Archon

The problem

Ten spreadsheets, and no room for an eleventh

Getting an item into an online assortment took a long chain of manual steps. An Assortment Manager pulled data from several systems, compiled it, emailed another team when an item needed a placeholder UPC, built a “wave file” for the core teams to review, and then tracked the item by hand until Item Setup processed it. All of it lived in Excel.

That worked until it didn’t. Kroger was opening new fulfillment centers, and each one meant more rows, more files, and more chances for a copy-paste error to reach a customer.

We literally can’t open another fulfillment center if we continue to use Excel. It’s completely maxed out.

Research

Earning the trust of people who’d been burned before

Our users had seen projects like this fail, and Excel was the devil they knew. Some of them had ended up on spreadsheets in the first place because nothing better existed. So before we designed much of anything, we spent months in discovery sessions with them: mapping flows and journeys, writing it all down, and checking our understanding against theirs.

It took a while. Kroger’s data architecture is a bowl of spaghetti. But by the time we had something to show, the Assortment Managers knew we understood their work, and that made every round of feedback after it more honest.

Testing

In front of users every two to four weeks

We tested on a steady rhythm for the life of the project. Usability tests showed us where people stumbled on real tasks. Preference tests settled questions when we had competing directions. User acceptance testing confirmed each release did what Assortment Managers needed before it shipped. The rhythm kept the build honest, and it made the skeptics part of shaping the tool.

The QA we didn’t have

For the first eight months, the project had no dedicated QA. Bugs and broken automations meant every step forward came with two steps back. My PM and I made the case for QA hires and automated testing, and we got two testers. Once they were up to speed, they had us back on the rails in short order.

The build

One place to run an assortment

Archon gives Assortment Managers a Product Library of sellable items, an assortment for each fulfillment center, and automation for the steps they used to do by hand, like moving an item from pending to active once its setup checks pass. It connects to the systems around it instead of re-keying their data.

Play video: Archon hype video

Play video: Archon: editing items in the Product Library

After launch

Creature comforts

Once Archon was live, we turned to quality of life: workspaces people could customize, keyboard shortcuts modeled on Excel’s, and Live Edit Mode for changing data right in the grid.

Play video: Archon: grid tab customization

What didn’t work

The shortcuts nobody used

Assortment Managers were Excel power users, so we assumed keyboard shortcuts would save them a lot of time. We mirrored Excel’s as closely as we could. Months after they shipped, follow-up testing showed almost nobody using them.

The timing was the problem. Our road to MVP ran longer than planned, and by the time shortcuts arrived, people had already built their Archon habits around the mouse. Habits form early. If a feature depends on changing how people work, it has to be there before they settle in.

Results

Off spreadsheets, and the source of truth

• 250 hours of work saved per Assortment Manager, per year.
• One team’s weekly process went from 5 hours to 10 minutes.
• 10 hand-kept spreadsheets retired.
• Assortment data opened up to 50+ view-only users.
• Fewer incidents needing an Assortment Manager to triage.
• Two new online fulfillment centers have opened since launch.

Today Archon is the source of truth for Kroger’s e-commerce assortment. Its data feeds downstream fulfillment experiences and APIs, and it’s named as an input to Plan-to-Shelf, Kroger’s roadmap for bringing assortment decisions and item readiness into one workflow.