Pricing art with comparable sales

Pricing art is part craft and part nerve, and the nerve comes from starting cold. A new work by an emerging artist, a medium you don't sell often, a size that doesn't map neatly to your last few sales — and suddenly you're staring at a blank field, knowing that whatever you type sets an anchor you'll have to live with.

The instinct is to reach for a round number that “feels right”. Sometimes that instinct is excellent. Often it's just a number with nothing behind it — too high and the work sits, too low and you've left money and the artist's standing on the table. There's a better starting point, and it's the one auctioneers and appraisers have always used: comparable sales.

Why a band beats a number

A single price pretends to a certainty that doesn't exist. A band — a low, a midpoint, a high — is honest about the truth of pricing: there's a reasonable range, and your job is to place this specific work within it.

A range built from comparables does three useful things at once. It tells you the floor below which you'd be underselling. It tells you the ceiling beyond which you're testing the market's patience. And it gives you a defensible midpoint to negotiate around — one you can explain to an artist or a collector with something more than “it felt right”. The band doesn't make the decision; it makes the decision informed.

How AI estimates the range

In Artfolio, the price band isn't a mysterious oracle — it's comparable-sales reasoning, run quickly. When you upload a work, the AI reads what it can see and considers it against relevant sales data to suggest a low–high range with a confidence indicator. What it weighs is roughly what you'd weigh yourself:

• The artist and the market — where comparable works have actually transacted.
• Medium and size — a large canvas and a small work on paper don't sit in the same range, and the estimate reflects that.
• The comparables themselves — surfaced so you can see what the band is built on, not just the number it produced.

Crucially, it shows its work. A range you can interrogate — here are the comparable sales, here's the size factor, here's the confidence — is a tool you can trust. A number with no reasoning is just someone else's guess wearing a confident face.

The estimate informs; you decide

Here is the line that keeps this honest: the AI estimates and you decide. The band is a starting point, never a final price, because the things that move a real price often aren't in the data.

Condition and provenance can lift or sink a work in ways a model can't fully see. The relationship matters — what you'd price for a first-time buyer differs from a patron of a decade. Timing matters — a fair, a show, a moment of momentum around an artist. And your read on where this particular artist's career is heading is judgement no estimate replaces. The band hands you a defensible range; you place the work within it as only a person who knows the context can.

From price to pipeline

Pricing isn't a one-off event — it's the front of a sales motion, and it's worth landing the work where the rest of that motion lives. Because the band is generated inside the same workspace as your pipeline, offers and invoices, the number you settle on flows straight into the deal. There's no re-keying it into a separate quote, no copy-paste into an invoice template, no risk of the agreed figure and the documented figure drifting apart.

Good pricing has never been about having the one true number — it's about starting from something real and then bringing your judgement to it. A band from comparable sales gives you the real starting point; your knowledge of the work, the artist, and the buyer turns it into the right price. Let the machine end the blank-field problem, and keep the decision where it belongs — with you.