The Price Comp Rabbit Hole: Why Comparing to 'Sold' Listings Gets You Wrong (And What Data to Use Instead)

Comparing your item to whatever popped up under eBay's "sold" filter, without adjusting for condition, season, and listing quality, routinely overstates the price you'll actually realize by 15-30%. The fix isn't more comps. It's normalizing the comps you already have against a fixed condition standard before you average them.

Most resellers know they're supposed to comp before pricing. Almost nobody does it right. You search the item, sort by "sold," eyeball the top five prices, average them, and list at that number or 10% above to leave offer room. That process feels rigorous because it involves real transaction data. It isn't rigorous. It's a sample of five to fifteen items that differ from yours in condition, season, photo quality, title SEO, and time listed — and you're treating the average like it's a single clean number.

Why sold listings lie to you

eBay's sold filter shows you outcomes, not comparables. Every listing in that grid sold under different conditions than the one you're about to list. The bias runs in a few consistent directions:

• Survivorship skew. You only see what sold. Items that sat for 90 days and got pulled or relisted at a lower price never show up in the sold filter — so your sample is pre-filtered toward faster, often better-priced, better-photographed listings.
• Condition blending. A search for "Levi's 501 32x32" pulls in everything from NWT deadstock to Fair-grade pairs with a repaired hem, all averaged into one number. eBay's condition field is self-reported and inconsistently used — sellers call the same defect profile "Excellent" and "Good" depending on how generous they feel that day.
• Season contamination. A wool coat that sold for $85 in November tells you almost nothing about what it will sell for in July. Sold comps carry a timestamp, and most resellers ignore it.
• Listing quality variance. Nine photos with a label shot and measurements versus three blurry flatlays produce different sold prices for identical garments. You're not comping the item — you're partly comping the seller's photography.
• Recency drift. eBay's default sold window can pull in comps from 90 days back. Trend-driven categories (Y2K, specific sneaker colorways, logo hoodies) move fast enough that a three-month-old sold price is stale.

None of this means sold data is useless. It means raw averaging without adjustment produces a number that looks precise and is actually just noisy.

The five variables every comp needs adjusted

Before you average anything, filter or adjust each comp for these five variables. Skip even one and your comp set drifts toward whichever variable you ignored.

Adjusting comps for condition — the part everyone skips

This is where the biggest pricing errors live. If you're using GradeThread's condition vocabulary — NWT, NWOT, Excellent, Very Good, Good, Fair, Poor — you already have a consistent framework. The problem is that the sold listings you're comping against almost never use that same standard. A seller's "Excellent" might mean no visible flaws under any of the five factors — Fabric Condition, Structural Integrity, Cosmetic Appearance, Functional Elements, Odor & Cleanliness — or it might mean "pretty good, I didn't look that closely."

Practical fix: don't take the seller's condition label at face value. Open the actual sold listing photos and grade it yourself against the same five-factor framework you'd apply to your own item. If a "Very Good" sold listing shows pilling under the arms and a small odor note in the description, that's closer to Good on Fabric Condition and Odor & Cleanliness — treat its sold price as a Good-tier comp, not a Very Good one, regardless of what the title says.

Once you've re-graded your comp set honestly, a simple discount ladder from your top-condition comp gets you close:

• NWT / NWOT: 100% of top comp (baseline)
• Excellent: 80-90% of NWT comp
• Very Good: 65-75% of NWT comp
• Good: 45-60% of NWT comp
• Fair: 25-40% of NWT comp
• Poor: liquidation pricing, rarely worth individual listing

These are starting ranges, not fixed rules — category and brand shift them. But they give you a defensible number instead of an average pulled from a condition-blended pile.

The comp methodology that actually works

Here's the step-by-step process to run before you price any item, in order:

1. Search the exact brand, model, and size on eBay filtered to "sold" and "completed," using the widest date range your tool allows (Terapeak or FlipDesk's comp module both work).
2. Pull every result into a list — don't stop at the first page. Aim for at least 10-15 raw comps before filtering.
3. Open each listing's photos and re-grade the item yourself against the five factors: Fabric Condition, Structural Integrity, Cosmetic Appearance, Functional Elements, Odor & Cleanliness. Assign your own tier (NWT/NWOT/Excellent/Very Good/Good/Fair/Poor), ignoring the seller's label.
4. Bucket the comps by your re-graded tier. Discard any bucket with fewer than 3 comps — too small to trust.
5. Within the tier that matches your item, discard comps older than 60 days for trend-sensitive categories (streetwear, fast fashion) or older than one full season for seasonal categories (coats, boots, knitwear).
6. Note listing quality on the remaining comps. If a comp has 3 blurry photos and no measurements and yours has 10 photos with a label shot, treat that comp's price as a floor, not a target.
7. Average the surviving comps in your matched tier. That average is your price anchor — list 5-10% above it if your listing quality and item specifics are stronger than the comp set, at or slightly below it if your inventory needs to move faster than average.

[Screenshot placeholder: FlipDesk comp module showing a filtered comp set grouped by re-graded condition tier, with average price and days-to-sell displayed per bucket]

A worked example

Say you're pricing a men's Patagonia Better Sweater, size L, in Very Good condition (light pilling on the sleeves, no odor, all zippers functional). A raw eBay sold search returns 14 results averaging $58. That number is the naive comp — and it's wrong, because the sample blends four NWT sales at $85+, three heavily worn Fair-tier sales at $28, and seven items closer to your actual condition.

After re-grading each comp by hand:

• NWT/NWOT (4 comps): average $88
• Excellent (3 comps): average $71
• Very Good (5 comps): average $54
• Fair/Good mixed (2 comps): average $29 — discarded, sample too small

Your real anchor is $54, not $58 — a small gap here, but on higher-ticket categories (denim, outerwear, boots) the naive-versus-adjusted gap regularly runs $15-40 per item. Multiply that across 200 monthly listings and the naive method is quietly costing you either lost margin (underpriced-then-discounted) or dead inventory (overpriced-then-stale).

Building this into a repeatable system

Doing this by hand on every SKU doesn't scale past a handful of listings a week. The methodology matters more than the manual labor — the goal is to bake condition-tier bucketing and season-adjustment into whatever comp process you run, whether that's a spreadsheet macro or a dedicated tool.

Three habits keep this sustainable at volume:

FlipDesk's comp module pulls sold listings and lets you tag each one by re-graded condition tier before averaging, so the anchor price you draft from is already normalized — not a blended number you have to mentally discount on the fly. Pull comps for one item you're about to list and run it through the condition-bucket method above before you draft the price. You'll likely find your naive average was off by more than you expected, in one direction or the other.