Company
Kroger
Role
Product Design Lead
Scope
Enterprise Cost Data Platform
Impact
Adopted across Merch, Finance & Sourcing
Timeline
2022 — 2023

A single source of truth for Kroger's category cost decisions.

A unified cost data model that replaced spreadsheet trackers and tool-by-tool reconciliation — and spread beyond the team it was built for.

Kroger cost data table — editing view

In low-margin grocery, every penny on a SKU matters. But Kroger's category managers were making cost decisions from data scattered across procurement systems, regional pricing tools, and personal spreadsheets — each with its own version of the truth.

I led design on a unified cost data platform: one model, one source, one place where category managers, finance, and sourcing could see the same numbers and act on them in time.

The Problem

Cost visibility wasn't the problem. Decision clarity was.

01
Costs that didn't tell the whole story
The net costs buyers saw didn't include back-end logistics or regional distribution variances. Margins looked healthier than they actually were — and the gap only surfaced after a deal had closed.
02
Negotiations that were always one step behind
By the time a cost increase was identified, analyzed, and routed through the right channels, the window to push back or substitute an SKU had already closed.
03
Different teams, different numbers
Merchandising and Finance pulled cost data from different sources. Meetings became arguments about whose number was right instead of what to do about it.
04
More time cleaning data than using it
Buyers spent the majority of their day stitching spreadsheets together — reconciling, reformatting, copy-pasting between tools. The actual decision-making got squeezed into what was left.

The fix wasn't another dashboard.
It was a shared model.

The Approach

Designing for the question, not the spreadsheet.

I started by sitting with category managers as they did their actual work — pulling cost reports, reconciling them against vendor agreements, exporting to Excel, layering in their own assumptions. The same moves, every Monday.

The new model wasn't a redesign of the old reports. It was a reframe: instead of optimizing for "what does the data say," we optimized for "what decision are you about to make, and what do you need to make it." Filters, defaults, and sort order all flowed from that question.

Kroger cost data filters
Unified cost data table — final view
The unified cost view. One table. One source. One set of numbers everyone could trust.
Component Detail

Historical cost, made legible at a glance.

Historical cost change view
1
One timeline, not five tools Past, current, and upcoming cost changes in a single view — no jumping between date-range reports to reconstruct a pattern.
2
Change reasons in line Why a cost moved — vendor, freight, promo — sits next to when it moved, so the story reads top-to-bottom without a second screen.
3
Forward-looking by default Upcoming changes get the same visual weight as historical ones. Buyers stop reacting to cost shifts after the fact.
One source.
One number.
Merchandising, Finance, and Sourcing stopped arguing about whose data was right and started talking about what to do.
Kroger Cost Data View
Cross-team adoption diagram
Adoption Beyond the Brief

An architecture win, not a feature win.

Each adopting team came looking for something different — forecast assumptions, vendor review prep, planning-meeting context. None of it was on the roadmap. None of it needed new screens or new permissions.

That's the kind of validation you can't ask for. It only shows up when the model has abstracted the right things — separated cost from interpretation, fact from judgment — so the same source can serve very different decision contexts without bending.

Shipped in 2023.
Revisited in 2026 with AI.

What this project would look like if I shipped it today.

Claude
Figma Make
Retrospective

After Kroger launched, I went back and re-ran two parts of the work with AI in the loop — the research synthesis and the prototype iteration — to see where it would have actually moved the needle.

Claude was strongest on the data side: clustering hundreds of buyer interview notes and cost-change logs into recurring patterns, and drafting plain-language explanations for why a cost moved (the kind of context buyers used to have to dig out manually).

Figma Make rebuilt the historical view as an interactive prototype in an afternoon. The kind of "what if upcoming changes had the same visual weight as historical?" exploration that used to take a week of static comps now happens in real time, with real data shape.

Kroger AI
The Outcome
One source of truth
Replaced fragmented procurement reports and personal spreadsheets with a single canonical cost model across teams.
Adopted beyond the brief
Picked up by Finance, Sourcing, and Merchandising leadership — none of whom were the original audience.
Past + future, one view
Historical, current, and upcoming cost changes consolidated into one timeline — no more reconstructing patterns from date-range reports.
Reflection

The hardest part of this project wasn't the data architecture — it was getting three teams to agree on what the number meant before agreeing on what it said. Cost data is a coordination problem dressed up as a technical one.