How we show numbers
Every number, and where it comes from
VoteColorado reformats official records. It doesn't score, rate or predict anything. These are the rules behind each number on the site.
Which sources count
Official government documents come first. The primary source for every item is a government record: the Secretary of State, a county clerk's sample ballot or TABOR notice, a city or town's own ballot resolution or notice, the legislature's nonpartisan staff (the Blue Book and fiscal impact statements), or the state Division of Local Government. When a town's legal notice is only online as printed in a newspaper, we use it and say so.
Other sources are context, never the final word. The Colorado Municipal League's preview, news stories and Capitol Lens add context. They never override an official document. When they disagree, we follow the official one and log the change on the corrections page.
Unconfirmed means we haven't found it in an official document yet. A city or town question we haven't found in its county's or town's posted documents is marked unconfirmed, with a note on what we checked. Your county's sample ballot settles it. Every item's source and status is in our source audit (CSV).
Counts of questions
One count per question a voter marks. A town asking two questions counts twice. Places are cities, towns and Niwot, which votes on becoming a town.
Why our numbers differ from CML's. The Colorado Municipal League's preview says voters in "at least 90 cities and towns" will see candidates or ballot measures, and that it tracked 107 issues. Its list names 68 places with ballot questions. We show those 68 plus Brighton, whose South Brighton General Improvement District question comes from a city resolution. The rest of the 90 are towns electing only council members or mayors, which this site doesn't list. CML doesn't publish the list behind its 107, so we can't match that number item by item. We count every question a voter marks separately (Niwot alone has seven). We also list Mountain View's 2C, which isn't in CML's list; see the corrections page.
County, school and special district questions are counted separately: 0 questions in 0 counties so far. A question from a district that spans counties, like the Front Range Passenger Rail District's 7A, is listed under each county where it's on the ballot and counted once in the totals.
Vote needed to pass
Thresholds come from the 2026 Blue Book. Constitutional amendments need at least 55% of the votes cast, except an amendment that only repeals, which needs a simple majority; the Blue Book lists Amendment 87 that way. Propositions and local questions need a simple majority, meaning more yes votes than no votes. Once results are in, the black tick on each bar marks the threshold.
Results
Results come from each county's official election-night site, with the vote counts behind every percent. Local contests aren't reported on the state's page. We show the yes share with its counts ("52% yes, 1,204 of 2,315 votes") and round to the nearest whole percent. Results stay marked unofficial until counties certify them, which is due by Nov. 25, 2026.
On election night we show who's ahead, not who won. From 7 p.m. on Nov. 3 the site reads the state's official results site, which every county reports to, about once a minute. A measure or question reads "Yes ahead" or "No ahead" (an amendment that needs 55% reads "Short of 55%" below it) until we enter the certified result. We tie one of our questions to an official contest only when a town has one question and the county lists one contest for it, or when we've checked the pairing by hand; each question page names the official contest it's counted as. Every other contest the county lists under a town's name still shows on that town's panel, under its official name.
Money figures for local questions
Only from official documents, as printed. Most figures come from the question's TABOR notice: the government's own estimate of first-year revenue and its reported fiscal year spending. When a notice isn't posted online, the first-year amount comes from the official ballot language, which TABOR requires to state it, and the card says so. We don't compute shares or comparisons from them. TABOR fiscal year spending leaves out some money, such as reserves and some grants, so it isn't the government's full budget. Every question gets the same lines in the same order; a line the source doesn't print is left out rather than shown as zero. Where we can't find a question's notice online, the card says that and links to the county clerk, whose office mails the notice to every address with a registered voter.
Tax rates show the rate now and the rate if the question passes. They come from the ballot language, the TABOR notice or the town's own tax page, named on each card. They're the rate of that one tax, not the total you pay at the register, which also includes state, county and district taxes. An extension shows the same rate on both lines. Each figure and rate is read from its source twice, independently, and appears only when both reads agree.
State fiscal impact of statewide measures
Copied from Table 1 of each measure's fiscal impact statement. Legislative Council Staff, the legislature's nonpartisan research office, writes these: the estimated change in state revenue, state spending, transfers and TABOR refunds, for each fiscal year the table shows. Colorado's fiscal year runs July 1 to June 30. We add nothing of our own; a measure with no estimated effect shows $0. The short notes under some measures quote the same statement. These are estimates made before the election, not results.
Candidate photos
Photos appear only when we can show one for every candidate in a race. Until then, everyone in the race gets the same initials circle, so no candidate looks more established than an opponent. We use only photos that are public domain, openly licensed or sent by the campaign with permission to publish. Campaigns can send a photo they own the rights to, with the photographer's name, to hello@votecolorado.org.
AI assistance
Every number comes from an official source and is checked before it appears. Claude, an AI model made by Anthropic, helped gather sources, draft one-line summaries and read figures from PDFs. Each money figure and candidate name was read twice, independently, from the official document, and appears only when both reads agree. Anything not yet confirmed is left out.
What this doesn't tell you
A summary isn't the ballot language. Your county's official sample ballot is the final word, and the TABOR notice mailed to your home has the arguments for and against each tax question.