Skip to Main Content

What your last AI vendor for returns evaluation missed

Enterprise retailers are funding AI that reports on fraud instead of stopping it. Patel on what the evaluation should ask instead.

Issue 11

A doctor with the correct scan but the wrong label will end up prescribing the wrong treatment plan.

Most enterprise retailers evaluating AI-powered returns management vendors have the result. They’re mislabeling the diagnosis and funding vendors accordingly.

Vishal Patel, Appriss Retail’s Chief Product and AI Officer, has spent a decade building AI that makes real-time, omnichannel decisions across 40% of all U.S. retail transactions. His read on where enterprise AI buying misses the mark is that the wrong diagnostic is driving the budget.

But first, your industry brain teaser of the week: 

Online return rates are nearly three times higher than in-store. What’s the in-store return rate?

Scroll to the bottom for the answer.

What’s in stock

What’s in stock

Here’s what we have in store for you this week:

  • The Rundown: The three criteria that predict whether an AI investment moves your return rate
  • Worth Your Time: What finance and LP leaders are reading on AI, returns, and measurable value
  • What We’re Up To:  Ain’t no party like a Retail Club party

The Rundown

The Rundown

Most AI vendor evaluations end with the wrong question answered. The pitch deck says “AI-powered.” The compliance team signs off and the system goes live. And six months later, the return rate hasn’t moved.

It’s a pattern Patel sees across the industry: by the time the post-mortem happens, nobody can measure the success criteria that were supposed to be tied to the contract.

Three criteria separate AI investments that show up in the P&L from ones that stall in pilot:

  • Complete omnichannel picture. Does the system score based on a unified customer identity that spans in-store, online, and customer service center behavior — or a partial one?
  • Determinism at the decision layer. Is the model making the risk call auditable and reproducible, or is a human still in the loop at every transaction?
  • Financial success defined before deployment. The right conversation isn’t about model accuracy. It’s about a return rate that was 15% before and 13% after.

If your last AI vendor evaluation didn’t consider these criteria, the full playbook is where to start.

Worth your time

Worth your time

We know time is money, so we won’t waste yours

  • 63% of finance teams have AI fully deployed. Only 21% of them can point to clear, measurable value from it. (Deloitte)
  • AI shopping agents now drive real traffic volume, and fraud is moving in alongside it. The case for treating AI-referred orders as a decision signal. (Loss Prevention Media)
  • Returns fraud and returns abuse call for different responses. New research separates the two and maps the tactics spreading online. (ECR Retail Loss)

What we're up to

What we're up to

We’re gearing up for our last few IRL events of the year and next month, we’ll be hitting the boardwalk in sunny Huntington Beach for the Retail Club AI Festival. What could be better than sand, spritzes, and getting serious about AI trends in retail? If you’ll be in attendance, keep an eye out for our team or join us for a beach happy hour to unwind. And if this edition of The Takeback wasn’t enough to convince you how important it is to approach AI vendor evaluations with care, we invite you to swing by our booth to see what modeling out the ROI of a tool before you buy can look like in practice.

Next up, Dallas roadshow. We’re bringing together leaders across LP, ops, ecomm, and finance to discuss the latest in total retail loss, what LP as a company-wide strategy really means, and how you can take the first steps toward uncovering blind spots in your business. Spots are limited – grab yours while you can.

The Archive

Here’s what you missed.

Stay informed

Monthly insights on the loss you can actually control.