The short version
Carfax prices the car's paper trail and KBB prices its condition on average; neither prices this car, in your town, this month. Use them as bookends: KBB's private-party range is the honest middle for a typical example, Carfax's history-based figure says whether this VIN's past should pull it up or down, and your own comp set — same trim, same mileage band, your region, private sellers only, listing age noted — is what sellers are actually accepting. When a seller quotes a book value, ask which one and for what kind of sale, then put your comps on the table. The comps win, because they are the only number in the room a buyer down the road already paid.
Type a VIN into Carfax and it hands you a value. Answer a few questions on Kelley Blue Book and it hands you another one, usually different. Then you open Facebook Marketplace and the same car is listed at a third number that matches neither. Buyers treat this as a puzzle with one right answer. It isn't. The three numbers are answers to three different questions, and the mistake is quoting one of them as if it settled the other two.
This guide takes the two tools apart — what each value is built from, what each can and cannot see — and then shows how to build the only number that actually decides a private sale: what the same car is selling for near you, this month. That comp set is twenty minutes of work, and it is the number you should walk into the driveway holding.
What Carfax's value actually is
Carfax calls its figure a history-based value, and the name is the whole idea. Instead of starting from a generic year, make, model, and trim, it starts from a specific VIN and adjusts a baseline for the events Carfax has on record for that car: reported accidents, the number of owners, service records, how the car was used, and the mileage in its history. Two identical cars with different pasts get different values. That is the feature, and it is a real improvement on a book value that has never heard of the car in front of you.
It is also the limit. The value can only price history that reached Carfax — a collision fixed for cash and never reported to an insurer or a reporting body shop is invisible, and so is every mile of hard use that left no paper trail. An unreported accident makes the history-based value too high, not too low. Depending on how you ask, Carfax shows a retail figure — roughly what a dealer would list the car for — alongside a private-sale or trade-in figure, and the retail one is the number sellers love to quote. A dealer's asking price is not what you should pay a private seller in a driveway.
The same values appear on Carfax's own used-car listings, where they exist to make a dealer's price look fair. That doesn't make the number wrong. It means it was designed to be read next to a dealer's ask, and you should read it that way.
What KBB's value actually is
Kelley Blue Book goes at the problem from the opposite side. It does not know your car's history; it knows the model, options, mileage, region, and the condition grade you choose from a short questionnaire. It then produces a value from aggregate market and transaction data, which is why the same car gets one figure in one metro area and another somewhere else. At the time of writing the site separates what it shows by the kind of sale you are pricing:
- A trade-in figure — what a dealer might offer to take the car off your hands, and the lowest of the set
- A private-party figure — what one individual might pay another, usually shown as a range rather than a single number
- A dealer retail or typical listing figure — what a dealer might ask, with reconditioning, overhead, and margin baked in
- A condition grade you pick yourself, from Fair up to Excellent, and the grade moves the value more than most people expect
Why both numbers drift from what cars actually sell for
Neither tool is lying. Each answers a question you did not quite ask. Carfax answers “what is a car with this reported history worth in general,” and KBB answers “what is a car in this condition worth in this region on average.” What a private seller will accept for this car, this month, on your side of town, is a third thing, and it drifts from both for reasons that are easy to name:
- Condition is self-reported. A seller grading their own car on KBB picks Excellent; by KBB's own definitions that is a rare car, and Good is where most used cars live. The gap between the two grades is the negotiation, before anyone has looked at a tire
- History is incomplete. Carfax adjusts for what was reported; a private seller knows what wasn't. The unreported fender-bender and the neighbor's cash repair never reach the number
- Local supply and demand swamp the model. A pickup in a rural county and the same pickup in a dense city do not sell at the same price, and neither tool tracks what is sitting unsold at your local Craigslist and Marketplace this week
- Lag. Both tools lean on data that is already in the past; a market that has turned in the last couple of months shows up in live listings before it shows up in a book value
- Dealer figures leak into private deals. The retail or typical-listing number is the one sellers screenshot, and it carries a dealer's reconditioning and margin that a private seller never spent
- The tools cannot see the car. Worn tires, a due timing belt, a mismatched repaint, a missing second key — the things that actually subtract dollars are found in a driveway, not in a database
Build your own comp set
A comp set is a short list of the same car for sale near you right now, tightly filtered, with the asking prices and how long each has been listed. It is what an appraiser builds and what a dealer's pricing tool does under the hood. Open two or three marketplaces — Facebook Marketplace, Craigslist, Autotrader, CarGurus — and narrow hard:
- Same model year, or one either side if the generation did not change in between
- Same trim, engine, and drivetrain — the loaded all-wheel-drive version and the base front-driver are different products with different prices
- A mileage band around this car's, not the whole range for the model
- Your region, at the distance you would actually drive to buy — prices shift with local demand, weather, and rust
- Private sellers separated from dealers, because dealer asks include reconditioning and margin
- Listing age wherever the site shows it — a car sitting for weeks at one price is a price the market has already rejected
- Sold or marked-sold listings when you can find them; one closed sale outweighs five live asks
Reading the three numbers together
You will end up with a cluster of asking prices and a couple of outliers. The cluster is your anchor, and remember that it is made of asks, not sales — the real transaction prices sit somewhat below it. Now put the two book values next to it. If KBB's private-party range and the comp cluster overlap, the market and the model agree and you can price with confidence. If Carfax's history-based figure sits well above the cluster, ask what the cluster knows that Carfax does not: usually local oversupply, or a model that is simply not wanted where you live. If it sits well below, the car's reported history is dragging it, and that is your first talking point with a seller who is quoting KBB Excellent. For a quick second read on whether an ask is out of line, paste the listing into a free browser check — but build the comps yourself, because the point is to be able to name them out loud.
Then subtract what this car needs that the comps don't. The comps are average examples for the mileage, so ordinary wear is already in the price; what comes off is the specific, quotable work — tires, brakes, a service that is due, body damage — that you find by walking the car with an inspection checklist and pricing each item with a real quote. That turns a comp cluster into an opening number, a target, and a walk-away, which is the whole job of building an offer.
One more thing the book values hide: neither one is a history report. A Carfax value is derived from the report, but seeing the value does not mean you have read the record behind it, and a seller waving a KBB screenshot has told you nothing about accidents or title brands. Decide separately whether the full report is worth paying for on the car you are about to visit — the history it shows is what explains a Carfax value that looked oddly low.