How DitchBook is built

This page documents the pipeline behind the directory, the map, and the statistics — what the sources are, how records are joined and verified, what we refuse to publish, and where the gaps are. It exists because a good question deserves a public answer: the numbers below are computed from the same data, in the same build, as the pages they describe.

The pipeline

Seed. The census began from the Open Water Foundation's Colorado ditch and reservoir companies dataset — open data, gratefully credited.
Census. 1,123 of today's 1,390 companies are that original OWF seed, carried over row for row and not individually re-verified by DitchBook beyond the reconciliation below. DitchBook has added 267 more since — each one checked individually against a Secretary of State record before inclusion, never bulk-imported, with ambiguous candidates left out rather than guessed in. The full list, seed and additions alike, is reconciled monthly against the Colorado Secretary of State's business-entity data for current standing (see below). Current census: 1,390 companies.
Monthly SoS refresh. A monthly pull of the SoS dataset re-joins standing, registered agent, and formation date by entity ID. Snapshots are kept, and the published "standings as of" date derives from the newest snapshot's own date (currently July 27, 2026) — never from the build clock, so a rebuild cannot dress stale standings up as fresh.
Water-rights matching. Companies are matched to their structures (WDIDs) in Colorado DWR's CDSS records. Only tier-1 matches are published — see the standard below. Currently 315 published matches (published July 28, 2026), each carrying its water source, earliest decreed right, and a citation link to the state record.
The map. Pins are placed only at real DWR coordinates for a matched company's structure — never a town-centroid guess. 314 of the 315 matched companies are pinned; the remainder have no digitized coordinates in the state's data.
One-pass publication. Every company page, basin index, the map data, the stats, and this page are generated from the same joined dataset in a single build. A number quoted anywhere on the site cannot disagree with the page it links to, because they are the same data. Builds hard-fail if a data source comes up short (a sparse map or a standings join covering less than half the census refuses to ship).

What "tier 1" means

A company-to-WDID match is published only when the state's own records assert it. The match key is the DWR structure record's associated-contacts field — the state's note of which organization is tied to the structure — which must equal the company's registered name exactly after mechanical normalization (case, punctuation, generic legal words; no fuzzy matching in anything published). Three guards then apply: the normalized name must map to exactly one registered company in the census (name collisions demote both candidates), all of a match's structures must sit in a single county (multi-county spread is the signature of a generic name conflating unrelated entities), and only ditch/reservoir-type structures are eligible (a company appearing as a secondary contact on a well must not be credited with the well's rights — a guard added after our own audit caught precisely that failure). The single-county guard belongs to the automated tier. A match promoted by individual verification is not subject to it, because a real irrigation company can and does cross a county line: 14 of the 315 published matches span adjacent counties — Larimer and Weld, Delta and Gunnison, Bent and Kiowa and Prowers among them — and each of those was individually verified against the state record before it was published. A fuzzy name-similarity tier exists in the pipeline for review purposes and is never published. Every published row carries a citation URL to the underlying state record. When post-publication review finds a conflict, the match is demoted and unpublished: on July 21, 2026 an internal audit demoted three previously published matches over basin conflicts. We treat the demotions as part of the record — a verifier with no public refusals is a rubber stamp.

See the full picture: how companies are matched to their ditches →

The numbers right now

Companies in the census: 1,390
Matched to a current SoS record: 1,372 (99%) — standings as of July 27, 2026
Published tier-1 water matches: 315 — published July 28, 2026
Companies pinned on the map: 314

Company profiles

Some company pages carry a fuller profile: the water rights decreed to the company's structures, how much land the state has mapped as irrigated by them, what the diversion records show, and where that land sits. These panels appear only where the underlying public data exists and has been assembled for that company — a page without them is not a company without a history, only one whose records we haven't compiled yet. Every panel carries its own source and date, and every figure is shown as the state records it.

Water-right portfolio — DWR's Water Rights Net Amounts, by structure (WDID): priority, appropriation and adjudication dates, decreed amount, whether the right is absolute or conditional, and the decreed uses. Each panel is stamped with the date the rows were pulled. Amounts are never converted, combined, or divided. DWR also records some net amounts only as alternate-point or exchange (APEX) entries; those are not listed in the table, so the panel says how many exist rather than letting the table imply it is the whole file.
Irrigated acreage history — the Colorado Decision Support System's Irrigated Lands snapshots, which map irrigated parcels in particular years rather than every year. The parcels are mapped by the state. Where a parcel is served by more than one structure, or is also linked to a well, the acres are not all attributable to one company's ditch, and the panel says so in plain language.
Recorded diversions — DWR's annual diversion records, labeled as the state labels them: published records are final; provisional records are not, and are drawn and marked differently so an unfinalized number never looks settled. Annual totals follow the DWR water year, November through October, not the calendar year — and the water year still in progress is marked partial, because a year that is only part-recorded would otherwise read as a collapse in supply. Diversions are never summed across a company's structures: structures on one system can feed each other, so the same water may appear on two records, and because each structure's record begins and ends in a different year a combined series would show a drop that is an artifact of the recordkeeping rather than anything that happened to the water. The chart shows one structure; the table shows them all. A diversion record shows what ran through a headgate. It is not a share, an entitlement, or a company allocation.
Share counts — published only when a document states one, cited to that document with its date, and quoted where the wording matters. A count is what that record said on that date, not a certification of the company's present share structure, and the page says so on its face. Where we have not verified a count, the page says so rather than estimating. DitchBook does not calculate water per share, and no page divides a decreed amount by a share count. Where a court has found a per-share figure, that finding is quoted with its case number, the shares it examined, and the study period it covered — a court's finding speaks to the case in front of it, not to the whole company.
Irrigated land linked to this company — the state-mapped irrigated-parcel footprint for the company's structures, drawn on a map. A service-area outline is sometimes shown behind it, dashed and labeled as context only; it is never used for any figure on the page.
Evapotranspiration — satellite evapotranspiration estimates are not shown on company pages. The renderer exists and stays switched off. An ensemble estimate covers all water reaching a field, including precipitation, wells, and shallow groundwater, so it is not a company's water use and could be badly misread as one. If it is ever published, it will carry that limit on its face.

These panels are a compilation of the public record, offered so a volunteer secretary can see the company's water history in one place without a records request. They are informational, accuracy is not guaranteed, and they are not a determination of historical consumptive use — quantification for water court requires a licensed engineering analysis. The corrections address applies here as it does everywhere else on the site.

Known gaps, stated plainly

18 census companies have no row in the current SoS pull — typically a name or registration mismatch still being reconciled, not a judgment about the company.
Roughly 1,100 further SoS entities are candidates for the census and await individual verification before inclusion; the census understates the true population until that triage completes.
The SoS status "Exists" is displayed as the state records it — it is not the same as "Good Standing," and we don't conflate them.
Water matching is deliberately conservative: many true company-to-structure relationships remain unpublished because they haven't yet cleared the tier-1 bar. The backlog is verification time, not missing data.
Basin assignment prefers the census's IBCC basin field; where blank, it derives from the DWR water division of the matched structure, and companies with neither appear under "Other."

Corrections

If anything on this site is wrong, tell us — corrections ship same-day, and material ones are noted rather than silently overwritten. Missing companies are added the same day they're verified.

Maintainer, disclaimer, and citing this work

This registry is maintained by Nick Davis (DitchBook, Loveland, Colorado — nick@getditchbook.com). It is a compilation of public records offered in good faith, with citations so any claim can be checked at its source; DitchBook is not affiliated with any listed company, the State of Colorado, or DARCA, and nothing here is a warranty — verify against the cited state record before relying on a row for any consequential purpose. Suggested citation: “DitchBook, Colorado ditch company registry, getditchbook.com/methods/, records as of July 27, 2026.”