← Back

Guide · Published November 14, 2026 · Reviewed November 14, 2026 · Ingenious AI team · Not legal advice

← All guides

What Counts as a Consequential Decision? The Lists in Colorado, California, and Connecticut

The short answer: Before any tool question comes the decision question: is this decision on the state's list? Colorado's consequential decisions: education, employment, housing, financial or lending services, insurance, health-care services, and essential government services and public benefits. California's significant decisions: the provision or denial of financial or lending services, housing, education enrolment or opportunities, employment or independent contracting opportunities or compensation, and healthcare services — with advertising expressly not significant. Connecticut's significant effects: financial or lending services, housing, insurance, education enrolment or opportunity, criminal justice, employment opportunities, and health care services. The lists overlap heavily and agree nowhere completely.

Check your duties and dates — free Build your first notice — free

What is a consequential (or significant) decision? It is the category of decision about a person that triggers the ADMT duties — the provision or denial of something that matters to their livelihood, housing, education, credit, care, or coverage. Each regime names its own list, which is why the free checker asks for the jurisdiction before the decision type, and why its answer for an unlisted decision type is 'check the list' rather than yes or no.

The overlap: the five decisions every list shares

Five areas appear on all three lists: employment, housing, financial or lending services, education, and health care. A business making automated-assisted decisions in any of those five should start from coverage in all three states and let the specific tests (thresholds, role, triggers) refine the answer — not the other way round. The sector guides in this series take each in turn: employment, housing, lending, education, and healthcare.

The overlap exists because the lists were written against the same reality: these are the decisions where an automated error costs a person something they cannot simply re-buy. That shared logic is a useful interpretive compass for edge cases — a decision with the same character (high stakes, hard to reverse, data-driven) deserves a careful listing analysis rather than a casual exemption.

The differences: insurance, criminal justice, government services, contracting

Outside the shared five, the lists diverge in instructive ways. Insurance is listed in Colorado and Connecticut, absent in California. Criminal justice appears only in Connecticut's list. Essential government services and public benefits appear only in Colorado's. Independent contracting is named only in California's employment wording — significant there because platform and agency work models turn on exactly that classification.

Each divergence is a trap for a national programme: an insurer assuming California parity, a platform assuming contracting is out of scope everywhere, a benefits administrator missing Colorado's express listing. The per-state treatment of insurance — the cleanest example — is worked through in ADMT Notices for Insurance Decisions.

Worked example: Brightpath Benefits Administration

Example — fictional business, real duty structure. Brightpath is a fictional administrator handling benefits, housing referrals, and contractor onboarding. It mapped its six automated-assisted decisions against the three lists:

California's express exclusion: advertising

California's list carries the only express exclusion in the kit's data: advertising is expressly not a significant decision. The exclusion matters beyond advertising itself, because it shows the list is deliberate — the drafters named what is out as well as what is in. Ad-tech tools that target, rank, or price advertising do not, by that fact alone, make significant decisions under the ADMT regulations.

Boundaries still need care: a tool that begins in advertising and ends up informing a listed decision (a credit offer's terms, a housing listing's allocation) has crossed from the excluded category into a listed one on its facts. The question is always what decision the output feeds, which is the same output-mapping exercise the security exception requires. When the facts are genuinely mixed, record the analysis and take advice — the kit's role is to state the lists accurately, not to stretch them.

Sub-decisions and partial effects: the decision inside the decision

Real workflows bundle decisions: an application involves eligibility, pricing, terms, and sometimes a referral. Listing analysis applies to the bundle's consequential parts. A tool that sets the terms of an offer (rather than its existence) is still influencing a listed decision — Colorado's checker language treats setting terms as material influence. A screening step that only decides who gets interviewed is inside the employment decision, not beneath it.

Partial effects also matter in reverse: a tool influencing an unlisted preliminary step, with the listed decision made independently on other grounds, may sit outside. The discipline is to describe the workflow as it is — which step, which effect, whose outcome changes — and test that description, rather than labelling the whole workflow by its least consequential part.

Using the lists: a working method

The method this guide recommends is mechanical: enumerate your decisions; map each to the three lists; for each listed combination, apply the state's remaining tests (role, thresholds, triggers, dates); record findings per cell, including the negatives with their reasoning. The checker implements the same method interactively — jurisdiction, decision type, role — and returns the duty set or the reason none attaches.

Keep the mapping with the system inventory, and review it when products change: a new feature can move a decision onto a list (an advertising tool that starts setting credit terms) faster than a legislature can. The comparison of full duty sets per state is in ADMT Duties in Colorado, California, and Connecticut Compared, and California's list gets its own close reading in What Counts as a Significant Decision in California?

  1. Enumerate your decisions first — before tools — including sub-decisions that set terms or narrow options
  2. Map each decision against all three lists and record the finding per cell, negatives included
  3. Watch the divergences: insurance (CO + CT, not CA), criminal justice (CT), government services and benefits (CO), independent contracting (CA)
  4. Respect California's express exclusion — advertising is not a significant decision — and map any tool whose output crosses into a listed decision
  5. Review the mapping when products or features change, not just when laws do

Sources, as carried in the kit's own data: the duty summaries in app/rules.py and the checker data in app/checker_data.py, verified against primary or authoritative sources on October 28, 2026. Run the free duties checker for your jurisdiction, decision type, and the tool's role, and compare the kit's plans for the record, builder, and review queue. More from Ingenious AI is on the hub.

Build your first notice — free

Related guides

See all ADMT guides for the full set.

Frequently asked questions

What decisions are on Colorado's consequential list?

Education, employment, housing, financial or lending services, insurance, health-care services, and essential government services and public benefits (SB 26-189).

What decisions are on California's significant list?

The provision or denial of financial or lending services, housing, education enrolment or opportunities, employment or independent contracting opportunities or compensation, and healthcare services. Advertising is expressly not a significant decision.

What effects are on Connecticut's list?

Financial or lending services, housing, insurance, education enrolment or opportunity, criminal justice, employment opportunities, and health care services, as the effects that bring profiling duties into play.

If my decision is not on a state's list, do the duties apply?

The checker's answer for an unlisted type is conditional — 'check the list': if the decision is on the list and the tool's role qualifies, duties attach; if it is not on the list, that state's ADMT duties do not. Record the finding and its date.

Does a tool that only sets terms count?

Colorado's checker treats setting terms as material influence, so yes — influencing the terms of a listed decision is influencing the decision.

Build your first notice — free