Preventing in-store fraud
Microsoft Dynamics 365 Fraud Protection loss prevention is a cloud-based solution designed to help brick-and-mortar retailers decrease in-store fraud costs and improve operational efficiencies and revenue. Using the power of AI, loss prevention identifies anomalies and patterns that might be indicative of in-store fraud.
Overview
Brick and mortar stores face fraud based on the improper discounts or returns applied by the cashiers. It takes a lot of time to analyze the data to identify where and when this type of fraud was committed, and act on these findings.
Key areas to explore were:
What is the current process of identifying in-store fraud?
What challenges do loss prevention managers face?
What meaningful data can we provide to help our customers detect fraud early?
How can the tool be integrated with Dynamics 365 family and other products?
Business goal
Create an MVP for retailers to act on in-store fraud early on that will reduce their losses.
Role and duration
Lead designer for this feature. I worked closely with a PM, user researcher, data scientists, and developers.
April 2019
Research
We conducted user interviews that helped us better understand the persona, process, current challenges, and opportunities for our product.
Persona overview
The loss prevention manager is responsible for loss prevention for a set of sites, typically in a geographic region. Their focus is on completing investigations and audits for sites in their region, as well as coordinating with the central loss prevention organization. Many loss prevention managers are also responsible for other areas, for example, defining safety policies and trainings.
Key research findings and common pain-points
Identifying and investigating in-store fraud is a multi-phase process that might take several months.
AI / ML was not broadly used. Some of the teams we’ve spoken reliant on manual reporting processes to identify anomalies.
The main challenges were communicating ROI, budget constraints, and lack of investment.
Ease of use was important to some participants. They referred that the systems they were using as super-complicated, and they needed to hire data analysts to navigate them.
Making sense of data
After a round of research, I closely collaborated with data scientists on the metrics we can provide for our customers. With the powerful (and sometimes complex) capabilities we needed to tell a straightforward story. The story which will immediately show the value of this tool and then lead to clean, understandable, and actionable data. I started off by digging into the data provided by the data science team and asked a lot of questions. I wanted to make sure I understood each available metric and the intention behind it.
Guiding through data
I had multiple rounds of iteration before we landed with the final deliverables.
Main goals that I wanted to achieve:
Show the value to the customer right away.
Provide a high-level view and then guide to an in-depth investigation.
Make sure the concepts and new terminology are clearly explained.
Provide a path for data import.
Bird’s-eye view
The main purpose of this report is to give an overview of the in-store fraud patterns, what the trends are, and what employees and terminals have the high-risk score.
In-depth view
This page gives more details about a specific employee or terminal, starting from the overview and moving to the actual transactions that affected the risk score.
Getting data into the system
The feature obviously doesn’t have any value without customer data. For the first version of this feature, we provide two integration options: connecting to Dynamics 365 family and manual data upload. Connecting to Dynamics is a simple task and provides ongoing sync. Although the second option might seem as cumbersome, it gives an opportunity for the non-Dynamics customers already to use this feature. Since auditing of in-store loss is not a daily task, uploading data manually won’t be frequent.
See in action
Customer feedback
By identifying fraud and loss quickly with Dynamics 365 Fraud Protection, we’re boosting business outcomes and building a culture where our core value of honesty is practiced by everyone every day.
- Kim Guldager, Senior Program Manager at BESTSELLER
Conclusion
Creating an MVP with limited time and resources is never an easy task. And, sure, the first version is never the perfect one and never the last one. However, with this feature, we expanded the capabilities of our product and received positive customer feedback. The most important thing for this version was to lay the foundation for scaling capabilities. And with more feedback from our customers and research participants, we will learn more about their needs and iterate to help our users do their job fast and successfully.