Most liquidation marketplaces don’t sell pallets

This is the whole finding in one table. Live lots only, measured 21 July 2026. These are the six of the nine marketplaces that run auctions with an end date — the other three get their own section below and contribute no rows here.

“Single-item lots” counts live lots on that marketplace whose unit count was exactly 1. Every units-per-lot figure on this page describes these six marketplaces, never nine.

The gap is not subtle. B-Stock’s median live lot held 68 units. Musick’s held one. Those are not two versions of the same product — they are two different businesses that happen to share a search term.

Three of the six, and not marginally. On Musick, 3,927 of 4,310 live lots were a single item — 91% of everything it had live. On QuickBidz, 3,464 of 4,279, or 81%. On GovDeals, 54 of 64, or 84%.

Across all six auction marketplaces, 7,617 of the 11,132 live lots where a unit count was readable — about 68% — were a single item. Not a pallet, not a case pack. One unit.

Lots live on QuickBidz at the moment of measurement included a single Nordic Ware baking sheet, one Threshold chair pad, and one iPhone case, each at a $2.00 opening bid. That is a real, functioning marketplace. It is not a pallet. And $2.00 there is an opening bid, not a price anybody paid.

Anyone landing on one of those three from a “liquidation pallets” search and bidding without checking the unit count is bidding on one item. BStalker’s coverage pages for Musick and QuickBidz describe what each house tends to run — but those are written platform profiles, not this measurement, so don’t read them as data. What this measurement says about Musick is narrower and firmer: on 21 July 2026, 3,927 of its 4,310 live lots were a single unit.

888 Lots, DirectLiquidation and Via Trading carried no auction end date at all in BStalker’s data on 21 July 2026. No clock, no closing bid, nothing scheduled to end. They read as fixed-price storefronts, so the number on a listing is an asking price rather than a bid climbing toward a close. Whether a storefront revises its own asking price over time is not something this measurement covers.

That matters more than it sounds. Most of the standard beginner advice about liquidation — bid late, set a maximum, watch the close — is auction advice. On a third of the marketplaces in this dataset none of it applies, because there is nothing to snipe.

It also sets the boundary on everything above. Every units-per-lot and single-item figure on this page is computed over the 11,132 live lots where BStalker could read a unit count, and all of those sit on the six auction marketplaces. These three storefronts contribute nothing to that set, so nothing on this page tells you what a typical lot on them contains.

If you take one thing from this page: check the unit count before you look at the price. It tells you which of these three businesses you are standing in, and no amount of price analysis will tell you that.

Look again at B-Stock: median 68 units, average 519.9. The average is 7.6 times the median.

That is a right-skewed distribution doing the work — some lots are far larger than the middle one, and they pull the mean up and away from anything typical. Read “B-Stock lots average about 520 units” and picture a typical lot, and you are picturing something roughly eight times bigger than the one in the middle of the pile.

Liquidation.com has the same shape: median 27, average 239.1. Musick: median 1, average 2.9. BlueLots is the mild one at 96.9 against 50, so this is not a universal law even inside the table — which is exactly why the medians are printed next to the averages rather than instead of them.

Whenever a liquidation site or a blog quotes you an average lot size or an average discount, the honest follow-up question is: what is the median, and what is n? A mean without a median in this category is usually a truckload wearing a costume.

BStalker’s canonical metric is cost per unit — the bid, plus shipping where the platform quotes it, divided by the unit count. Across all live lots on 21 July 2026 that figure averaged $51.16.

We publish that number and then immediately tell you not to use it the obvious way, because the obvious way is wrong twice over.

First: it is a current bid, not a price. A live auction lot shows what someone has bid so far. It is not what anyone paid, not where the lot will close, and not what you would pay. Only closed auctions support “sold for” language — and even there, our statistics page states the caveat plainly: a median close is the highest bid we recorded before the auction closed, a strong signal but not a confirmed sale price.

Second, and worse: cost per unit is not comparable across marketplaces that sell different unit types. A $2 per-unit figure on QuickBidz is a single returned item with a $2 opening bid. A $200 per-unit figure on B-Stock may be one item inside a pallet of televisions. Putting those two numbers in the same column and concluding that one marketplace is a hundred times cheaper is a false comparison — and heading it off is the reason these unit counts were measured in the first place.

That has to point at our own pages too. The best liquidation sites page shows a median $/unit beside each marketplace wherever the live sample clears its 30-lot floor, and BStalker’s own copy describes cost per unit as making lots comparable across marketplaces. Measured against the unit-type split above, that phrasing is too strong, and this page is the caveat: read that column one marketplace at a time, not as a ranked list where the smallest number wins. Between rows, the thing being counted changes.

Cost per unit is a useful metric within a marketplace and within a category. Across the nine, it is close to meaningless.

Those eleven rows add to 3,932 of the 11,527 lots BStalker had live that day. The rest either fall in categories not listed here or carry no category at all — and across the full 24,400-lot catalog, 13,639 lots (56%) have no category. So this table shows the shape of the categorised slice, not the whole market. Note also that the 56% is measured against the full catalog while these counts are live-only, so the two are not meant to reconcile.

Don’t build a thesis on “there are only 114 jewelry lots in liquidation.” Build it on “BStalker had 114 categorised jewelry lots live on 21 July 2026, in a catalog that is 56% uncategorised.”

That last row is the uncomfortable one, so it is worth saying out loud: only about 8% of live lots had a manifest BStalker could parse at all. Manifest availability is the practical limit on how much anyone — us, you, or any tool — can know about a lot before buying it. One honest caveat on that 8%: it counts manifests our parser could read, which is not the same as manifests published, so read it as a floor on availability rather than a measured publication rate.

If you want the vocabulary for what a manifest actually promises, we keep a plain-English glossary and a guide to reading one. The comparison between a manifested and a blind lot is in manifested vs unmanifested pallets.

• This is not the liquidation industry. It is what BStalker tracked, on one day, across nine marketplaces. Other marketplaces exist, coverage depth differs per marketplace, and this page does not claim complete coverage of any of them.
• It is not all nine, either. The units-per-lot and single-item figures cover the six marketplaces that run timed auctions. Three of the nine contribute no unit counts to this measurement at all.
• No lot here is called a good deal.BStalker deliberately has no opinion about individual lots — no scoring, no ranking, no “we think this one is worth it.” That is a product decision written into its own non-goals, and it applies to this page too.
• Two rows are thin. BlueLots (n=83) and GovDeals (n=64) are the smallest samples in the table. A median drawn from 64 lots can move a lot in a week. Weight them accordingly.
• Medians move. This is one measurement, not a trend. This page gets re-run and re-dated rather than replaced with a second post claiming the same thing with different rounding.
• Single-item marketplaces aren’t worse. They’re different. Buying one unit and buying a pallet are different transactions carrying different risks, and this page takes no position on which one you should do. The failure mode isn’t using these marketplaces — it’s using them by accident.

This works whether or not you ever open BStalker.

1. Find the unit count first. Before price, before photos. If the listing doesn’t state one, that absence is itself the answer.
2. Find the end date. No end date means a storefront rather than an auction, and no clock forcing a decision.
3. Divide, then sanity-check the divisor. Price per unit only means something once you know what one unit is.
4. On a live auction, remember the number is a current bid. It only moves up. Whatever you compute now is a floor, not a cost.
5. Ask whether there is a manifest at all. Only 888 of the 11,527 live lots in this dataset — about 8% — had one BStalker could parse.
6. Add freight before you compare anything. A per-unit figure that doesn’t include landed freight is incomplete by construction — the shipping-cost guide covers how to estimate it before you bid.

Measured: 21 July 2026, from BStalker’s own production database. Every figure on this page is computed from the lots BStalker scans and the closed auctions it has archived. These are our own tracked numbers, not an industry estimate.

Population:“live lots” means lots BStalker had marked live at the moment of measurement — 11,527 across nine marketplaces, out of a 24,400-lot catalog. 24,302 of those catalog lots had been re-checked within the prior 24 hours.

The denominator, stated plainly: the units-per-lot and single-item figures are computed over the 11,132 live lots where a unit count was readable, every one of them on the six auction marketplaces. They are not computed over all 11,527 live lots and notacross all nine marketplaces. 888 Lots, DirectLiquidation and Via Trading carry no auction end dates and contribute no live-lot rows. One arithmetic note you can check against the table yourself: the n column sums to 11,133 live lots, one more than the 11,132 with a readable unit count. The live set also changes by the minute as auctions close, so treat these as “about 68%” and “roughly 11,100,” dated to 21 July 2026.

Median units per lot: computed per marketplace across that marketplace’s live lots only. n for each row is printed in the table (B-Stock 1,100 · BlueLots 83 · Liquidation.com 1,297 · Musick 4,310 · QuickBidz 4,279 · GovDeals 64). Median is the middle value; the average column is shown alongside precisely so you can see how far apart they are. Unit counts are whatever the listing published — we count that field, we do not audit it against what comes off the truck.

Prices:every price figure on this page is a current bid on a live auction, not a sale price and not a purchase price. The $51.16 average cost per unit is bid-based, computed across live lots. Closed-auction figures — where “closed for” language is defensible — live on the statistics page instead, under a floor of 30 priced closings per category.

Categories: category as recorded in BStalker’s data at the moment of measurement. 13,639 of the 24,400 catalog lots (56%) carry no category at all, so the category table is explicitly not exhaustive.

Known limits: coverage per marketplace is not complete and this page does not claim it is; manifest parsing succeeded on 888 live lots, which is a floor on availability rather than a publication rate; and a single day is a snapshot, not a trend.

• 2026-07-27 — First published on bstalker.com. Figures unchanged from the 21 July 2026 measurement; no number on this page has been re-run since.
• 2026-07-21 — Measured: 24,400 lots in the catalog, 11,527 live, nine marketplaces; units-per-lot figures over the 11,132 live lots with a readable unit count.

When the medians move, the table and this changelog move with them — this page gets re-measured and re-dated rather than replaced. If you find an error in any figure here, there is a contact route on the about page; corrections land here as a dated note rather than a quiet edit.

• Liquidation auction statistics — closing prices by category, discount against stated retail, and when auctions close, with the same methodology-in-public treatment.
• The best liquidation sites — the nine compared side by side, including the median $/unit column this page puts a caveat on.
• Live sample feed — no account needed, if you want to check any of this against live listings yourself.
• Auctions ending today — the auction-clock view, which by definition covers only the six marketplaces that run auctions.
• All buying guides — the process, cost anatomy and manifest-reading cluster this measurement sits under.