Product Leadership

I build products
people actually use.

10-year product leader at the intersection of data, consumer products, and monetization. This isn't a resume rehash — it's a look at how I think, decide, and ship.

↳ Gap Inc — Sr Director, Customer Journey Nike Pro Football Focus Coca-Cola
See My Work →
Marty Styles

Product at the hard intersections.

I've spent my career at the intersection of data, consumer products, and monetization — most recently as Sr Director of Customer Journey in Ecommerce at Gap Inc, and before that building AI-powered features and betting tools at Pro Football Focus and scaling checkout and launch experiences at Nike.

I manage product managers and UX designers across multiple industries and specialize in turning complex systems into experiences that users actually come back to.

My approach: figure out where the real leverage is, cut the noise, and ship things that move the metric that matters. I've done this at global scale — Nike.com's checkout across 20+ markets — and at startup speed, like PFF's prop tool built on a tight timeline to drive a measurable lift in conversions.

10+
Years leading product across ecommerce, AI, and consumer apps
20+
Global markets shipped across Nike and Gap Inc ecommerce
58%
Hit rate on AI-powered prop recommendations at PFF

A few high-impact projects.

The work that shows how I think, the calls I made, and what shipped.

Pro Football Focus Betting · AI
Player Prop Tool — Building a Monetization Differentiator
58%
Hit rate on recommendations
+5%
Free-to-paid conversion
+15%
Session frequency
Read case study +

Context

PFF's betting product lacked a differentiated monetization angle in a crowded market. DFS and sports betting were exploding, but most tools gave users raw stats with no actionable edge. We had the best data in the industry — we just weren't surfacing it in a way that drove conversion or retention.

The Move

I identified that player props were an underserved format — high engagement, high repeat usage, and perfectly matched to PFF's grading strengths. Killed the "just add a props tab" proposal in favor of a standalone experience built around a clear hit-rate story.

Key Decisions

  • Chose PFF's proprietary grades as the core differentiator — not odds data, which competitors already had
  • Pushed data science to focus on hit rate as the north star metric, not volume of picks
  • Built for casual bettors, not sharps — intentionally simplified the UI to reduce friction
  • Navigated stakeholder tension between editorial (who wanted to own recommendations) and product (who wanted an algorithmic approach)

Impact

58%
Hit rate on prop recommendations
+5%
Conversion, free to paid tier
+15%
Session frequency among prop tool users

The hit rate number became the product's entire story. In a trust-deficient category like sports betting, a single verifiable outcome metric is worth more than 10 feature bullets. If I did it again, I'd have instrumented hit-rate tracking earlier so we could build the marketing story in parallel with the product.

Pro Football Focus AI · LLM
AI Key Insights — LLM-Powered Product at Scale
+8%
Time on page
+15%
Return visit rate
Core
Sales differentiator
Read case study +

Context

PFF had a goldmine of football intelligence locked inside analyst brains and proprietary grades. Users were getting the data, but not the so what. LLMs had matured to the point where generating credible, contextual insight at scale was actually feasible. The question wasn't if — it was how to do it without turning the product into a generic AI chatbot.

The Move

Scoped and led the build of AI Key Insights: a feature that generates contextual, grade-backed narrative summaries for players and matchups. The key call was to constrain the LLM to PFF's own data — no hallucinations, no generic takes. Every output had to be anchored in a real PFF grade or stat.

Key Decisions

  • Designed prompt architecture to anchor all outputs in structured PFF data before the LLM generated language — hard constraint, not a suggestion
  • Chose breadth over depth for launch: coverage across all NFL players vs. deep dive on a few — drove broader engagement faster
  • Built a review/flagging layer for the first 30 days post-launch to manage editorial team concerns about AI-generated content
  • Partnered with data science to define quality thresholds before shipping — rejected outputs below confidence score

Impact

+8%
Time on page for AI Insights pages
+15%
Return visit rate among exposed users
Core
Feature in subscription sales materials

The hardest problem wasn't the LLM — it was defining "good." Building that evaluation framework before shipping would have made QA faster and stakeholder buy-in easier. The editorial team's skepticism turned into advocacy once they saw the quality bar. Getting them involved earlier would have saved weeks of alignment.

Nike Global Ecommerce
Nike Checkout — Global Scale & Conversion Optimization
+3.5%
Global checkout conversion
-20%
Payment step abandonment
20+
Markets shipped
Read case study +

Context

Nike.com's checkout handled hundreds of millions of transactions globally but had accumulated years of technical debt and regional inconsistencies. Conversion rates varied significantly across markets, and the underlying system made it expensive and slow to ship improvements. At Nike's scale, every basis point of checkout conversion had massive revenue impact.

The Move

Led product for checkout innovation and optimization with a focus on reducing friction in the critical path and standardizing across global markets. Prioritized the highest-traffic markets and highest drop-off steps over trying to fix everything — a constraint-based approach that forced ruthless prioritization.

Key Decisions

  • Mapped the full checkout funnel across markets to find where drop-off was highest vs. where it was actually fixable — not the same list
  • Pushed for a shared component library over one-off regional builds — created short-term tension but paid off in shipping velocity
  • Navigated a complex stakeholder landscape: payments, legal/compliance, regional ops, engineering, and brand all had a seat at the table
  • Ran rapid A/B tests on highest-impact steps before committing to full rollouts

Impact

+3.5%
Global checkout conversion rate
-20%
Abandonment at the payment step
20+
Markets on a consolidated codebase

At Nike's scale, small UX wins compound into enormous revenue impact. The move that unlocked everything was creating a shared north-star metric — checkout conversion rate, segmented by market — that every team could see and rally around. Without that shared scoreboard, everyone was optimizing for their own thing.

Nike Consumer Mobile · Growth
Nike App — Product Launch Experience
+15%
Launch day stability
+3%
Post-launch retention
+20%
App annual revenue growth
Read case study +

Context

Nike's app product launches — sneaker drops, member-exclusive releases, SNKRS — were high-visibility, high-pressure moments. Millions of users would hit the app simultaneously during a launch. Technical failures or a bad lottery experience were immediate brand problems. At the same time, the launch experience was a major driver of app engagement and member acquisition.

The Move

Owned the product experience for launch flows — entry, queue, and confirmation states in the Nike App. The key strategic call was to treat product launches as a loyalty moment, not just a transaction. That reframe changed how we thought about the experience for users who didn't win the lottery, not just those who did.

Key Decisions

  • Redesigned the post-loss state to retain non-winners rather than letting them churn — introduced personalized recommendations and early access to the next drop
  • Simplified the entry flow after data showed drop-off at confirmation was higher than at entry — people were bailing because the process felt uncertain
  • Worked with engineering to build load-testing protocols that could simulate peak launch traffic before a major Jordan release
  • Balanced the brand team's desire for premium visual treatment with the performance constraints of a high-concurrency environment

Impact

+15%
Launch day stability (crash rate reduction)
+3%
Post-launch retention for non-winning users
+20%
Nike app annual revenue growth

The most underrated move was focusing on the losing experience. 95%+ of launch participants don't get the shoe. Treating that moment as a retention opportunity instead of a dead end changed the product's impact on LTV. Finding leverage in the "empty state" is something I now apply to every product I build.

How I think about AI products.

Not a framework deck. Not buzzwords. This is how I actually evaluate whether to build an AI feature — and why most of them fail.

01

LLMs matter when the data exists but the synthesis doesn't.

Users have access to data but can't process it fast enough to act on it. LLMs collapse time-to-insight from hours to seconds. That's where I built AI Key Insights at PFF — the grades existed, the language layer was the missing piece. Generic LLMs produce generic output. The moat is always in what you feed them.

02

AI should be a conversion lever, not just a feature.

AI features have a monetization problem: they often make the free tier too good. The right move is to surface AI outputs at the exact moment a user hits a wall in the free tier. At PFF, the prop tool's 58% hit rate wasn't just a product stat — it was the monetization story. That number made the paywall feel like a no-brainer.

03

Most AI products fail because they ship a demo, not a product.

The demo works because someone hand-curated the prompt and cherry-picked the output. The product fails because real users generate edge cases at scale and the system breaks in ways that erode trust fast. If you can't define what a bad output looks like before launch, you'll ship bad outputs. No quality floor defined = no product.

Got a hard problem?

I'm always open to interesting conversations about product, ecommerce, and AI. Reach out directly.

marty.styles@88consulting.co