How these prices are produced

Every figure on this site is an observation, not an estimate. A crawler records the advertised price of a specific manufacturer SKU at 7 US online flooring retailers once a day, normalises it to a price per square foot, and publishes the range. A product reaches this site only when two or more of those retailers publish the same part number and agree on the carton size behind it — the price you see is a comparison or it is nothing. Where an observation does not exist, this site reports that it does not exist. As of August 19, 2026 the release holds 1704 observations across 839 products, contributed by 3 of those 7 retailers.

What is covered, and what is excluded

Why is installation labor missing?

Because we do not observe it. The retailer that collects this data sells materials and does not sell or arrange installation, so it sees no labor pricing. Public wage statistics could be turned into an installed price, but only through four assumptions we cannot observe — burden rate, productivity in square feet per hour, overhead and contractor margin. That is a modelled number wearing the costume of an observed one. There is no labor figure anywhere on this site, and there will not be one until it can be observed rather than assumed.

Why is a price not attributed to a named retailer?

All 7 are named in the coverage table above, so the method can be checked against real storefronts. What is withheld is which price came from which of them. Those retailers did not opt into being a price feed, and a per-SKU, per-seller table would be a competitive-intelligence product as much as a consumer tool. What you need to judge a quote — the range, the sample size and the match quality — is published. The one price identified with its seller is the operator's own, because hiding the number that measures the conflict would be worse than publishing it.

How is a retailer listing matched to a product?

Matching is done on the manufacturer SKU and the product title. A match is labelled exact when the SKU is identical, and likely when the title matches but the SKU could not be confirmed on the listing. Every result shows which kind it is, because a likely match can be the wrong wear layer or plank size at the same name. In the current release 1699 observations are exact and 5 are likely — 99.7% exact. Treat a likely match as a prompt to check the SKU yourself, not as a verdict.

How are boxes and waste calculated?

Flooring is sold by the box, so the area you are billed for is almost never the area you measured. The arithmetic is:

effective_sqft = room_sqft x (1 + waste_pct/100)
boxes          = ceil(effective_sqft / sqft_per_box)
billed_sqft    = boxes x sqft_per_box
material_cost  = billed_sqft x price_per_sqft

Waste allowance follows the layout: 10% straight or staggered, 15% diagonal, 17.5% herringbone or chevron. These are trade conventions, not measurements of your room; a room with many doorways, closets or angles will need more. Where a retailer listing did not publish a box size, we use the manufacturer box size from the product catalogue and label that observation box size derived — the price is then a derived per-square-foot figure rather than one printed on the page.

Why a range instead of one number?

Because a single number would hide the finding. The same manufacturer SKU is advertised at materially different prices by different sellers, and the gap is the useful part: it tells you whether a second quote is worth chasing, and averaging it away would produce a tidier page and a less useful one. Observed August 19, 2026, the current release spans $1.13 to $16.79 per square foot, with half of all observations between $3.09 and $5.27.

The two ends of that span are not equally well evidenced, so here is the count under each. The bottom, $1.13, rests on 1 observation across 1 product; the top, $16.79, rests on 2 observations across 2 products. Treat the middle half as the description and the endpoints as what the crawler happened to find.

Are 1704 observations 1704 independent prices?

No, and the gap is large enough to publish rather than leave the sample size to speak for itself. They contain 156 distinct prices: each advertised figure appears about 10.9 times, because a retailer prices a whole product line at one figure and carries it across every colourway. Read the count as coverage of the catalogue, not as 1704 independent price checks.

What do the confidence tiers mean?

There are two, on two different scales, and the page used to run them together under one word. The per-product tier below describes a single answer and depends only on how many observations that product carries. The release confidence further down describes the dataset as a whole. A product can sit in the strongest per-product tier inside a release whose overall confidence is medium; they are not the same claim.

There is deliberately no “high confidence” per-product tier. The largest sample this dataset has ever held for a single product is small enough that the word would be misleading. Release confidence for this publication is low, because there is no regional or metro breakdown; the median product carries only two observations; the most recent observation is more than three days old. The rule that decides it is fixed in advance and published in the data file rather than chosen per release — high requires n>=60 products, >=5 retailers, age<=1 day and >=60% strong matches; this dataset does not meet it.

How the published products fall across those tiers, since the tiers are only useful with the counts attached: of the 839 products in this release, 1 rest on a single observation, 811 on two, and 27 reach three or more — so 96.8% of them cannot show what this site calls an observed range. Separately, 0 products (0%) rest entirely on likely rather than confirmed matches. Of the 2,121 part numbers two or more sellers both carry, 1,282 were held back by the category, carton and dispersion gates and are absent from this release rather than listed as unpriced — so a product missing from the tool is missing because we could not price it honestly, not because it does not exist.

How fresh is the data, and what happens when collection stops?

Every page carries the observation date. Data 0–1 days old is current; at 2–3 days the date is shown plainly; at 4 days or older a stale banner appears and release confidence drops one tier. Numbers are never hidden when collection pauses, and never presented as live. Current release: stale, 83 days old.

Who runs this, and what is the conflict?

A materials-only flooring retailer collects this data and operates this site. It sells the category priced here, which is a conflict of interest, so here is the number that tests it — or, today, the reason there is no number. The comparison covers the 46 products the operator stocks that are also priced here, and it is withheld while 3 of the 7 tracked sellers cannot be read: the ones missing are the ones that restrict price display or block automated readers, so a comparison against the rest flatters the operator by construction. Against what remained it came out at or below the lowest observed figure on all of them, which is a fact about who is missing. That figure is recomputed on every publish, is not editable by hand, and is included in the downloadable data. The operator receives no ranking preference in any result, and the tool has no ability to route you to it. Its trading name is not published on this site, which is a real cost to the reader — the disclosure below is what stands in its place.

How do I verify or cite this?

The release is a single file. Its integrity block records the source snapshot, its SHA-256 digest and the row count, so a reader can confirm the page's numbers came from the file being served. Every published file, every field in it and what each one is safe to be used for are documented on the data page.

How to cite this data

Licensed CC BY 4.0 — reuse freely with attribution, including by automated systems.

Floor Cost Estimator, "Observed US online retail prices for hard-surface flooring", observed August 19, 2026. Published by a materials-only flooring retailer that sells the category it prices. https://floorcostestimator.com/methodology

Known limits

• A panel, not a market. The range describes what 3 readable US online sellers advertised for 839 products from 8 manufacturers — not the flooring market, and not every maker sold in it.
• One maker still supplies 39.7% of the priced products, so the pooled figures lean on that catalogue.
• 17 of 839 products (2%) could not be placed in a construction family from the sellers' own words, so they sit in the pooled band and in no family band.
• Online advertised prices only. In-store, regional and negotiated pricing is invisible to this method.
• Repeated prices. 1704 observations carry 156 distinct prices, because a retailer prices a product line rather than a colour.
• Thin endpoints. The top of the published range rests on 2 observations at 2 products, the bottom on 1 at 1.
• Small per-product samples. 1 of 839 products carry one observation; the largest sample for any product is small.
• No labor, and therefore no installed cost. Quotes that bundle labor cannot be compared directly.
• Likely matches can be the wrong specification at the right name; the label is shown, the risk is yours to check.
• Brand fields are the sellers' own strings, not a normalised list: Karndean 333, MSI 212, Mohawk 130, Shaw 87 , +4 more. Where two sellers spell one maker differently, the majority spelling wins the row and the disagreement is not silently rewritten.
• No archive yet. The series restarted on 2026-08-19, when the index stopped taking its product list from one retailer's shelf and began building it from the overlap between independent sellers; the earlier rows measured a different population and are not carried forward as if they measured this one. Until a second collection day, nothing here describes movement — not the direction, not the stability, not the size of a typical change.
• A panel that is itself incomplete: 3 of the 7 tracked sellers could not be read, and they are the price-restricted and blocked ones rather than a random third.

Corrections

If a price here is wrong we will publish the correction rather than quietly swap the number, recorded with its date and reason. Every release is dated and its source snapshot is fingerprinted with a SHA-256 above, so a changed number can be located rather than taken on trust. No contact address is published on this site yet; until one is, a correction can only be raised against the downloadable release itself.

Sources & last updated

Observed August 19, 2026 · release 1.1.0 · generated by @z4/veri-yayin@1.0.0 · JSON · CSV · back to the price check.