What WorksBetter found by reading across whole sets of public records. We read every finished highway contract, every bill and recorded vote of a session, and every project on this site’s ledger. No single publisher states these findings. Each one shows its test or its formula, names its records, and says how to check it yourself. Published .
How to read a finding
Our own work, with the working shown
A finding reads across many public records at once, so no single publisher states it. Each one is the work of WorksBetter, and each is dated. A finding labeled ANALYSIS makes a judgment, such as “usually”. It shows the test it was made against, with the evidence for each part. A finding labeled CALC is arithmetic, with the formula shown.
Every number in a finding is listed under “The numbers” with its formula. A number copied from a record names its publisher instead. Each finding is worked out again from its records whenever this page is rebuilt. If the records stop passing a finding’s test, the finding is not printed until it is reworded or withdrawn. A finding we get wrong goes in the corrections log.
Projects and contracts
Where the project money goes, and what happens to a contract
Of 3,021 finished CDOT construction contracts, 1,935 (64.1%) ended above the amount CDOT awarded them at. Another 1,085 (35.9%) ended below it. 1 ended exactly at its award. The middle contract grew by 3.23%, and 783 (25.9%) grew by more than 10%.
Ended above the award 64.1%
Ended below it 35.9%
Older contracts grew more often. Of contracts awarded from 1994 to 2009, 71.3% ended above the award. For 2010 to 2026 it was 58.7%. A recent contract that is already finished is more likely to be a short one. So this change shows a direction, not proof.
What counts as finished. The newest pay estimate CDOT lists for the contract is marked final (FINL or FINAL). 3,023 contracts qualify. 2 print an amount of zero and are left out, which leaves 3,021.
What is compared. The Award Project Amount against the Current Projected Amount. CDOT prints both on that final estimate.
“Usually” means more than half ended above their award.1,935 of 3,021 is 64.1%, which is more than half.
Open CDOT’s pay-estimate index. For each contract, take the newest estimate, and keep it if it is marked FINL or FINAL and both amounts are above zero. Subtract the award from the current amount and divide by the award. Count the results above zero, below zero and above one tenth, and take the middle value.
Limits
Grew means the contract amount on the final estimate is above the award. The award is one contract’s starting amount, not a project’s budget, and this finding does not say why a contract changed.
Dollars are never added across these three decades of contracts, because a dollar in one decade and a dollar in another are not the same amount.
No contractor is named. This site does not store the names a pay estimate prints.
How often does a finished highway contract use more than its contract time?
On 435 of 3,020 finished CDOT contracts (14.4%), CDOT’s own Percent Time on the final estimate was above 100%. The middle contract used 98.97% of its contract time.
Take the same final estimates as the finding above. Read the Percent Time printed on each, count the values above 100%, and take the middle value.
Limits
We have not confirmed whether CDOT measures Percent Time against the original contract time or the time after approved extensions. So this finding does not say that any contract finished late.
Five of the 20 projects with money approved on this site’s ledger hold 67.2% of their $5.32B. Central 70 Project alone holds 22.6%. Transportation projects hold 85.5%.
Which projects count. The ledger’s own total counts projects that are done, active, funded or halted. Requests are left out. That is 20 of its 50 projects, $5,316.22M in all.
The five largest.Central 70 Project, I-70 Floyd Hill Improvement Project, I-270 Corridor Improvements, I-25 South Gap: Monument to Castle Rock and I-70 West Vail Pass Auxiliary Lanes. Together they hold $3,574M.
“Most” means the five largest hold more than half.$3,574M ÷ $5,316.22M is 67.2%, which is more than half.
Share held by the five largest= (Central 70 Project$1,200M + I-70 Floyd Hill Improvement Project$905M + I-270 Corridor Improvements$725M + I-25 South Gap: Monument to Castle Rock$419M + I-70 West Vail Pass Auxiliary Lanes$325M = $3,574M) ÷ $5,316.22M
On the Projects page, take the ledger projects that are done, active, funded or halted. Sort them by named total, add the five largest, and divide by the sum of all of them. Add the transportation projects the same way.
Limits
This is the ledger of flagship projects this site follows, not all of the state’s capital spending.
Each figure is that project’s own named total. Some are a finished highway’s whole cost and some are one year of money for a building, so the shares compare named totals, not spending.
How much of two big highway projects has CDOT paid so far?
By its September 2026 dashboard, North I-25 Segment 5 (Mead to Berthoud) had paid 74.0% of its contracts’ current value. By its April 2026 dashboard, I-70 Floyd Hill had paid 35.6%.
CDOT’s monthly dashboards for these two projects print, for each contract, its “Payments to date” and its “Current Contract Value”.
We add those across the contracts on the newest dashboard. North I-25 Segment 5 (Mead to Berthoud) has three contracts, and I-70 Floyd Hill has four. Then we divide paid by current value.
CDOT’s monthly project dashboards for the two projects, the newest of each: September 2026 and April 2026. Each project’s pacing page shows every month.
Open each project’s newest monthly dashboards on CDOT’s site. Add “Payments to date” across the contracts, add “Current Contract Value” the same way, and divide the first sum by the second.
Limits
The two projects are at different stages, and their newest dashboards are from different months, so they are not ranked.
Open the figure-setting document. Take its printed “Total Not Recommended”, add the totals of its three request tables, and divide the first by the second.
Limits
This is the recommendation stage, before the budget act. It is not what lawmakers finally funded, and a request is not spending.
In the 2026 regular session, at least 395 of 626 bills (63.1%) were signed into law. At least 133 (21.2%) were postponed indefinitely. In 2025, at least 434 of 658 (66.0%) were signed and at least 115 (17.5%) were postponed indefinitely.
Signed into law, 2026 63.1%
Postponed indefinitely 21.2%
Every other outcome, or none recorded
The rest, in the legislature’s own words. 2026: Deemed Lost: 33; Lost in House: 1; Lost in Senate: 2; Referred to Voters: 1; Vetoed by Governor: 12. 2025: Deemed Lost: 35; Lost in House: 3; Lost in Senate: 5; Vetoed by Governor: 11.
Every bill the General Assembly lists for the session counts once, by the outcome its own page records.
49 bills in 2026 and 55 in 2025 have no outcome in the records we read. They stay in the count. Some of them were signed, the budget bill (HB26-1410) among them, so each share here is a floor.
In both recent sessions, yes. The group of costly bills is small, so this shows a direction, not proof.ANALYSIS
In 2026, 29.6% of final votes on bills with a $0 fiscal note had no “no” votes at all. On bills whose latest note showed $1 million or more in state spending, the share was 6.5% (2 of 31). In 2025 it was 25.4% against 12.2% (5 of 41).
Which votes count. Third-reading votes on the bill itself, in each chamber: 945 in 2026 and 1,031 in 2025. A bill passed by both chambers counts twice.
The two cost groups. We read the state spending in the bill’s latest fiscal note, for its first budget year. One group is $0: 665 votes in 2026 and 666 in 2025. The other is $1 million or more: 31 and 41. Votes on bills in between are in neither group (164 and 198). Votes on bills whose note we could not read are left out (85 and 126).
What is counted. A vote whose printed tally shows zero “no” votes.
The pattern holds if, in each session on its own, the share is lower for the costly group. 2026: 6.5% against 29.6%. 2025: 12.2% against 25.4%. Lower in both.
For each regular session, take every third-reading vote on a bill. Join it to the state spending in that bill’s latest fiscal note. Within the $0 group and the group at $1 million or more, count the votes with no “no” votes and divide by the votes in the group.
Limits
These are tallies of votes. No member, district or party is counted.
The costly groups are 31 votes in 2026 and 41 in 2025, so this finding shows a direction, not proof.
A fiscal note is a forecast for one version of a bill, not its final cost.
Of 3,675 bills whose first and latest fiscal notes both print a budget-year cost, 2,594 (70.6%) never changed. 549 fell, 354 rose and 178 were amended down to $0.
The fiscal notes Legislative Council Staff wrote for each bill, first and latest, as read for Before It’s Law. Each count has its own record on Ask WB.
For each bill, read the state spending in its first fiscal note and in its latest one. Keep the bills where both print a figure. Count the bills where the two are equal, and divide by the bills kept.
Limits
A fiscal note is a forecast for one version of a bill, not its final cost.
A bill whose first note is a sentence rather than a figure is not counted.