How do you price eBay items from sold comps?

The single most expensive habit in reselling is finding the one sale that says your item is worth a lot and stopping there. I have done it. It feels like research and it is closer to shopping for a number you already picked.

A sold comp is evidence, not an instruction. It tells you what one person paid, for one item, in one condition, on one day. A price you can defend comes from a group of them, put on the same footing, read next to whatever a buyer could order instead this afternoon.

What follows is deliberately manual. It is slower than clicking a suggested price, and it produces a worksheet you can argue with later, which is the point.

Which eBay sold-data view should you use?

Product Research, if you have Seller Hub. It searches back three years, gives you calculated metrics, and it shows what a Best Offer actually sold for. Find it in the Research area of Seller Hub and in the eBay mobile apps.

The ordinary Sold Items filter is fine for a quick look and it hides something that will cost you money. When a listing sold through Best Offer, the number on the card can be the original Buy It Now price rather than the lower amount somebody actually paid. In a category where offers are normal, that is not a rounding error, it is a systematically inflated picture of your market. eBay describes Product Research as the view with the actual accepted-offer price. So: quick glance in Sold Items, real decisions in Product Research.

How do you build a comp query that matches the actual item?

Search with identifiers, not adjectives. Read the model number, part number, size tag, edition mark, capacity or pattern name off the thing in front of you, then filter by category and condition. "Sony WH-1000XM4 black used" is a query. "Sony wireless headphones nice condition" is a wish.

Then be strict about what you let into the set. Every one of these has burned somebody:

• Exact model, generation, storage capacity, size, or edition.
• Condition and function, including parts-only listings, repairs, stains, and missing pieces.
• Included accessories, packaging, certificates, chargers, cases, and manuals.
• Quantity. A lot of four is not four single-item comps.
• Sale format. Auctions can behave differently from fixed-price listings, especially with few bidders.
• Sale date and season. A coat sold in October may be weak evidence for July.
• Seller and shipping geography when those materially change delivery cost or buyer trust.

If that leaves you with almost nothing, widen one variable at a time and write down which one. Dropping color is usually harmless for a commodity cable. Dropping the model year on a camera body can make the whole set meaningless. Five real matches beat fifty things that look about right, and it is not close.

How do you normalize shipping, lots, and discounts?

Compare what the buyer paid to get the thing to their door, not the number in bold. This is the correction that changes the most prices in practice:

Delivered comp price = accepted item price + buyer-paid shipping

Do not add buyer sales tax to the comparison. It varies by buyer and jurisdiction. If a listing is a multi-item lot, use another line:

Delivered price per unit = delivered comp price / comparable units in the lot

Free shipping is not free. It is baked into the asking price, and pretending otherwise makes two identical sales look like different markets. A $35 sale with $8 shipping and a $43 sale with free shipping are both $43 delivered. Line them up next to each other and then start thinking about condition.

How do you use the median without hiding outliers?

Sort the delivered prices low to high and take the middle one, or the average of the middle two. Keep the low and the high where you can see them, because the spread is telling you how confident to be and the median alone will not.

And here is the rule I hold myself to: you may delete a result only when you can name what is wrong with it. A rare accessory bundle, an untested unit, a lot of four, local pickup, the wrong generation. All fine reasons. "It is higher than I want" is not, and neither is "it is lower than I hoped". A genuine high sale of the same item stays in the evidence even when you decide not to price there.

Here is an illustrative worked example, not a prediction of what any item will sell for:

After removing the documented accessory mismatch, the sorted delivered prices are $39, $40, $41, $45, $48, and $50. The median is ($41 + $45) / 2 = $43. If you plan to charge $8 shipping, a $35 item price creates the same $43 delivered ask. Fees, returns, and your cost of goods still determine whether that price works for your business.

How should active listings affect the price?

Asking prices are not sales, and people over-correct on this and ignore them entirely. They are worth exactly one thing: they show you what the next buyer can choose instead of you. Note how many close matches are live, their delivered asks, their condition and terms, and whether the same ones are still sitting there next time you look. That last part is the useful signal.

If the sold median is $43 but many equal-condition items are currently offered at $36 delivered, listing at $43 needs a defensible advantage. That might be better condition, complete accessories, faster handling, or clearer photos. If no advantage exists, the active supply is warning you that the historical median may be stale or that competition has increased. It does not tell you when any price will sell.

What should you do when exact sold comps do not exist?

Climb down a ladder, and label which rung you are standing on. The failure here is not using weak evidence, it is using weak evidence and then talking about it as though it were strong:

1. Extend the Product Research date range and note possible seasonality or market drift.
2. Remove one nonessential variant, such as color, while preserving model, condition, and quantity.
3. Compare the nearest model or edition and document the feature difference.
4. Check active substitutes and their delivered asks as a ceiling or competition signal, not a sold value.
5. Check completed evidence on another marketplace only if you can account for different buyers, fees, shipping, and return terms.

If it is still thin after all that, widen the range and protect yourself on the buy rather than on the sell. You can list with offers on, test a fixed price, or simply wait. An auction can find you a price, and with a thin audience it can also find you a bad one, so it is not the escape hatch people treat it as.

How do you turn the comp set into a listing price?

Start at the median delivered price and then write one sentence saying why your item sits below it, on it, or above it. Condition, what is included, the season, what else is live, your return policy, how long you can afford to hold it: all reasons. Hope is not a reason, and writing the sentence is what exposes that.

• Aggressive position: below the central comp evidence when recovering cash or reducing inventory matters more than seeking the top of the range. It does not set a sale date.
• Market position: near the median when the item closely matches the comp set and active competition has not undercut it.
• Maximum position: near the supported high end only when condition, completeness, scarcity, or seller terms justify it and you accept uncertain holding time.

In the worked example, approximate delivered asks might be $39 aggressive, $43 market, and $50 maximum. With $8 buyer-paid shipping, those translate to item prices of $31, $35, and $42. They are positions within observed evidence, not promised transaction prices or sale dates.

How does Vendlists approach price suggestions?

It reads the photos and whatever it can support, and it offers approximate aggressive, market and maximum positions rather than one confident number. For niche, vintage or collectible items it will do extra research first. For an ordinary commodity it usually does not need to. When it is unsure, it says so instead of picking.

What those positions are not is a completed-sales median. They are a starting point, and the work on this page is what turns a starting point into a price. Check the exact model, the condition, what is included, and the final number yourself. The walkthrough shows where drafting sits, and the plans and limits are on the pricing page.

What Vendlists reads from the photos, and what it will not infer from them, is on the photo to eBay listing page.

What should you record in a reusable pricing worksheet?

One row per candidate comp, one summary block per item. Boring, and it is what makes the next repricing a decision instead of a mood:

• Search query, filters, marketplace, and research date.
• Sale date, accepted item price when available, buyer-paid shipping, and delivered price.
• Quantity, condition, exact variant, included accessories, and sale format.
• Include or exclude decision, with a written reason for every exclusion.
• Low, median, high, and count of usable sales.
• Count and delivered-price range of close active listings.
• Your planned item price, shipping charge, evidence level, and next review date.

Save it against the SKU. When the item has not sold by your review date, rerun the same query and see what moved. That is a very different act from dropping the price because there are watchers and no offers, which is what most of us do instead, and it is the reason the sheet is worth the ten minutes.