Retailers Pocket a Third of Supplier Profits – Called Deductions

01 /The Concept

Glimpse built an AI platform that helps suppliers claw money back from the deductions trap – making it dramatically faster, cheaper, and more accurate to audit deductions and execute the dispute procedures needed to recover unjustified chargebacks.

With the platform, suppliers can verify at least 95% of incoming payments and cut the recovery cycle by 15–20 days.

The Glimpse platform has three core modules.

Module one's AI agents analyze purchase orders, invoices, payment records, and retailer deduction notices – ingesting data from documents both sent and received, as well as from retailers' B2B portals.

Module two's agents cross-reference that data to surface discrepancies and flag potentially unjustified deductions.

Module three's agents handle the actual dispute workflow – filing claims, responding to retailer pushback, attaching supporting documents, raising credit memos, and tracking whether recoveries actually land in the supplier's account.

Glimpse launched in 2024 and already serves more than 200 brands shipping into retail. As a16z put it, "the results have been phenomenal" – suppliers have already recovered millions in disputed deductions, and the startup's revenue is "growing practically vertically" Pricing appears to combine a per-document processing fee with a percentage of recoveries obtained.

Glimpse raised its initial $10 million in early 2025, and has now closed a new $35 million round led by a16z.

One notable backstory: Glimpse went through Y Combinator as far back as 2020, but only launched this specific deductions platform in 2024 – spending the intervening years exploring various ideas in the consumer packaged goods space, attracting roughly $8 million along the way. The constant thread was simplifying operations for brands selling boxed products into retail. The deductions angle turned out to be the one that actually took off.

02 /Why It Matters

Start with an obscure problem most people have never heard of – because without understanding it, the startup's value proposition makes no sense.

When a brand ships large volumes of product to retail chains, something called "deductions" enters the picture. Retailers pay suppliers less than the agreed invoice amount – offsetting the difference against chargebacks for damaged or spoiled goods, shortfalls, vendor compliance violations, or promotional discounts that were supposedly negotiated upfront.

The problem is that retailers calculate these deductions unilaterally. The supplier either accepts the numbers on faith or assigns staff to audit every line item and fight back.

It might seem like a niche edge case – something that happens occasionally and can be safely ignored. Absolutely not.

Deductions happen so frequently and at such scale that they're often called the "hidden tax" on retail supply – typically running 2–10% of invoice value. Given the thin margins baked into wholesale pricing, those deductions can consume a third or more of a supplier's profit.

Major chains like Walmart, Amazon, and Kroger routinely apply deductions of 5–6% of shipment value. At high volumes, auditing every deduction is enormously time-consuming – so most suppliers simply let it go and absorb the loss, even when those deductions aren't fully justified.

Small and mid-sized suppliers feel the pain most acutely. A $35,000 deduction on a $500,000 shipment is genuinely material – but those suppliers rarely have the bandwidth to dispute it.

Manual audits and chargeback battles typically recover no more than 50% of disputed amounts, and the resolution process takes at least three months. During that window, suppliers either can't reconcile their books or have to restate financials and file amended reports with the IRS – compounding the hassle considerably.

The net result: a mid-sized brand doing $50 million in retail annually can easily have $2–5 million per year stuck in the deductions trap – money it must either write off or fight hard to recover.

03 /Opportunities

The deeper lesson here is that the critical variable isn't the idea – it's the audience. Pick a customer segment you can genuinely access and serve, then keep digging until you find what actually matters to them.

Don't hunt for a brilliant idea – choose a target audience. You can approach that audience with almost any initial concept, but don't fall in love with it. Keep looking for what's truly important to those people, because the most valuable problems are rarely visible from the outside. You need to get in the room first before you can find them.

Customer interviews alone won't get you there. You don't yet know what to ask – because you're not inside yet. The better approach is to enter a space with a concrete product that gets you real relationships in the industry, builds genuine understanding of how things actually work, and eventually reveals the non-obvious angle worth going after.

A closely related example illustrates the same pattern. A founder builds software for US residential moving companies – 7,000 in the country, 900 already customers. The product handles scheduling, routing, and customer management. At first glance those looked like the core value. Wrong.

The feature customers actually buy the product *for* is claims management. Moving companies face at least one damage claim for every four moves, averaging $800 per claim. Most carriers just pay it to avoid negative reviews. The software now includes a mobile app where movers photograph every item before and after the move, AI generates a condition report, and customers sign off at both ends. Carriers who ran the old way paid an average of $180,000 per year in claims. Customers on the platform now pay $60,000 – covering only legitimate damages. That's $10,000 per month recovered against a $525/month subscription.

When the founder led with the claims module instead of the standard demo, the sales cycle shrank from 45 days to 8.

This story maps almost perfectly onto Glimpse. So – what's the non-obvious feature that could anchor the product for your target audience?

If no answer comes to mind, it probably means you haven't gone deep enough yet.