Cashier Performance Scorecard
Nature of product/system: Retail point of sale (POS) reporting system
Persona: Retail store managers
Stage at which I join the project: Near the start of the contract competition
My role: (Primary) UX Strategist (Secondary) UX Interaction designer and researcher
Type of research: Formative and Summative, Qualitative, Business rule definition and user acceptance
Research Technique(s): In-person interviews and prototype evaluations
What was the problem?
My company, NCR, was in a contract competition to replace all of Home Depot’s point of sale terminals. As part of this, Home Depot’s CEO wanted each team to include reports that measured the value of their purchase. The competitors were asked to design cashier performance reports, based solely on data from the POS terminals.
The Teradata division of NCR led this side project. They had lots of experience with analytics for the retail industry. However, their users were all data analysts, people who understood their data intimately, designed their own reports, and spent much of their day gleaning insights from them.
On this system, the users were store managers who would likely allot only 10-20 seconds to reports per week, preferably less. Worry that Teradata didn’t know how to create quickly actionable reports lead the team to add me to the project.
As requirements discussions progressed, it became clear I was the team member best suited to lead conversations around:
The store manager’s needs
Cashier user scenarios,
How to develop a scorecard algorithm that was fair and accurate, yet resistant to cashier manipulation.
What were the stakeholder assumptions?
From POS data alone, we could design reports that accurately reflected cashier speed and accuracy, and indicate the possibility of stealing.
We could develop reports that store managers could quickly comprehend and would use consistently.
The reports could be accurate, and fair enough cashier employees would accept them.
How did you determine the forms of research to conduct?
The discovery process and the personnel we could work with was laid out by Home Depot Operations executives.
Which research method(s) did you choose and why?
Contract competition phase
This project followed a waterfall methodology, so our options were limited.
In working sessions with Home Depot operations personnel, the project to discuss:
How cashier performance is taught and evaluated at present
The behaviors they wanted to measure and incent with the new system
The metrics that could be derived from POS data that accurately reflected those behaviors
In our proposal, we laid out how we would design the scoring algorithm and reports.
After winning the contract
We dug into the details with Operations and store managers to determine:
The balance of cashier behaviors the client thought of as optimal.
How triggers indicating the need for cashier retraining would be set
How we could construct the scoring model to minimize cashiers’ ability to “game the system.”
In parallel, I developed report designs that presented the weekly results. I paid particular attention to making the reports as quick to read and interpret as possible. As soon as practical, we produced reports from live POS data and confirmed with store managers that the reports were:
Easy to scan visually
Easy to understand
Accurately presented their cashiers’ performance.
How were your recommendations tied to both user and business value?
The objective of this project was to give store managers the information they needed to improve their checkout operations.
Did you iterate and improve each cycle?
The development of the report algorithm and report design went through numerous iterations as potential issues surfaced. Once Store Operations executives were satisfied, we conducted one formative and one summative feedback exercise with selected store managers.
How was your research received?
Both major stakeholder groups (HQ Operations and store managers) accepted the findings of our work sessions and store manager feedback.
How did your results bring impact?
Our cashier scorecard proposal was called out as a significant factor in winning this major equipment sale for NCR. Three weeks after the system was deployed, the accuracy and utility of the reports were confirmed by both store managers and cashiers (half of whom said they liked the reports). Anecdotally, one manager shared how the reports called out an employee who was later found to be stealing.
Within six months, average transaction times across Home Depot improved by 10%. To quote one of the help desk analysts who supported the system, “If I had these types of reports when I was a front end supervisor, I would have never left the stores.”
PUBLICATION
I presented a paper on the design principles behind this scorecard at the 2006 Human Factors Convention in 2006 in San Francisco.