Every marketplace will tell you what an item costs. None of them will tell you whether you can fit it in your car.
That gap is the single most common failure in a local sale: a buyer drives across town for a dresser, discovers it does not fit through their own front door, and both people have wasted an afternoon. We built ClearList partly to close that gap, which means we now measure things nobody else publishes.
So we looked at what we have.
Methodology
This analysis covers 300 listings created across 46 separate sales on ClearList, queried on 28 July 2026. Every listing is real. These are items people actually put up for sale, and many of them sold.
Two caveats you should weigh before quoting anything here.
First, the sample is small. Three hundred listings is enough to see shape, not enough to make confident claims about the whole world of moving sales. Treat everything below as a first measurement, not a settled fact.
Second, the distribution is lopsided. The two largest sales account for 179 of the 300 listings, about 60% of the corpus. Both are genuine sales with items that genuinely sold, so excluding them would throw away real data. But one household's contents can visibly move a category distribution.
Rather than pick one number and hope, we report both: the figure across all 300 listings, and the figure with the two largest sales removed (121 listings, 44 sales). Where those two numbers differ, the truth is somewhere between them, and the gap itself tells you how much to trust the estimate.
The query that produced these numbers is published in full at Cuneiform-LLC/clearlist-research, along with the findings and their caveats. It is a single read-only aggregation over our items collection: it counts categories, conditions, price statistics, dimension-confidence levels, and the truck and helper flags, and it emits only totals. No seller identifiers, email addresses, or listing titles appear in any output.
The headline: moving is a logistics problem wearing a pricing costume
| Measure | All 300 listings | Excluding the 2 largest sales |
|---|---|---|
| Requires a truck | 19.3% | 32.2% |
| Requires two people to lift | 14.0% | 24.8% |
| Median price | $35 | $105 |
Somewhere between one in five and one in three items cannot go home in a normal car. Between one in seven and one in four cannot be lifted by one person.
An item is flagged as needing a truck when any dimension exceeds 48 inches or estimated weight exceeds 50 pounds, and as needing help above 75 pounds.
Sit with that for a second. A buyer browsing a typical listing page has no way to know which items those are. They find out in the driveway. The information exists, it is knowable from a photograph, and the entire industry has simply decided not to surface it.
What people actually sell
| Category | All 300 | Excluding the 2 largest sales |
|---|---|---|
| Furniture | 19.7% | 30.6% |
| Electronics | 20.0% | 21.5% |
| Other | 25.7% | 18.2% |
| Decor | 12.7% | — |
| Clothing | 4.7% | 8.3% |
| Kitchen | 7.3% | 5.8% |
This is the measure most affected by the skew, which is exactly what you would expect. Household inventories are idiosyncratic. One person downsizing a full home moves the numbers more than fifty people selling a couch each.
The stable finding across both views: furniture and electronics together are roughly half of everything sold, and they are the two categories where dimensions and weight matter most. The logistics problem and the volume are in the same place.
What AI can actually tell from a photograph
This is the part we did not expect to be the most interesting.
| Measure | All 300 | Excluding the 2 largest sales |
|---|---|---|
| Item received dimensions | 96.0% | 96%+ |
| Exact manufacturer model identified | 48.3% | 54.5% |
| Estimated from visual cues only | 49.7% | 41.3% |
| Measured by a human | 0.7% | 1.7% |
Two things stand out.
About half the time, the AI is not guessing. For 48% to 55% of items it identified the specific product, which means dimensions come from a manufacturer specification rather than an inference about proportions. That is a meaningfully different quality of answer, and it is why we show a confidence level on every dimension instead of presenting all numbers as equally solid.
Almost nobody measures anything. Under 2% of listings had human-measured dimensions. This is the finding we would most like other people to check, because it reframes a common assumption: the choice is not between AI estimates and accurate measurements. It is between AI estimates and no information at all. Ninety-eight percent of the time, nobody was ever going to get out a tape measure.
Weight follows the same pattern. 93.7% of items carried an estimate, with a median of 5 pounds, which tells you how many small things are in a house relative to how few large ones.
Pricing reality
Across 291 priced listings:
- Median: $35. Excluding the two largest sales: $105
- Mean: $189.42, dragged upward by a small number of high-value items
- 56.4% of items are priced under $50
- 10.3% are $500 or more
- 3% are listed free
- Range: $5 to $15,000
The distance between the $35 median and the $189 mean is the whole story of a moving sale. Most of what you own is worth very little individually. A handful of things carry the value. The tedium is that both kinds require the same listing effort, which is the specific problem automatic listing generation exists to solve.
Condition, and a note on optimism
Sellers and the AI together graded items as: Good 69.7%, Like New 23.3%, Fair 4.7%, Used 2.3%.
Ninety-three percent of everything is rated Good or better. We do not think 93% of used household goods are genuinely in good condition, so read this as a measurement of grading behavior rather than of physical reality. Optimism is the default setting, for humans and models alike. It is worth knowing that your "Good" and someone else's "Good" are not the same claim.
What we could not measure
Being honest about the holes matters more than the findings.
- No sell-through timing. We know 16% of listings are marked taken and 64% are still available, but not how long items sat before selling.
- No price accuracy. We cannot yet tell you how often an AI-suggested price was edited by the seller, or how final sale prices compared to suggestions. This is the number we most want and will publish when we have it.
- Sale size is not usefully summarized. The median seller listed one item; the largest sale had 96. There is no meaningful "typical" sale size at this sample size.
- Zero prohibited items were flagged across all 300 listings. Either the filter is well-calibrated for household goods or it has never been genuinely tested. We cannot distinguish those from this data.
Why publish this
Partly because it is useful. Mostly because almost nobody who runs a marketplace publishes what they see, and the resulting advice ecosystem runs on intuition and anecdote.
If you are moving: expect roughly a third of your furniture to need a truck, expect most of your things to be worth under $50, and expect that nobody will measure anything unless a machine does it.
If you are building in this space: the dimension data is the underrated asset. Price is what everyone competes on. Whether the thing fits in a Civic is what actually determines if the sale happens.
We will re-run this analysis quarterly as the dataset grows. The numbers above are a snapshot of 28 July 2026, and we expect several of them to move.
If any of this looks wrong, open an issue. We would rather be corrected than cited incorrectly.