Forget "how many SKUs" they have. That’s a vanity metric. I want to know their SKU velocity-to-bin-location ratio. In an apparel brand with 50 variations of a single t-shirt (size/color), if the high-velocity units are tucked in the back of the warehouse because "that's where they fit," your picking time will explode. You audit this by looking at their pick-path logs. If the picker is walking more than 10 meters for a Top-20% velocity item, the operation is fundamentally broken.

This is where CFOs get burned. A "ghost" inventory happens when the storefront says an item is in stock, but the physical bin is empty. This usually stems from poor integration between the ERP and the WMS. I’ve seen brands with a 12% "short-ship" rate because their sync cycle was lagging by hours during flash sales. You need to see the delta between promised inventory and actual available-to-promise (ATP) stock at 10:00 AM, 2:00 PM, and 8:00 PM on a high-traffic day. If that gap exists, your customer service team is just a band-aid on a systemic hemorrhage.

In the Indian market, Return to Origin (RTO) isn't just an operational headache; it’s a margin killer. You need to audit their pincode-specific RTO data. Is the brand trying to ship to high-risk rural zones with "premium" courier partners who just swap boxes in a warehouse for 48 hours? That’s wasted freight. A healthy scale model uses automated routing logic that switches carriers based on historical success rates per zip code. If they aren't filtering by pincode reliability, they are subsidizing the courier’s inefficiency with their own margins.

Specifically for FMCG or electronics, you need to audit the "dead space." I’ve seen brands ship small items in oversized boxes because they didn’t want to negotiate custom packaging layouts. They pay for volumetric weight on every single shipment. If your average box occupancy is below 70%, you are literally throwing money into a furnace. You check this by comparing the 'Declared Weight' vs. 'Actual Product Weight' across their top 100 SKUs.

Finally, look at the manual intervention point. Every automated system has a breaking point where human eyes must intervene—handling exceptions like "address not found," "partial shipments," or "weight discrepancies." If more than 15% of their daily orders require a manual "push" to get out the door, the automation isn't "advanced"—it’s just thin.

The Reality Check: A Case Study in Failure I once consulted for an apparel brand that grew from ₹20Cr to ₹150Cr in 18 months. They were "winning" on paper. However, during a major festival sale, their system collapsed because they hadn't accounted for multi-node inventory visibility. They had one central hub and three regional hubs. Because their API couldn't route orders based on the nearest available stock—only the total stock—trucks were crossing half of India to deliver a single shirt that was sitting 50km away from the customer in a different warehouse. They ended up with a 22% RTO rate and a massive spike in "out-of-stock" cancellations because the central hub got overwhelmed by orders it couldn't fulfill. They were moving boxes, but they weren't managing a network.

The Implementation Matrix: How to Audit the Tech Don't take their word for it. Ask for these three specific data exports during your discovery:

• Sync Logs : Show the timestamp of the last successful sync between the OMS and WMS. (Anything > 5 minutes means they are vulnerable to overselling).

• RTO Heatmap : A list of the top 50 RTO-heavy pincodes over the last 90 days. (If it's not a concentrated geographic cluster, their routing logic is failing).

• Picker Productivity Metrics : Average "Units Per Hour" (UPH) during peak vs. off-peak. (If UPH drops by more than 20% during peaks, their warehouse layout isn't built for scale).

Stop looking at their growth charts and start looking at their error logs. That’s where the real story is told.