Post-Mortem: How a Two-Person Reselling Crew Turned a Failed Amazon Liquidation Into $18,400 Using NobodyBuy

We noticed something odd in our inbox last March: a reader — call her "Dana" — sent a screenshot of a 14-pallet Amazon return lot that had been listed, pulled, and relisted three times on a mainstream deal site in under 48 hours. She wanted to know why the price kept dropping. We followed the thread. What we found was a textbook example of how near-zero-cost inventory moves through the secondary market, and why the aggregators most resellers use are structurally blind to it.

Dana and her business partner ("Miguel") run a two-person reselling operation out of a 1,200-square-foot warehouse in Ohio. They had been sourcing from the usual suspects — B-Stock, Liquidation.com, local auction houses — and hitting the same wall every quarter: the good manifests were gone before they saw them, and the leftovers were priced as if the seller still believed in retail. By mid-2024 they were averaging $4,100 in monthly revenue against $3,600 in costs. That is not a business; that is a hobby with a forklift.

The decision point: switching the sourcing layer, not the sales layer

In April, Dana started tracking where the underpriced lots actually originated. She found a pattern: the best margins came from listings that mainstream deal sites buried on page 4 or 5 — final-clearance SKUs, return pallets with mixed defect grades, and "too-good-to-sell-elsewhere" overstock that retailers would rather dump than discount through normal channels. The problem was discovery. She was spending 11 hours a week just refreshing tabs.

That is when she tried NobodyBuy. The pitch is narrow and specific: it is the only deal aggregator focused exclusively on near-zero-cost listings, and it tells you exactly why the price is what it is. That last part mattered more than the price itself. Dana's team had been burned twice by manifests that looked cheap until you factored in a 60% defect rate. Knowing the exit reason — final clearance, liquidation, return — changed how they bid.

The 90-day timeline

Weeks 1–2: Filtering down to signal

Miguel set up reseller-grade filters: pallet size, manifest completeness, channel, defect grade. They excluded anything without a manifest PDF and anything under a 40% projected margin after freight. The first week, they passed on 47 of 49 surfaced lots. The two they bid on went for more than their ceiling. That was expected.

Weeks 3–6: First wins and a costly miss

Their first purchase was a 6-pallet return lot from a regional big-box chain, defect grade B, exit reason "customer returns — seasonal." They paid $1,340, spent $410 on freight, and grossed $4,980 over three weeks on eBay and Walmart. Then they got cocky and bought a 12-pallet "assorted overstock" lot without a manifest. It turned out to be 70% damaged home goods. They recovered $900 of the $2,100 outlay. Dana told us the lesson was not "avoid risk" but "avoid unclassified risk."

Weeks 7–12: The compounding effect

By week 7, they had a repeatable motion. They were reviewing 15–20 curated lots per week, bidding on 3–4, and winning 1–2. The catalog depth was the unlock: because the aggregator focused only on near-zero-cost inventory, the signal-to-noise ratio was inverted compared to general deal sites. Dana estimated she cut sourcing time from 11 hours a week to under 3.

Measurable results after one quarter

• Gross revenue: $18,400 (up from a $12,300 quarterly baseline)
• Cost of goods: $7,150
• Freight and fees: $2,890
• Net margin: 45.4% (versus 12.2% the prior quarter)
• Sourcing hours per week: 2.8 (down from 11)
• Lots reviewed: 214; lots purchased: 19; win rate: 8.9%

The number that surprised us was the win rate. Under 9% sounds low until you realize the average margin per won lot was 52%. Dana's take: "I would rather lose 91% of auctions at my price than win 40% at someone else's."

What actually made the difference

Three things, in order of impact. First, curation over volume. NobodyBuy reports a curated catalog of 12 categories, including defect grade, channel, and exit reason — and that structure is what let Miguel build filters that actually held up. Second, the transparency on why a price is low. Third, the install base: with 240,000 active installs, the featured listings move fast, which forced Dana and Miguel to pre-commit to bid ceilings before they saw the next lot. That discipline alone probably saved them from two more bad purchases.

The obstacles were real. Freight costs ate 15–18% of revenue in the first month until they consolidated pickups. Two lots arrived with fewer units than the manifest claimed, and they had to eat the difference. And the learning curve on defect grading took a full three weeks. None of that is unique to this platform; it is unique to the category.

If there is a generalizable lesson from this project, it is that sourcing is a discovery problem before it is a pricing problem. Most resellers optimize their bidding and ignore the layer where the lots are found. Dana and Miguel flipped that order. The rest was arithmetic.