What this is, and what it is not

The standings rank 41 remodeling contractors in Thurston County, WA on public signals: their Google reviews and their Washington contractor registration. Placement is earned. No company can pay to appear, to rank higher, or to be removed.

It is not an inspection, a recommendation, or an assessment of workmanship. We have not visited a job site, spoken to a customer, or reviewed a contract. A high rank means a business has a strong public review record and a verifiable registration. It does not mean the business is right for your job, and a low rank — or no rank at all — is not evidence of a problem.

Always verify a contractor's registration, bond and insurance directly with the Washington State Department of Labor & Industries before hiring. Our data is a snapshot taken on 22 August 2026.

Who was included

We ran 36 searches against the Google Places API — every combination of 6 trade terms and 6 cities — and collected up to 20 results each, de-duplicated on place_id. The full query list is published with each edition.

Location bias

47.0379, -122.9007 with a 25 km radius, centred on Olympia, WA — county seat

Area

ZIP 98501-98516, plus 98530, 98540, 98556, 98576, 98579, 98589, 98597, matched on the first five digits

Categories included

General Contractor, Remodeler, Home Builder, Kitchen Remodeler, Bathroom Remodeler, Construction Company

Categories excluded

Deck Builder, Roofing Contractor, Plumber, Electrician, HVAC Contractor, Landscaper, Painter, Flooring Contractor, Window Installer, Fence Contractor, Concrete Contractor, Handyman

Collected

22 August 2026

A business appears in this edition only if the search found it. A contractor who does not maintain a Google Business Profile will not appear, however good their work. That is a limitation of the method, not a judgement.

How the score is built

Each company is scored out of 100 across four components. A company needs at least 3 reviews to be scored at all.

Reputation quality — 35 points

A raw average is unreliable when it rests on few reviews: three five-star reviews would otherwise outrank ninety at 4.8. Each rating is pulled toward the field mean in proportion to how little evidence supports it.

adjusted = (v / (v + m)) × R + (m / (v + m)) × C

points = ((adjusted − 3.0) / 2.0) × 35, floored at 0

R is the business's average rating, v its review count, C the mean rating across the ranked field (4.859), and m the prior weight (10) — roughly, how many reviews of evidence it takes before a business is judged mostly on its own record. A 3.0-star adjusted rating scores zero; a 5.0 scores the full 35.

Review volume — 35 points

More reviews mean more evidence, but with diminishing returns — the hundredth review tells you far less than the tenth. Scaled logarithmically against the largest count in the field.

points = log₁₀(v + 1) / log₁₀(v_max + 1) × 35

v_max for this edition is 122.

Review recency — 25 points

A strong record from four years ago describes a different business than a strong record from last month. Scored in bands by days since the most recent review.

Profile completeness — 5 points

Whether the business publishes the two things a homeowner needs to contact it: a website (3 points) and its opening hours (2 points).

Registration date is published as a matter of record. It is not scored, and it is not a measure of years in business — see below.

What we decided not to measure

Three signals were dropped because they did not measure what their labels claimed. We publish them here because a scoring system is only as trustworthy as its account of its own failures.

Years registered

Earlier versions awarded points for how long a business had held its Washington registration, presented as tenure. It is not tenure. A registration record belongs to a legal entity, and it resets when a business restructures — so a firm operating continuously for years can show a registration date of months. In this county, nine businesses with substantial review histories carried registration dates under two years, while a quarter of the field sat at the old twenty-year cap, flattening everything from twenty-one to thirty-eight actual years into an identical score. It penalised restructuring and erased genuine longevity at the same time. Relabelling it would have left a meaningless weight ordering the standings, so it was removed.

Photo count

Effectively every profile in the field returned the same number. A signal that does not vary cannot distinguish anything.

Owner response rate

The data source does not report it reliably enough to score.

How businesses are matched to registrations

Matching a Google listing to a state registration is the step most likely to produce a false statement about a real business, so the rule is deliberately strict: a similar name is never sufficient on its own.

Company names collapse together under normalisation — many businesses named "<Something> Construction" reduce to the same tokens. Publishing a registration number on name similarity alone would attach one company's licence to another company's listing. A match is only published when an independent field agrees:

• Near-identical name and the ZIP or phone number agrees
• Similar name and the phone agrees and the ZIP or city agrees
• Anything else goes to human review, with the reasoning recorded

A business that resolves to two different registrations is never auto-resolved. Known matching artefacts are blocked outright, and the build refuses to publish if one reaches a ranked row.

Businesses we could not verify

16 businesses in this edition were collected but could not be matched to a registration confidently enough to publish. A name mismatch is common and is not evidence of a problem. Businesses trade under names that differ from the ones on their paperwork, and our matching is conservative by design. These companies may well be properly registered.

We list them by name and city, and publish no registration data for them at all — including the candidate records our matching rejected, which would be misleading to show. 3 further businesses had fewer than 3 reviews and could not be scored.

Corrections

If something here is wrong about your business, tell us and we will correct it. That includes a registration we failed to match, a business we placed in the wrong county, or a listing we missed. Corrections are made to the published edition and noted.

We do not remove a company from the standings on request, and we do not change a rank on request. Those are the two things that would make the ranking worthless. Write to corrections@contractorstandings.com.

Version history

v1.3 — 22 August 2026

Years-registered removed and its weight redistributed to volume and recency. Registration matching now requires independent corroboration; name similarity alone no longer resolves a match.

v1.2 and earlier

Owner response rate and photo count removed as unreliable.

Changes to the scoring are decided from evidence about what a field measures, and are applied before the ranked set exists — never chosen after seeing which companies a change would favour.