Director, Product Management, Marketing Technologies (2020 to 2022)
How I got to Gap is part of the story. By early 2020 I needed to be home. My son Cole, born with microcephaly, was growing up, and I wanted work that was local and off the airplanes. Over a lunch, my very first manager from my consulting days, by then a Gap executive, told me he could not offer a big title but had hard problems that fit my skills and a role that would keep me near my family. After years as a CMO and then a chief strategy officer, I took a Director’s title on purpose, knowing exactly what I was trading and why. The Elastic Path chapter tells that part in full.
I joined Gap Inc. in June 2020, in the strange middle of a pandemic, to bring “personalization at scale,” the idea McKinsey framed in Marketing’s Holy Grail: Digital personalization at scale, to life inside one of America’s largest apparel retailers. After years building and selling commerce software to retailers, this was the other chair: being the retailer, in-house at a $16 billion Fortune 500 company with four iconic brands, Gap, Old Navy, Banana Republic, and Athleta, thousands of stores, and about 170 million known customers. The company thought about digital the way a merchant does; the deck they hand you in your first week put it plainly, think of the online business as our biggest store, the homepage its window, the product page its sales associate. The challenge I was handed was the unglamorous half nobody writes white papers about: make it real inside a company with all that surface area and half a century of retail habits. It is the marketer’s version of John Wanamaker’s old lament that half his advertising spend was wasted and he could never tell which half. In retail, where margins are thin and a season turns on getting the right product in front of the right customer, that problem is about as concrete as it gets.
What drew me was the starting position. Plenty of companies talk about personalization without the raw material to do it; Gap had the opposite problem. It already had a rich customer file, strong instrumentation, and a data-science function doing sophisticated marketing-mix modeling, and now, finally, it had the budget and the appetite. A new CEO had laid out Power Plan 2023, and just before I arrived the company had called off its plan to spin Old Navy into a separate company, committing to run the brands together the way it largely already did, on shared infrastructure down to a universal shopping cart. That set the stage for investing in one digital platform for all four brands. The pieces were on the table. The job was to assemble them.
Where I sat, and what I found
My team was a product group in the seam between GapTech, the company’s technology organization, and the brands themselves. That seam turned out to be the most important place to stand, because the problem was a seam problem. I led three product managers, and the leverage came less from headcount than from driving the platform decisions that dozens of brand and central teams built on top of.
What I found was a mature martech stack, only bigger: not a dozen tools but dozens upon dozens of vendor systems across marketing and commerce, unevenly adopted, stitched together by slow manual work, and, tellingly, no central content management system at all. Content lived wherever it had landed: a Box folder here, a SharePoint site there, a chain of email attachments. The paradox took a few months to see clearly. This was not a company that failed to personalize; by its own public measure, roughly three quarters of site visits and eight in ten emails already carried personalized content. As Gap put it plainly in its own investor materials, the coverage was strong, and the sophistication and coordination were the next opportunity. It had once been ahead of the curve and was now a step behind a few of its peers, sitting on excellent data science and years of genuinely sophisticated targeting, but mostly at the level of affinity groups and segments. We could know a customer richly and still serve them roughly the same experience as everyone else in their segment. The leap in front of us was from the segment to the individual, true one to one, acting on what we knew about a single shopper in the moment of the visit. That distance, between how well we could know a customer and how completely we could act on it at the moment that mattered, is the gap I spent two years closing. It is also the thread that runs through everything I have built since.
Choosing the platform, and choosing against Adobe
My first job was selecting a new content management system, and it put me in a seat I had rarely occupied. I had spent most of my career on the vendor and consulting side, the one pitching; now I was the buyer, responsible for a choice the whole company would live with. We did it properly: more than thirty interviews across five brands to map the real pain before we watched a single demo, then evaluations over several months tested against how work actually flowed. The answer was a modern, headless platform, and we chose Amplience.
The twist was Adobe. The assumption going in was that we would land on Adobe Experience Manager, the establishment choice, and Adobe was my own former employer. I had been on the Adobe.com team when Adobe acquired the company whose technology became that product, so I knew it from the inside, its strengths and its limits for what Gap actually needed. And it was genuinely close: after the first scoring round the two finished as co-leaders, and one of our four brands preferred Adobe to the end. So we let evidence break the tie. We gave our own developers sandboxes in both platforms and ran a final round of demonstrations across dozens of evaluators from every discipline, and the result was decisive. AEM was, at heart, an experience suite built to own the glass, still evolving toward headless, while Gap needed a headless hub from day one, serving many different sites and apps. Our engineers saw a faster path on a platform that matched the storefront they had just built rather than one that asked them to retrofit it. And the winner’s retail-grade scheduling and content calendaring, sharpened with retailers like Crate & Barrel, fit the way our producers actually worked better than anything Adobe could show us. We even sat down with Crate & Barrel ourselves to hear what the journey had done for them. Recommending against my own former employer, on the merits and from inside knowledge of its product, was the harder call and the right one.
What I would do differently
Here is the part I got wrong. I moved too fast. I came in with deep software-selection experience, from my Accenture years and from the buyer’s side at Adobe, and I wanted a quick win on a decision that was already in flight. What I underestimated, in a company that large, was how little alignment actually existed underneath the mandate. GapTech and the digital teams had other content platforms in use and under construction, and there were real differences of opinion about how far this new system should reach. The quick start was followed by friction that slowed us down later, friction I could have eased by spending my first months learning every stakeholder and what they were really optimizing for before I drove for a decision. At that scale, that patience is not overhead. It is the work. I relearned the same lesson with our design system, which I wish I had reached for sooner instead of first trying to standardize components by force across brands fighting to look nothing like each other.
From weeks to days
Speed was the promise. The example that stuck with me was a producer who needed a landing page live in a few hours to chase a fast-selling item, working against a system that took six to seven weeks to give it to her and capped the site at two publishes a day. That gap was the whole problem in a single sentence. The new platform collapsed that toward near-instant publishing, no engineer in the loop, the same building blocks reused across site, email, and landing pages. We turned a process measured in weeks into one measured in days, fast enough that a producer handed fresh content in the morning could have a landing page live that afternoon, and a brand could react to a fast-selling item or a breaking trend mid-season instead of watching the window close. That is not a back-office win in retail. It is a merchandising one.
We proved it on a live call where a producer changed an Athleta landing page in real time while everyone watched, and the relief in the room was real, because these were people who had been getting by on workarounds. It was not a brand-new capability; teams had done things like it before and lost it somewhere in the move to a new React storefront. To cope in the meantime, they had pressed a personalization and recommendation tool into service as a makeshift content manager, forcing changes onto pages whose normal schedule was too slow. Getting real controls back felt like a homecoming. What none of us clocked yet was what that workaround had been doing to how fast the page loaded. Hold onto that detail. It comes back.
Rolling the platform into a four-brand retailer was the real work, and we did it deliberately: a pilot, then a content structure that would hold for every brand, then migrating brand by brand and section by section, so we never bet a holiday season on a big-bang cutover. Only then did I turn to the rest of the stack, running the same disciplined process for campaign management and the customer-data layer while a peer drove the parallel work on our recommendation and search tools. Each time the pattern repeated: find out what we were actually paying for, separate best-in-class from merely incumbent, and recommend what deserved to stay, what should be replaced, and what we could live without.
Building the engine, and a privacy-first bet
By the time I turned to personalization, the portfolio question had a voice. The build-minded SVP of Product I worked under, who trusted outside vendors only when they were exceptional, asked it plainly: we spend millions on martech, do we actually need all of it? The honest answer was no. He handed me a quiet, off-the-books mission to go with the official one: keep watch over everything we owned, and keep finding the things we did not need. So the personalization work carried a parallel mandate, question what we had while building what we lacked, earning a skeptic’s trust one defensible recommendation at a time.
The build itself was a real-time personalization engine that, the moment someone landed on a site or opened the app, pulled from our key systems to tailor what they saw to their loyalty level and purchase history. We thought about it as a ladder of recognition, from an anonymous visitor to a recognized returning shopper to an authenticated cardmember, with more of the experience unlocking at each rung as we earned the right to know them. In practice that meant small, specific moments of relevance instead of a generic storefront: greeting a returning cardmember with the loyalty rewards they had waiting to spend, or lifting a product’s best customer-review highlights onto a brand’s homepage. It was deliberately a blend of commercial platforms and internal systems rather than a single product. One thing is a point of pride about Gap, and worth stating exactly: we were not starting from nothing. Years of embedding data scientists directly in the teams meant the company already ran a genuinely sophisticated homegrown identity and customer-data capability. My goal was not to invent identity resolution but to find a commercial system that could beat ours at lower cost, freeing engineering from maintaining a legacy database. That was the one piece still in flight when I left, a deliberate build-versus-buy call, not a gap in capability.
None of this was invented in a vacuum. The whole program was Gap operationalizing McKinsey’s personalization-at-scale framework, whose research put the opportunity at a five to fifteen percent revenue lift, the promise that justified the investment. The cross-functional “pods” that ran experiments and scaled the winners came straight from that playbook. My real work was turning a framework on a page into a roadmap a Fortune 500 company could actually run, and earning each win the only honest way, through controlled A/B tests, letting what worked earn its way into the always-on experience.
It was also a bet on where the industry was going. Third-party cookies were dying and privacy rules were tightening, so the advantage was shifting to companies that could earn and use first-party data well, and a retailer with a deep, responsibly used customer file held a better hand than it realized. That is why privacy was strategy for me, not a compliance afterthought. I think of personalization as a barter: a customer trades data for an easier experience, and the trade only holds if they get real transparency and real, easy control. Gap took that seriously enough to adopt formal privacy principles to guide every project, and I wrote about the philosophy publicly, including a guest lecture at the University of Virginia.
Atomic Design: one system, four brands
A multi-brand company on a headless CMS has a design problem hiding inside its technology choice. Four distinct brands each have to look unmistakably like themselves while dozens of teams build without colliding, all on shared infrastructure that made a single rollout genuinely hard. The answer my experience-design leadership proposed, and I embraced and evangelized, was Atomic Design, Brad Frost’s method of composing pages from small reusable parts, with a target of roughly eighty percent shared components and twenty percent custom brand moments. The lasting lesson was that a design system is not a tool you install but, as the team put it, a team sport. It lives or dies on governance and adoption, so getting an organization to genuinely use a shared system rather than quietly build around it was harder, and more rewarding, than any of the technology.
What I took from it
Gap was where I learned retail from the inside, not as a vendor selling to it but as the company on the hook for the season: omni-channel, multi-brand, loyalty-driven, with a relentless calendar and a store on the other side of every digital decision. It is where I learned the buyer’s side of the table for good, where I learned that the real constraint in a sophisticated retailer is rarely the data or the budget but the seams between the people who know the customer and the systems that touch the customer, and where I learned, the hard way, that in a big company you earn the right to move fast by first understanding what everyone around you is trying to win. Close the seams, with discipline about what is worth keeping and a real respect for the customer’s privacy, and the rest follows.
The thread to what came next
There is one more thing Gap gave me, almost by accident. Remember the team using a recommendation engine as a makeshift CMS? That was one small symptom of something I slowly could not unsee. The Old Navy homepage, at one point, took nine seconds to render for a real shopper, buried under third-party scripts, and the strangest part was that our own monitoring never caught it. New Relic watched the React storefront but not the third-party JavaScript that was actually breaking the experience, so by its numbers everything looked fine. I went to New Relic and to Splunk for help, and at the time neither knew what to do. I did not have the answer either; I had not yet learned the name for it. What I could do was prove it, with synthetic tools like Pingdom and Catchpoint, and the lesson stuck: all the personalization in the world is worthless if the page does not load before the shopper moves on. It was the same gap I had been closing all along, now in its most stubborn form, the page itself. I left Gap and went straight to Yottaa, a company built to solve exactly that, and recognized the pattern instantly in its retail customers’ data, because I had only just lived it from the other side. But that is the next story.
This chapter is part of My Work, my career told one company at a time.