M&R Finance Intelligence HubSnowflake transition governance, transformation knowledge and working workflows
Static interim hub · OOSS-ready
Medicare & Retirement · Finance
One place to see the Snowflake transition, understand the operating model, and find the workflows underneath it.
This interim Finance Intelligence Hub brings the current transition-governance snapshot, architecture and certified-layer guidance, data-ecosystem reference, and the first detailed Finance workflows into one static experience.
The hub is organized by the task a Finance leader or team member is trying to complete, not by the file the information originally came from.
What changed this week?
Portfolio readiness, team movement, blockers, milestones and the operating rhythm.
Where does my team stand?
Readiness percentage, stage standing, 20 proof gates, evidence and the next unlock for each team.
What is the target data model?
Shared source access → team working schemas → Certified Outputs, with the production database context underneath.
What makes an output certified?
Certification criteria, documentation, checks, publication sequence, version handling and implementation controls.
How do Membership and Revenue actually work?
Executive workflow stories plus the source, control and validation detail needed to trust them.
Current portfolio movement
Movement statements are carried directly from the Nov. 6 working governance snapshot.
Source: Governance workbook · Nov. 6, 2026
1.1 · Project status
Snowflake transition governance
Progress is measured by proven future-state capability, not activity completed. The portfolio uses five stages and 20 binary readiness gates per team.
Working governance snapshot: Nov. 6, 2026 · sourced from the calibration workbook and maintained through the governance cadence.
Portfolio readiness at a glance
Where are we overall, what moved, and what needs attention?
Portfolio movement and escalation
The weekly executive pulse reports change, not activity.
Portfolio movement
Decisions / escalation
Proposed transition milestones
Each target is expressed as a capability that must be proven, not simply work completed.
Proposed targets · pending executive calibration
1.1 · Project status · team detail
Team readiness
Each team has the same 20 proof gates. Select a team to see the evidence, open questions and next unlock behind its readiness percentage.
Snapshot: Nov. 6, 2026 · 11 process groups.
Team / process group
Lead
POC
Readiness
Movement this cycle
Attention
1.1 · Key data sources status
Key Data Sources
A working roll-up of foundational source families emerging from Process & Source Discovery—what is mapped, what has a Snowflake target, and where validation remains.
Standing: initial reference, not yet an authoritative source inventory.
Working source-domain roll-up
“Available in Snowflake” reflects what is captured in the working register; it does not by itself establish M&R access, authoritative status or workflow validation.
Domain
Standing
Reach
Mapped relationships
Targets identified
Working status
Next focus
CMS DWHSE · first worked source profile
The source profile exposes upstream enablement, M&R dependency, requested objects and what must be proven before Stage 2 is met.
Draft for validation
Current path
CMS Data Warehouse remains the current intake while the enterprise path is being migrated through MAGNUS / Databricks toward a Snowflake product / subscription path for M&R.
Requested scope: M&R member subset using HICN / MBI identifiers and applicable date filters such as PaymentDateYM or RunDate.
Shared Source Access · final Snowflake mapping to validate
Part D
MOR / Part D actual inputs
CMSMORPartD · CMSMORPartDPOnly
SAS connection to CMS DW · team attribution to validate
Shared Source Access · final mapping to validate
Part C / Bid
Part C Bid Code actual inputs
CMSMORTypeM · CMSMORTypeD
SAS connection to CMS DW · team attribution incomplete
Shared Source Access · final mapping to validate
Other portfolio teams
To be established through process/source discovery
—
Current path to validate
Shared Source Access once dependency is confirmed
CMS DW → MAGNUS / Databricks → Snowflake Data Product → M&R governed source access → team workflows.
Confirm the final Snowflake product / database name and account.
Confirm upstream business, product, engineering and access owners.
Map each requested object to its final Snowflake object and availability date.
Confirm HICN / MBI and date-filter implementation and any fine-grained access requirements.
Complete the 11-team dependency map and document each business use.
Define M&R validation / reconciliation evidence required before Stage 2 is met.
1.2 · Knowledge base · vision
Facts first. Decisions faster.
Build a more connected M&R Finance organization where trusted facts, repeatable forecasts and decision-ready intelligence move across teams with less friction.
Operating principle: standardize the handoff, not the work. Teams keep the expertise, judgment and ownership of how their work is performed.
A more connected Finance organization
Clearer governed handoffs replace fragmented file exchanges. Trusted outputs become reusable and dependencies are easier to understand.
A more rewarding FP&A employee experience
Less time moving files, reconciling versions and rebuilding logic; more time analyzing, challenging assumptions and influencing decisions.
Deeper insight & more intelligent forecasts
Connected inputs and repeatable processes create a stronger foundation for scenario thinking, forecasting and guided intelligence grounded in governed facts and human judgment.
What changes for teams
The future state removes friction around the work without removing the expertise that makes the work valuable.
What stays with the team
Business expertise and context
Forecast judgment and assumptions
Team-specific analytical logic
Interpretation of results and trade-offs
Ownership of contributed outputs
What gets easier
Finding the right inputs
Knowing which version is current
Tracking what changed and when
Sharing trusted outputs downstream
Reusing another team’s proven work
Reconciling across teams and tracing where a number came from
Governance first. Intelligence follows.
AI is not the source of truth and does not replace team judgment. Guided intelligence becomes useful because the facts, definitions, assumptions and outputs underneath it are governed and traceable.
Layer 1 · Governed facts
Trusted definitions, source, lineage, refresh and reusable certified outputs.
Layer 2 · Repeatable forecasts
Controlled assumptions, versioned forecasts, surfaced exceptions and explicit validation.
Layer 3 · Guided intelligence
Decision-oriented questions and scenario exploration grounded in governed facts with human review where judgment is required.
One proof model for every team, with escalation only where broader action is required.
The transition governance model separates visible movement from proven readiness. Every process group advances through the same five stages and 20 gates while keeping its own workflow expertise and implementation choices.
Reference standing: working governance model carried from the Nov. 6, 2026 calibration snapshot.
Five stages define the transition path
The handoff standard is common; the way teams do their work is not forced into one implementation pattern.
Scoring rule: each gate is binary for readiness credit. Four gates × five stages = 20 gates per team. “In Progress” shows movement but does not count as proven capability.
Dry runs pressure-test the governance model before broad adoption
The calibration plan uses real Finance work to expose friction, ownership gaps and control points before the pattern is generalized.
Working plan
October → November · Prepare & align
Document the use cases, map current handoffs, identify performers and consumers, and align on what the future-state simulation needs to prove.
Week of Dec. 1 · Run the dry runs
Work through both use cases end to end using the proposed Stage 1–5 pattern; capture friction, exceptions, ownership gaps and control points.
Week of Dec. 8 · Calibrate & adjust
Refine gates and handoffs based on what worked and what did not before broader application.
Revenue YoY Bridge
Pressure-test a cross-functional path from source inputs and assumptions through the Revenue YoY Bridge and final Finance output.
Question: does the future-state model make assumptions, transformations, unresolved variance, ownership and the final handoff easier to trace and resolve?
TFM Forecasting at HPlan Level
Pressure-test whether forecasting can operate at HPlan level while preserving the controls, assumptions and reconciliation needed for broader FP&A consumption.
Question: can the future-state pattern support HPlan-level forecasting without making the process harder to control or reconcile?
What FP&A owns in the rhythm
Enable the transition, connect teams, make dependencies visible and protect the certification standard.
What remains with each team
Workflow expertise, forecast judgment, team-specific analytical logic, interpretation and ownership of the outputs it publishes.
1.2 · Knowledge base · architecture overview
PROD separates source access, team work and certified consumption.
The long-term operating model keeps teams flexible inside controlled working schemas while standardizing the governed interface used for downstream reuse.
Important distinction: PROD is the environment. Team data schemas are working areas inside PROD. Certified schemas are the read-only handoff for shared Finance information.
1 · Shared source access
Enterprise and shared data made available to Finance teams.
Stepwise | Group MA
FIN360 FICHGL
Epsilon
NDAR / NDB
FICHMMR
UDW / Galaxy
M&R HCE
SMART
CMS / MAGNUS
GPS
These are the production source / data-share names carried by the Certified Layer guide.
2 · Team working schemas
Where teams build, transform, test, adjust assumptions, reconcile and iterate.
MML
RISK_ANALYTICS
MEDICAL_TFM
MEMBERSHIP
REVENUE
MEDICAL_NON_LAG
MR_IND
PHARMACY
SGNA
BCASI
MEDSUPP
PHARMACY_ACTUARY
PHARMACY_DATA_TM
MEDSUPP_FPNA
GROUP_MA
CONSOLIDATIONS_TM
3 · Certified outputs
Read-only for consumers. Documented and registered outputs are published for downstream use.
Operating rule: a working object does not become certified because it exists in PROD. Certification happens when the output is documented, registered and published through the certified process.
Confirmed PROD database context
The operating model is schema-led; these names provide the physical production context currently documented.
Purpose
Production database
MML tenant
UTXZB_PRD_UHC_MR_MML_TENANT_DATA_DB
Main M&R Finance database
UTXZB_PRD_UHC_MR_FIN_DATA_DB
Risk Analytics
UTXZB_PRD_RISK_ANALYTICS_DATA_DB
Transition-state context
The governance architecture describes a PROD-only long-term model and carries a prior DEV transition target of Early November. That target remains source text, not a statement that the transition is complete.
Transition state · DEV
Existing development activity transitions out as the future-state PROD model is established.
Future-state · PROD team working area
Teams retain flexible working space for build, transform, test, reconcile and iterate.
Governed handoff · Certified
Ownership, definition, rules, lineage, refresh, QA, assumptions/version and change history travel with reusable outputs.
1.2 · Knowledge base · Certified Layer requirements
Build where you work. Consume from Certified. Publish through a documented Certified Output.
Certification gives a consumer a stable, read-only handoff plus the context required to understand what the output is, who owns it, what it passed, which version is current, and where it goes next.
Source guide: M&R Finance · Snowflake production operating guide · 21 October 2026.
What makes an output Certified
A Certified Output is a documented table or view published in a team’s CERT_* schema. It is not the same thing as an enterprise Data Product.
01DocumentedPeople can understand what it is and how to use it.
02RegisteredIt is configured in the Certified Publishing application.
03OwnedA named team or owner is accountable for it.
04Quality-checkedIts defined publication checks have passed.
05Version-controlledConsumers can identify which published version they are using.
06Published through the certified processIt reached CERT_* through the established publishing path.
Document the Certified Output once
Downstream users should not have to rediscover meaning, ownership or lineage every cycle.
Documentation area
What a consumer should understand
Purpose and ownership
Business purpose, producing team, business owner, technical contact and support path.
Structure and grain
Certified location, what one row represents, important keys, fields, units and null meanings.
Definitions and calculations
How important measures are defined and calculated, including aggregation or filtering rules.
Time, scenario and version
Period, forecast cycle, scenario, revision, cutoff and as-of context.
Assumptions
Material assumptions, scope and effective period, plus what changed in a new version.
Sources and transformation
Origin and important transformations or dependencies between source and output.
Checks and reconciliation
What must pass before publication and what the output reconciles to.
Publication and history
Who can submit/review, how active version selection works and how prior versions are referenced.
Downstream consumers
Who uses it next, what for, frequency and whether publication notification is useful.
Required: identify who depends on the Certified Output. Optional: notify registered consumers each time a new certified version is published.
Publishing is short because the contract is already known
The output must be documented and registered before it appears in the application.
Select Certified OutputChoose a registered output.
Choose data / versionSelect the candidate revision.
Key distinction: documented means we understand the output. Registered means it is configured and available in the application. Certified means a specific submitted version completed its defined publishing requirements and was released for read-only consumption.
First 45–60 days
Teams keep running existing workflows while priority outputs are identified, documented, registered and progressively moved into the certified publishing process.
Progress is not how many workflows are blocked. It is how much important Finance information flow becomes documented, discoverable and ready to publish.
Version model
Working: revisions can change without affecting consumers.
Approved / Active: the certified consumer interface points to the approved revision for the selected context.
Locked / historical: prior official versions remain identifiable for reproduction and traceability.
Technical implementation controls
Control
Implementation expectation
Certified Output registry
Stores certified schema, owner, documentation, checks and downstream-consumer information.
Submission identity
Each candidate receives a durable request / revision identity tied to the exact payload and documentation version.
Validation
A failing required check stops publication; results remain tied to the tested revision.
Review separation
Working/review revisions are restricted; ordinary consumers see only the approved certified interface.
Approved selection
Approval changes the consumer selection atomically so partial change cannot expose an inconsistent state.
History and traceability
Publication records preserve output, cycle/version, sources/assumptions, checks and human decisions needed to reproduce the handoff.
Consumer notification
Notification remains separate from the publication transaction so a messaging failure does not undo publication.
Reference notes
External links preserved from the source guide; they are references, not runtime dependencies.
Understand where the data lives before deciding how to connect it.
The goal is not to move every source into M&R. The goal is to establish an authoritative, governed path into M&R—and an equally governed path for M&R outputs to be reused downstream.
Standing: draft for validation · source location, ownership, account relationship and access mechanism still require source-by-source confirmation.
Two source situations, one M&R operating model
First establish whether the authoritative source is already in the enterprise Snowflake ecosystem; then establish the governed M&R access mechanism.
Already in UHGSDRP
Authoritative data exists in Snowflake outside the M&R consumption boundary. Sharing / subscription / applicable governed access is still required.
Already in another SDRP Snowflake account
Cross-account access depends on the actual account and tenancy relationship.
Not yet in Snowflake
Identify the authoritative owner and the upstream ingestion / transformation / publication work before M&R can consume it.
Availability ≠ access. A source can already exist in Snowflake and still require sharing, subscription, role configuration and validation before M&R can use it.
M&R platform position · working view
Account, resource group and tenancy are different boundaries.
UHG validation required: confirm UTXZB tenancy type and the authoritative hierarchy used for M&R documentation.
Consumption patterns
The source location and account / tenancy relationship determine the technical mechanism; every path converges on governed M&R source access, team-owned work and certified publication.
Scenario A · UHGSDRP upstream
Source → UHGSDRP producer → Data Share / applicable mechanism → M&R Shared Source Access → Team Data Schema.
Scenario B · another SDRP account
Source → other SDRP-managed account → approved cross-account access → M&R Shared Source Access → Team Data Schema.
Team A Data Schema → validate / certify → Certified Data Schema → Team B / FP&A consumption.
Outside M&R: a Certified M&R output can become an enterprise data product published to the enterprise / HCP Data Catalog for governed subscription. Internal certification and enterprise publication are related but not identical.
10 things to remember
The working mental model retained from the ecosystem artifact.
Start with where the authoritative data lives today, not where we want it to end up.
Separate “already in Snowflake” from “upstream enablement required.”
Data in Snowflake may still require governed share, subscription, role or other access.
Snowflake account, resource group and tenancy are different boundaries; UTXZB is the M&R tenancy in the current working model.
Do not rebuild an authoritative upstream product just to move it into M&R when an approved reuse path exists.
When a source is not in Snowflake, the dependency shifts upstream to the authoritative owner and enablement work.
“Shared Source Access” is a business-friendly consumption boundary, not a claim that every source lives in one literal schema.
Teams retain flexibility in how they build; what gets standardized is the handoff.
Reusable M&R outputs cross the Certified boundary with ownership, definition, lineage, refresh and validation context.
A source is transition-ready only when its authoritative path is accessible and validated for the workflow.
Plain-language glossary
Snowflake Account
Top-level Snowflake boundary where platform resources and data are hosted; account relationship helps determine access.
Tenancy
Logical construct grouping environments and Snowflake objects under customer ownership; the current M&R working reference is UTXZB and tenancy type remains to be confirmed.
Resource Group
Administrative container for tenancy resources and deployment / administration visibility.
Data Share
Snowflake mechanism for sharing selected objects read-only across accounts without recreating the source.
Data Catalog / Subscription
Enterprise discovery and subscription path for published data products; related to but not synonymous with a Snowflake data share.
Gateway Role
Working SDRP mechanism described for governed read-only schema access across supported tenancies within the same resource group; applicability depends on the actual relationship.
Shared Source Access
Business-friendly name for the incoming M&R consumption boundary where approved enterprise sources become available to teams.
Certified Data
M&R output published for downstream reuse with ownership, definition, rules, lineage, refresh, QA/reconciliation and version context.
1.2 · Workflows · Membership
How the Membership forecast is built
The workflow is a two-process chain: first build controlled monthly forecast detail, then enrich and publish it in a form downstream Finance users can consume.
Working view · owner validation required. Source-supported synthesis from the Membership workflow deck and discovery register.
The answer in one line
Actual membership + approved assumptions + plan structure are allocated and rolled forward, then enriched with business / accounting tags and published as detailed forecast data plus tie-out / reporting materials.
Process 2 · Enrichment + publication Actuals + forecast
Combine historical actuals + forecast months
Apply market + product tags
Apply GL, risk, legal entity + customer tags
Normalize to Type / Type2 / Count
Run controls + publish final detailed data
Final reporting grain: one row per detailed intersection, month and Type2; measure = COUNT. Process 1 detail is HPBP · Segment · SCC · DEC ID · Group · Month.
Outputs
Detailed forecast data
Monthly Adds, Terms, Switches and Ending Membership at detailed reporting dimensions.
Locate, assign owner, document versioning and decide which rules become governed reference data.
Already established
Two distinct processes: forecast build, then enrichment / publication.
Actual membership, assumptions, structure/mapping and product factors drive Process 1.
Market/product and accounting/risk/organizational references enrich Process 2.
Process 1 output becomes a Process 2 input; final datasets are published for broad reuse.
Controls span allocation, tagging, duplicates, negative values and reconciliation.
Still required for an authoritative source contract
Confirm the two process / model names and primary vs secondary business output.
Resolve ownership conflicts across Switch_Forecast, Structure Grid, tags, legal entity and actual membership.
Confirm Excel storage locations, sheets/ranges and Group / C&S DSNP variants.
Confirm refresh/as-of rules, history depth and authoritative vs supporting classification.
Identify the PS9 GL object and exact LE Novations / bid-group files.
Validate whether archive, 31JAN23 and New Cap Membership are truly in the workflow.
Evidence standing: source-supported synthesis. Process naming, ownership conflicts, consumer completeness and several file/table details remain open for Membership-team validation. Snowflake locations shown are only those already captured in the discovery register.
1.2 · Workflows · Revenue FP&A
Two chained revenue models, one shared input spine.
The Current Year model completes risk scores against actuals; the Future Year model projects them forward using bid trends. Most of the input spine is produced outside Revenue FP&A.
Working view · owner validation required. Discovery status remains “In Discovery”; source authority, refresh/as-of, history depth, access and migration standing are not established in the supplied material.
The answer in one line
General Ledger actuals + the Membership forecast + member cohort buckets + mapping / tagging reference data + team assumptions are staged in SAS/FIN360, modeled at HPBP / Grp_nbr / SCC grain, tied out to the RAF accrual and ledger, and published as two forecast tables.
Six critical input families
General Ledger actualsThree years of GAAP ledger, M&R and C&SPS9_GL_MR/CS_GAAP 2024 · 2025 · 2026 · owner conflict documented
Membership forecastForecast membership base from MembershipMBR_FORECAST_31JUL26(_CS) · vintage conflict across evidence
Member cohort / risk bucketsMembership by cohort used to complete YTD risk scoresMBR_Bucket_2026_GRP · bucket summaries · informed by restated MMR
The source evidence points to governed access and handoffs as the main transition work—not blind reconstruction of every current implementation.
Consolidator, not a rebuilder
Twelve of the fourteen named current-year inputs are owned outside Revenue FP&A. The gating work is governed source access plus ownership—not recreating every upstream logic path.
Membership is the first real cross-team test
Four inputs come from Membership. A governed Membership output is a natural first consumer handoff into Revenue rather than a file copy.
Staging names are not production contracts
REVENUE_SANDBOX and MMLTARG names document development staging in the source material; they are not treated here as approved production or certified locations.
Keep the judgment, move the mechanics
The register already notes FP&A development to remove two Excel dependencies: prospective manual adjustment and ESRD/Hospice override.
Discovery trace · conflicts and validation required
Where the flowchart and discovery register disagree
Neither source is silently preferred. The difference stays visible as validation work.
Item
Flowchart says
Discovery register says
Why it matters
Ledger ownership
FICHGL — Accounting
FDM Team
The ledger tie-out is a release control; accountable ownership must be clear.
Future-year trend ownership
Bid Actuarial Team (Jamison T.)
Revenue FP&A
This input drives the future-year projection.
Membership forecast vintage
MBR_FORECAST_31JUL26(_CS)
31JUL26 with note referencing 31MAY26
The Membership artifact publishes 31AUG26; a single governed vintage is required.
Member premium ownership
URS Group FP&A Team
URS / Revenue
The PMPM override needs an accountable approver.
Four staged Excel inputs
Names REVENUE_SANDBOX objects
Snowflake target blank
Need to distinguish actual staging from planned staging.
Seven workbooks + two bucket summaries
Not shown
Recorded as source rows
The flowchart understates the manual surface.
RAF accrual input
2026 EOL / Final RAF Accrual → REVENUE_SANDBOX.C03_RAF
No matching row; C03_RAF_Cap_Adjustment exists separately
The current-year tie-out target is missing from the controlled register.
Current-year forecast as FY input
REV_FCST_2026_5_7_2026_YOS
No row for the internal handoff
Vintage convention and internal handoff both need recording.
Process basis
Forecast only
Two rows carry Actual basis
There may be an actuals process not covered by the flowchart.
Future-year summary line
Closing line describes the output as Current Year Revenue Forecast
n/a
Appears to be drafting carry-over; confirm intended wording.
Confirmed by supplied evidence
Two chained models exist and Current Year output feeds Future Year.
Both are staged in SAS / FIN360 by connecting Excel to existing SAS tables.
Output grain is HPBP / Grp_nbr / SCC.
Both output tables and five downstream consumers are documented.
Release is gated by two tie-outs per model plus SAS-and-Excel review.
Twelve of fourteen named CY inputs are owned outside Revenue FP&A.
FP&A development is already underway to remove two Excel dependencies.
Working interpretation
C01 / C02 / C03 / C07 prefixes likely correspond to numbered model components.
“_CS” likely indicates a parallel Community & State run.
Three years of ledger history appear required because three ledger years are consumed.
Revenue FP&A appears to act primarily as a consolidator plus its own risk-score and assumption logic.
Membership is the strongest first governed handoff candidate.
REVENUE_SANDBOX / MMLTARG appear to be development staging names.
Validation required before a build plan
The seven modeling steps and which are genuinely Revenue FP&A logic.
Authoritative standing, refresh/as-of and history for every source.
Snowflake availability and access standing.
The ownership conflicts and Membership forecast vintage.
Functions of C02_Retro, HPBP Yearly Mapping Tag and the bucket summaries.
Location / owner of the SAS reformatting program for C03_CY_Final.
Whether GPS actual rows belong to this process.
The governed future-state output and tie-out approval owner.
Discovery standing: synthesis of the Revenue forecast flowchart and 28 Revenue FP&A rows in the Process & Source Discovery Register. No validated as-of date is recorded for this group; access and migration are “Not Assessed” and multiple governance fields are blank.
2 · Executive Intelligence
Executive Intelligence
Reserved for the governed fact, forecast and decision experiences that will sit on top of the certified data foundation.
Placeholder · not yet implemented
No executive-intelligence product is published in this temporary hub yet.
The intended direction is guided intelligence grounded in governed facts, repeatable forecasts, explicit assumptions and human judgment. This placeholder intentionally contains no invented KPI, forecast, decision feed or AI capability.
3 · Support request
Support request
This section will point to the future support / ticketing system when that destination is available and approved.
Placeholder · link not yet supplied
Support intake is not connected in this version.
No ticket URL, owner, SLA or routing logic was supplied, so this hub does not invent one. Add the approved destination here when it exists.