ADU builder statistics

What the licenses actually say

Classification counts across the 735 companies in the store. Because a license may carry several classifications, the column sums to more than the company count — 584 firms carry exactly one, 94 carry two, and a handful carry six or seven.

HyreADU count over the company store, 2026-09-05. The B row is 100 per cent because holding B is part of the promotion rule.

What is informative is the shallowness below it: outside electrical and plumbing, no additional classification appears on more than ten companies, so the ADU trade is a general-building trade that subcontracts almost everything.

The full classification list and what each one permits is published by CSLB at cslb.ca.gov.

How often the license record is not clean

License status at harvest is the number a homeowner should care about most, and it is the reason the attrition study exists. Of 965 distinct licenses held by firms that were building ADUs in five California jurisdictions, roughly one in four was not in a state that permits contracting when we looked.

Every row is a HyreADU count over CSLB license detail pages harvested 2026-08-29 to 2026-08-30. CSLB LicenseDetail.aspx rejected an automated request on 2026-09-05 (web application firewall). Every status on this page is the 29-30 August 2026 harvest, not a 5 September 2026 live census.

The median-age column is the more interesting one: a revoked license in this cohort was a median 12.3 years old and an active one 15.3, so the failure mode is not simply young firms disappearing.

Enforcement outcomes — what CSLB actually does when a complaint is upheld — are the subject of the enforcement study and the enforcement record.

Why licenses get suspended

Suspension sounds like a finding about workmanship. In this cohort it usually is not. CSLB records the reason on the license detail page, and the pattern is overwhelmingly financial-compliance rather than disciplinary.

HyreADU count over the same harvest. Bond and workers’ compensation between them account for 59.6% of recorded suspensions. A license can be suspended because a bond lapsed on a Tuesday and reinstated on a Thursday — which is precisely why a status is a moment rather than a verdict.

Reasons are recorded per license and one license may carry more than one reason, so the total here (57) can exceed the 44 suspended licenses in the status table.

Does an older firm mean a safer bet?

The intuitive answer is yes, and the record is more complicated. Splitting the cohort by the year the license was issued shows the newest licenses are less likely to still be active — but the oldest are not the most likely to be either.

HyreADU calculation from CSLB issue dates. Whole-cohort median license age is 14.8 years, with a quartile range of 9.2 to 21.7 years. In the company store — a different and narrower frame — mean license age is 17.09 years, with 5.4% licensed five years or less and 29.1% more than twenty.

What license age is and is not. It measures how long this license number has existed.

It does not measure how long the people have been building, and it survives changes of ownership badly: 138 records in this cohort (14.3%) carry the entry “license reissued to another entity”.

Treat license age as a weak signal, and treat a firm’s claimed years in business as an unverified one.

Small firms, and what size predicts

The ADU trade in this frame is a trade of very small operators. Grouping the cohort by how many distinct ADU addresses a firm appears at shows both the shape of the market and a modest relationship between scale and a clean record.

HyreADU calculation over the harvested cohort. Read the direction, not the magnitude. The differences between bands are small and the bands are unequal in size, so this is a weak signal and not a screening rule. What it does establish is the shape of the market: 501 of 965 licenses sit in the smallest band.

Firm-size counts here are per jurisdiction. A contractor working across two cities is counted once — 964 of the licenses appear in a single jurisdiction and 1 in three.

Active license, inactive business

A current license tells you the paperwork is in order. It does not tell you the firm is still building ADUs. Crossing license status against the date of the firm’s most recent ADU permit separates the two.

HyreADU calculation joining the CSLB harvest to the permit frame. Only 185 of the 738 active licenses — 25.1% — were named on an ADU permit issued in 2024 or later. The rest hold a current license and last appeared in this trade earlier in the window.

That gap is the single most useful thing on this page for a homeowner. “Licensed” and “currently building accessory dwelling units” are different questions, and only the first one is answered on a business card.

How these businesses are structured

Entity type, bond and workers’ compensation are all published on the license detail page, and all three are worth a homeowner’s attention before a deposit changes hands.

HyreADU counts over the store and the harvested cohort. Entity shares are computed over 737 license files in the store; bond and workers’ compensation over the 965 harvested licenses.

The workers’ compensation row deserves a sentence. 32.6% of harvested licenses were recorded exempt. An exemption is legitimate for a genuinely single-person business. It is not legitimate cover for a crew, and the consequences of an uninsured injury on a residential site land on the property owner. Ask, and ask for the certificate.

Who actually builds the unit

The only survey evidence on this question belongs to Chapple, Ganetsos and Lopez, who asked 674 California ADU owners who did the physical construction. It is cited as theirs, it describes a 2018–19 cohort, and the question allowed multiple answers.

Chapple, Ganetsos and Lopez, CCI, 22 April 2021, p. 35 (Terner-reviewed). Question: “Who did the physical construction on your ADU?”, n = 674, select all that apply — which is why the shares sum above 100 per cent.

Terner’s own footnote on the related ease-of-construction question (n = 664) notes that it “This may include issues finding a contractor”.

Every row here is a survey estimate from a 2018–19 cohort, not a 2026 measurement. It sits on this page as the one piece of evidence about owner behavior, not as a description of the current market. Our own frame measures contractors of record on permits, which is a different universe entirely.

Where the firms are

The store is concentrated in the jurisdictions we harvest, which is a fact about our method rather than about California. We publish it so nobody mistakes the concentration for a finding.

HyreADU count, 2026-09-05. Headquarters addresses are more dispersed than the qualifying jurisdictions: 158 distinct cities appear, led by San Francisco (122) and Los Angeles (79). County is recorded on 322 of 735 files, which is why we do not publish a county map.

This is not a ranking of ADU markets. Los Angeles has 61.6% of the store because Los Angeles publishes the most complete permit data, not because it necessarily has the most builders. The permit data directory explains which jurisdictions publish what.