The situation
Kroger was treating relevancy as one team’s job. The companies winning customers were treating it as infrastructure.
- The business problem. We were losing share of wallet among our most digitally engaged customers while Amazon, Costco, and Walmart grew. Part of the reason was that we couldn’t make the experience feel relevant.
- The org problem. Personalization, Promotion, and Monetization worked against each other. Nobody had defined “relevant,” and nobody could measure it. So teams built one-offs, and every one-off cost more to coordinate than the last.
- The customer problem. Our promotional content had stopped earning attention. Shoppers were direct about what they wanted: to feel known without feeling watched, and to save time and money without doing the work of hunting for it.
I was asked how to personalize better. I answered a different question: what makes an interaction feel relevant, and how do we deliver that without hand-building every one?
What I built
Two reframes did the heavy lifting.
- Personalization → relevancy. The word was too narrow and too loaded. Relevancy meant getting two dials right at once: how precisely you target, and how well you tell the story. Running them together at scale turned a data science problem into something the whole company could own.
- Strategy → operating system. A strategy document was never going to change how 2,400 people work. So I built three connected things.
The Dynamic Experience Matrix
Scroll sideways to see the full matrix.
Targeting ↑
Storytelling →
Storytelling →
RecommendGet what you came for.
InfluenceDiscover what you didn’t know you wanted.
InspireFeel known.
One to oneHousehold or account
HighYour usual order, rebuilt
HigherSwaps that fit this basket
PeakA plan built around you
One to manySegments and cohorts
ModerateCategory picks for a segment
ModerateSegment-aware promo rail
HighCollections for a way of living
One to allMass audience
BaselineBest sellers grid
LowSeasonal campaign banner
ModerateBrand story moment
Targeting on the vertical axis, storytelling on the horizontal. Value climbs toward the top right.
The Relevancy Decision Engine
What goes in
Customer signalsWhat they’ve told us, what they’ve shown us, what they’re doing right now.
Business signalsMerchandising priorities, promo plans, media-eligible inventory, strategic initiatives.
What’s inside
Data layerThe part everyone assumes already exists.
One to allBannerization
One to manySegmentation
One to onePersonalization
Ranking with explainabilitySo business needs shape content ranking upstream instead of fighting for placement downstream.
How teams use it
One place to compose, target, and shipNo custom build required.
Every channel it plugs intoSite and app, merchandising surfaces, retail media supply, marketing campaigns.
What comes out
A relevant experience across every channel — right offer, right customer, right moment.
Engagement data feeds back in, so the system gets smarter.
The Digital Experience Value Equation
What we earn
Experience performance
- Profitability lift from the experience
- Sentiment lift from the experience
−
What it costs
Cost to decide
- The coordination tax
- Time from idea to agreement
- How many times we re-argue the same decision
+
What it costs
Cost to deliver
- Hand-built hours
- Orchestration overhead
- One-off features somebody maintains forever
Nobody was measuring this
=
Realized value
One number two teams can argue about honestly, instead of four scorecards that never meet.
How I operated
- I challenged the brief before I solved it. A discovery, an audit, and a cross-functional Experience Vision Workshop gave me the evidence to tell senior leaders their framing was the problem. Moving the org from “a unique experience for every customer” to “target and tell the story together” meant calling the current model incomplete without dismissing the work that got us there.
- I put a price on the drag. Cost to Decide and Cost to Deliver were a deliberate provocation. Most metrics measure outcomes. I measured what bespoke work was costing us, because that was the thing killing our ability to scale.
- I said the uncomfortable part out loud, then offered a way forward. Personalization, Promotion, and Monetization were working against each other, and teams built one-offs because the system forced them to. That implicates everyone in the room, so I paired it with a path nobody had to lose face to take.
- I built for adoption. I stood up three workstreams — Foundational, Experiential, Cultural — and ran the whole thing like a program. I picked adoptability over analytical sophistication every time, because a framework that dies in the deck it was born in changes nothing. I spread ownership across a three-person leadership model from the start, and pitched relevancy to each executive sponsor in their own terms rather than presenting at them.
What happened
This work is still in flight, so what follows is reframes, funding, and adoption rather than booked results.
- The language changed. “Relevancy” replaced “personalization” as the shared term across Product, Design, Tech, Digital Experience, eComm, Data Science, Marketing, Merchandising, and Kroger Precision Marketing. The Dynamic Experience Matrix is vocabulary that didn’t exist at Kroger before this.
- We put a number on the cost nobody was tracking. Manual orchestration across Digital Experience, Site Production, and Digital Merchandising was running 200+ hours a week combined. That turned Cost to Deliver from a rhetorical device into a line item, and it’s the baseline the platform investment gets measured against.
- Teams started asking for the frameworks. Partner teams began requesting the Matrix for their own planning, which is the clearest proof the strategy works without me in the room.
- It got funded. The work became the Relevancy Program: millions in funding, owned departmental KRs, three workstreams, cross-functional co-leads, and sponsorship from Kroger’s Chief Data & AI Officer, GVP Customer Growth, and VP Customer Loyalty.
- I sized the opportunity honestly. The long-range plan models a 30% lift in promo redemption, a 12% AOV increase, and a 5–20% retention improvement, laddering to billions in value by 2030. These are models with stated assumptions, not results.
- We started proving value. Six A/B tests delivered against zones of the Matrix, with more in the queue.
Who owned what
Data Science owned the models and signal infrastructure. Digital Experience owned the customer-facing surfaces. Kroger Precision Marketing owned retail media supply and monetization. Experience Management Platforms owned the tooling teams would touch day to day.
I owned the shared frameworks and the strategic argument, plus the cross-workstream rollup, dependency management, and the shared accountability model. My co-leads on Relevancy Capabilities and Experience Management Platforms ran their own workstreams. My influence ran up to three C-suite and VP sponsors, across 16+ partner teams, and down to the domain leads adopting the Matrix.
What I’d do differently
- Stand up the value equation first. Its whole power is making Cost to Decide and Cost to Deliver visible, and I baselined it mid-flight. That means I proved the system’s value after asking teams to adopt it. I’d flip the order next time.
- Watch for reframe fatigue. I reframed twice: personalization to relevancy, then strategy to system. Both were right, and both cost partner teams real energy to absorb. I’d build one durable source of truth earlier, so the story can evolve without re-arguing the foundation every time.