Data products

DeepGarden Data

DeepGarden Data is our external data-product and Data-as-a-Service (DaaS) offering. We collect published datasets that are genuinely hard to use — inconsistent file formats, identifiers that change between releases, entities that appear under six different names — and turn them into something you can join and query on the day you receive it. License them on their own, or harmonize them with your proprietary data through Roots.

Discuss a data product See how it works

0

Datasets maintained today, with a working collector

0/yr

Refreshes on enrollment and penetration

0

Contract years crosswalked, 2024–2026

0

Write paths back to any source system

Data catalog

The catalog

Everything we maintain, with coverage, history and refresh cadence. Datasets that are not ready say so. Open any card for the full record, caveats included.

In development: Being built now. We will tell you where it stands and what it does today rather than what it will do.
Planned: Committed direction, not yet started. Do not buy on the basis of this.

Intelligence feeds

Custom intelligence feeds

A feed is more than a dataset: continuous collection, entity resolution, enrichment and the analysis on top, maintained rather than delivered once.

Medicare Advantage market monitor

The six maintained CMS datasets, crosswalked and joined, refreshed on their publication cadence, with market share, penetration and share shift already defined as certified metrics. This is what most healthcare engagements start against, and it needs nothing connected on your side to be useful on day one.

Monthly enrollment and penetration refresh
Annual Stars, landscape, crosswalk and benefit refresh
Crosswalk-aware year-over-year comparison
Certified market-share and share-shift definitions

Network and market disruption monitoring

In development

Market signals collected from trade and regulatory sources, resolved to the same organizations, facilities and counties as the datasets above, and filtered to the markets you actually operate in. Where your membership data is connected, the intent is to estimate potentially affected members — that part is not built.

Continuous collection from trade and regulatory sources
Entity resolution onto organizations, facilities and counties
Filtering to your defined service areas
Member-impact estimation against your data (planned)

Commissioned datasets and enrichment

Custom engagement

A dataset that does not exist yet, or an enrichment layer over one that does. Scoped as an engagement with a defined deliverable, a refresh commitment and an agreement about who owns the result — which is a conversation worth having explicitly rather than discovering at renewal.

New collection against a defined source
Entity resolution against your existing keys
Agreed refresh cadence and quality checks
Explicit ownership and licensing terms

How the data reaches you

Two of these work today and two do not. If your use case depends on one of the latter, say so early and we will tell you exactly where it stands.

Inside Canopy

The datasets are queryable in the workspace immediately, with definitions and entity resolution already applied. This is the fastest route and the one we recommend, because the value is mostly in the resolution rather than the rows.

File extractAvailable through implementation

Scheduled extracts of the normalized tables, scoped to what you license. Set up as part of the engagement rather than self-serve, because the scope and the refresh cadence both need agreeing.

API accessPlanned

A programmatic path into the same governed query execution. Committed direction; not available today. If your use case depends on it, say so early and we will tell you where it stands.

Warehouse sharingPlanned

Direct sharing into your own warehouse. Under evaluation, and dependent on which platform you run. Not something to plan against yet.

Methodology

Where it comes from and what we do to it

Enough to judge whether the work is real, without publishing the pipeline.

Collection

Files are pulled from their official publisher on that publisher's cadence. Source URLs are verified rather than pattern-guessed, because these publishers change URL structures without notice and a silently failing collector is worse than a loud one.

Parsing and normalization

Column layouts move between releases, so parsers are written against the actual file rather than the documentation. Identifiers are normalized: counties carried in both SSA and FIPS, plan numbers zero-padded, segments present even when unsegmented.

Entity resolution

Contracts to plans to segments, plans to counties across both coding systems, contracts to organizations to parents so a competitor question returns the whole group, and years through the official crosswalk. This is the part that takes the time.

Suppression and quality

A suppressed small cell is preserved as suppressed rather than becoming a zero, and any aggregate it affects is reported as incomplete. Ingest runs are checkpointed, so a partial failure is retried rather than leaving a gap.

Provenance

Every row traces to the source file and the ingest run that loaded it — the same chain of custody your own data gets.

What is public, what is ours

The underlying CMS data is public and stays public. What you license is the collection, parsing, normalization, entity resolution, crosswalking and maintenance. We say which is which so you can judge the value honestly.

DeepGarden Data — external data, resolved and ready to use. License it alone, or combine it with your own.