The Invisible Market: What Real Estate Offer Data Can Reveal

A closing price is the last line of a longer story

Most of what the public sees about a housing market is recorded at the end of a transaction: the list price, the sale price, how long the home took to go under contract. Those numbers are valuable. They are also incomplete.

Behind every recorded sale is a negotiation that the record does not describe. How many buyers made offers? How far apart were they? Which contingencies did each include, and did the winning buyer receive a concession that a recorded price does not reveal? In public data, the offers that lost leave no trace at all.

This article looks first at what public data does show for the Washington, DC metropolitan area, then at what it cannot. The figures describe the metro area as a whole — a region spanning the District and parts of Maryland and Virginia — not the District alone, and they are not broken out by jurisdiction.

What public data shows: 2022 compared with 2025

Zillow publishes monthly metro-level measures that practitioners commonly read as signals of competition. Averaging each year's monthly values gives a simple comparison between 2022 and 2025.

The share of sales closing above list price averaged 47.38% in 2022 and 36.05% in 2025, a difference of 11.33 percentage points. Over the same years, the share of listings with a price cut rose from 19.98% to 24.05%. The median sale-to-list ratio averaged 100.70% in 2022 and 100.00% in 2025 — the average of monthly medians moved from slightly above list to exactly at list.

Timing and supply moved in the same direction. Median days to pending averaged 11.00 in 2022 and 15.83 in 2025, while average monthly for-sale inventory rose from 14,124 to 15,456 listings.

Taken together, these are the patterns practitioners commonly read as less intense competition than in 2022. They do not say why. Interest rates, household finances, seasonality, the mix of homes listed and many other factors could each play a part, and this analysis does not separate them.

The same months, one year apart

Zillow data runs through August 2026, so a fair comparison with the most recent year has to use the same months. Redfin data in this analysis ends in December 2025, so this comparison uses Zillow only. Comparing January–August 2026 with January–August 2025:

• The share of sales above list was 36.8%, compared with 39.9%.
• Median days to pending was 14.5, compared with 11.4.
• Average for-sale inventory was 16,825 listings, compared with 15,200.
• The share of listings with a price cut was 22.8%, compared with 22.7% — essentially unchanged.

Three of the four measures continued in the direction seen between 2022 and 2025. Price cuts did not: the share was flat year over year. That is a useful reminder that market signals do not always move together.

A second source, a different yardstick

Redfin publishes related measures for the same metro area. They are useful as a check on direction — but not as a check on levels, because the definitions differ. Redfin's above-list share is measured against the original list price; Zillow's is reported as sold above list price, and this analysis does not establish whether Zillow's basis is the original or the most recent list price.

Redfin's share of homes sold above their original list price fell from 42.54% in 2022 to 31.15% in 2025. Its average sale-to-list ratio moved from 100.80% to 99.72%, and its median days on market rose from 27.92 to 39.58.

Each Redfin measure moved in the same direction as its Zillow counterpart. The levels, however, are different — sometimes substantially. Days on market and days to pending are defined over different intervals, so the gap between them should not be read as one source contradicting the other. Reading two sources side by side is a good discipline precisely because it forces attention to definitions.

What the record leaves out: a fictional example

Even with two sources, every measure above is built from listings and closed sales. None records the offers themselves. A simple, invented example shows why that matters.

A home is listed at $600,000 and receives three offers:

1. 01Offer A: $610,000, conventional financing, with a financing contingency and an inspection contingency. Not accepted.
2. 02Offer B: $600,000, cash, so no financing contingency, and no inspection contingency. Not accepted.
3. 03Offer C: $605,000, conventional financing, with a financing contingency, no inspection contingency and a $5,000 seller concession. Accepted.

In this example, a sale-level record would show one sale at the gross contract price of $605,000, a sale-to-list ratio of about 100.83%, counted among homes that sold above list. The concession is not subtracted from that recorded price. The record would not show that two other offers existed, that the highest offer did not win, or that after the $5,000 concession Offer C's illustrative net — $600,000, before commissions, fees and other costs — equals Offer B's cash price.

The example cannot say why the seller chose Offer C. That is the point: the reasons live in the terms, the conversations and the seller's own circumstances, none of which reach public summaries.

What offer-level data could add — and what it could not

Structured records of offers, captured consistently, could help answer questions that sale-level data cannot:

• How many offers did homes typically receive, and how did that change over time?
• How often did the highest price lose to different terms?
• How common were concessions, and how large were they relative to price?
• How did financing type and contingencies relate to which offers were accepted?

Such data would also bring its own limits. Offers are commercially sensitive, and any collection has to respect the confidentiality expectations of buyers and sellers and the obligations of the professionals involved. Records gathered from participating agents or brokerages may not represent the wider market. And even rich offer data would describe associations, not prove what caused a particular outcome.

Where OfferIntel fits

The challenge of observing offer activity helped shape my work on OfferIntel, a live market-intelligence and offer-management platform. Its OfferHub workflow collects and compares offers and carries an accepted offer through to closing, which is the kind of structured record that public summaries lack.

To be clear about the boundaries of this article: no OfferIntel data was used. Every market figure comes from public Zillow and Redfin downloads, and the offer example is fictional. The connection is the question, not the data.

What this means

For practitioners, the public measures are a reasonable starting point for describing a market's temperature, provided their definitions are read carefully. They are not a substitute for knowing what happened in an individual negotiation.

For anyone working with housing data, the lesson is about the unit of observation. A dataset built from closed sales can describe outcomes well and still say nothing about the competition that produced them. Recognizing what a dataset cannot see is as important as analyzing what it does.