Issue 9
Somewhere right now, an investigator is spending four hours on an alert that leads nowhere.
That time is gone. That dollar is gone.
And the actual fraud pattern that triggered nothing? Still growing.
Volume-based detection systems are looking for trees when they’re big and when they’re a clear problem. A Director of Loss Prevention covering 11,000+ locations with 10 people serving as coverage knows that below the big ones, a lot more seeds and saplings are spreading their roots.
By the time a pattern is big enough to trip a threshold, you’re looking at a forest of them. That director built a system that can identify those sneaker saplings, and pull them out before they wreak havoc.
But first, your industry brain teaser of the week:
What percentage of flagged work items should confirm real fraud or abuse in a well-tuned detection system?
Scroll to the bottom for the answer.
What's in stock
Here’s what we have in store for you this week:
- The Rundown: How changing your alert system can triple your case volume
- Worth Your Time: Retail leaders are betting big on physical stores this week, while powers both fraud detection and evasion
- What We’re Up To: We’re coming to NYC (and why nobody cares about “AI-powered”)
The Rundown
Who had the most refunds. Who had the most voids.
Most exception analytics surface things related to volume, but high volume isn’t a fraud or abuse pattern. It’s a starting point that burns investigator time before anyone finds a real case.
By the time a pattern is big enough to trip a volume threshold, it’s already been growing for months, and investigators are spending full days on alerts that don’t lead anywhere.
The director rebuilt their detection system around a different idea: fraud and abuse have a shape before they have a size. Five steps help them see what that shape is:
- Build channel-specific detection rules. Drive-through fraud works nothing like café fraud. The rules have to reflect that.
- Measure every rule’s effectiveness. Track outcomes on every work item. Rules that don’t confirm fraud or abuse get rebuilt or retired.
- Correlate labor data. Off-shift benefit abuse looks like a clean transaction unless you know the employee wasn’t scheduled.
- Prioritize the queue systematically. Always cut the highest-risk item in each category before moving to the next.
- Build regional dashboards. Work item counts by market show where POS risk is concentrated, and whether it’s coming down.
Worth Your Time
We know time is money, so we won’t waste yours
- Tecovas customers who shop in-store and online generate 32% higher lifetime value than single-channel shoppers. (Retail Brew)
- 92% of shoppers say AI-generated content makes it harder to tell if a site is real. (Chain Store Age)
- Counterfeiters are using AI to fake reviews and alter product images, making fraud harder to detect. (Modern Retail)
What we're up to
Appriss Roadshow NYC is July 30 at The Ned in Nomad. Half-day and built around one question: what is returns and shrink actually costing your business, and what would it take to change that?
The product team previews what’s coming in H2. And we’ll run your actual returns exposure number live in the room. Most retailers have never seen it.
Space is limited. Claim your spot.
Btw, I wrote something for CMO Alliance on why nobody cares that your product is “AI-powered” anymore — and what actually builds credibility with buyers who’ve heard it all. Read it here.
