — conversion, AOV, revenue per visitor, confidence level
- Iterate — report to Nike and the partner, feed the next cycle
Sticky PLP filters — Innovasport
Hypothesis: making filter and sort CTAs sticky on product listing pages — so users can filter regardless of scroll position — increases filter use and improves order conversion and average order value.
+3.32% order conversion
2.59% → 2.68% at 99% confidence.
+6.32% average order value
At 99% confidence.
+8.10% revenue per visitor
Conversion and AOV compounding together.
+4.90% filter or sort activation
The behavior change the variant was designed to unlock.
MX$19.9M estimated annual revenue
Roughly US$934K, projected from the measured lift.
Deliverables
Experiments across Invictus, Coppel and Liverpool
Each partner had a different shopping context, so each hypothesis was framed specifically. Detailed results for these three remain confidential.
Invictus — PDP clarity & trust signals
Surfacing size guide, delivery promise and return policy directly on the product page instead of behind links, to reduce uncertainty at the decision moment.
Coppel — financing visibility at checkout
Bringing installment financing options into cart and checkout, not just the product page, to reduce sticker-price hesitation.
Liverpool — cart cross-sell & recovery
Contextual “complete the look” recommendations inside the cart to lift units per transaction and reduce abandonment.
Outcomes & impact
Program results at a glance.
Nike retail partners in Mexico optimized through structured experiments
Users tested in the Innovasport sticky filters experiment alone
Order conversion lift at 99% statistical confidence
Revenue per visitor improvement on Innovasport
Small UX friction has large revenue leverage
A sticky filter bar sounds trivial. At 1.6M visitors and +3.32% conversion the compounding effect is dramatic. The highest-impact changes often just remove friction from a behavior users were already attempting.
Confidence level is the whole argument
Every result was reported with statistical confidence. At 99% on all primary metrics the case to implement was unambiguous. Designing for measurability matters as much as designing the variant.
Partner diversity requires hypothesis discipline
A specialty athletic store and a mass-market checkout are different contexts. Framing hypotheses per partner prevented assuming learnings were universally portable.
Next case
Santander México — Nómina DigitalOnboarding Mexico's workforce onto digital banking — no branch required.