Partner data room
What the partner gets, and what it never gets.
The partner receives the program insight it is contracted for, in days, as governed aggregates. It never receives a cardholder row. The fictional partner here is called Aurora.
The tension this solves
The partner owns the brand relationship; the bank owns the account and the regulated data. Neither can act alone after the card is issued. The data room has to let each side see exactly what it is contractually entitled to, fast enough to run the program, without ever handing the partner a copy of cardholder data.
One governed plane, two lenses
Internal teams and the partner read the same governed tables through different, database-enforced lenses. No copy is made for the partner.
One set of governed tables. No copies, no extracts; lineage by construction.
Analysts, data scientists and fraud teams: row-level access across the portfolio, masked by role and by purpose, every read logged.
The co-brand partner: contracted metrics by contracted dimensions, each tied to a contract clause. Any cell under 25 cardholders is suppressed by the database.
How much to share: the spectrum
- 1Atlas ships thisContracted aggregates with small-cell suppression
- 2Co-marketing attribution on aggregates
- 3Overlap analysis in a clean room, no rows exchanged
- 4Event-level sharing for real-time offers
Act as the Partner Aurora analyst in Access controls. Ask for rows and the database returns none. Ask for spend by merchant and most cells come back suppressed.
Act as the partnerWhat the partner sees: the contracted catalogue
Every metric maps to a contract clause and a fixed set of breakdowns. Cells under 25 cardholders are suppressed, value and count both.
By category, region, segment, month, merchant, or program total.
By category, region, segment, month, merchant, or program total.
By category, region, segment, month, or program total.
By category, region, segment, month, or program total.
By category, region, segment, month, or program total.
By category, region, segment, month, or program total.
By dispute reason, region, segment, month, or program total.
By dispute reason, region, segment, month, or program total.
Seven data domains, and which lens each serves
Two of them are named products the partner co-owns: rewards liability and revenue-share settlement.
| Domain | Contents | Consumers | Partner lens |
|---|---|---|---|
| Spend and transactions | Authorizations, settled transactions, merchant category, channel, recurring spend | Partner reporting, marketing, finance | Aggregates only |
| Accounts and cardholders | Lifecycle, balances, credit line, status, tenure | Risk, operations, servicing | Issuer only |
| Rewards and loyaltynamed product | Points earned and redeemed, pre-bought currency liability, transfers | Loyalty finance, partner | Aggregates plus liability share |
| Credit and risk | Underwriting, delinquency, charge-off, fraud, exposure | Risk, regulatory reporting | Issuer only |
| Partner economics and settlementnamed product | Revenue share, marketing-fund accruals, contract KPIs | Partner finance, deal team | Shared, named product |
| Marketing and acquisition | Campaigns, application funnel, approvals, activation, attribution | Marketing, partner | Aggregates plus co-marketing |
| Servicing and disputes | Disputes, contacts, digital engagement | Operations, customer experience | Aggregates only |
- Spend and transactionsAggregates only
Authorizations, settled transactions, merchant category, channel, recurring spend
Used by partner reporting, marketing, finance
- Accounts and cardholdersIssuer only
Lifecycle, balances, credit line, status, tenure
Used by risk, operations, servicing
- Rewards and loyaltyAggregates plus liability share
Points earned and redeemed, pre-bought currency liability, transfers
Used by loyalty finance, partner
- Credit and riskIssuer only
Underwriting, delinquency, charge-off, fraud, exposure
Used by risk, regulatory reporting
- Partner economics and settlementShared, named product
Revenue share, marketing-fund accruals, contract KPIs
Used by partner finance, deal team
- Marketing and acquisitionAggregates plus co-marketing
Campaigns, application funnel, approvals, activation, attribution
Used by marketing, partner
- Servicing and disputesAggregates only
Disputes, contacts, digital engagement
Used by operations, customer experience
Rewards liability and revenue-share settlement are first-class, named products. They are what the deal team and partner finance argue about, so they are the first things to make trustworthy.
With the transaction function roughly break-even after rewards, analytics that grow engaged spend are where a co-brand program earns its keep. Speed of insight to the partner is part of the product.
Tokenization and rewards liability cut across every domain; neither belongs to one team.
Onboarding: partner number N+1 is configuration, not a build
The controls are reusable, so each new program adds rows of configuration against the same, already-tested policies.
- 1Register the program
One program code and one partner persona. Name the data owner and the product manager accountable for the partner's data.
- 2Publish the data contract
The metrics, dimensions, freshness SLA and contract clause behind each tile. The contract is the interface.
- 3Bind entitlements
The partner persona gets aggregate-only access and its own program scope. The existing policies apply unchanged; nothing is re-implemented.
- 4Prove it before go-live
The policy self-test runs every persona, purpose and data product against the database. Go-live is gated on it passing.
- 5Wire settlement and liability
Instantiate the named revenue-share and rewards-liability products for the program.
- 6Sign off and go live
The data owner signs off. The partner is live in days, with nothing partner-shaped copied out of the bank.
How success is measured
| Metric | Target | Note |
|---|---|---|
| Days from signed agreement to first governed report | 10 business days | Target. The differentiator against a bespoke build. |
| Partner metrics traceable to a contract clause | 100% | Target. Every tile names its clause. |
| Cardholder-level rows delivered to the partner | 0 | Enforced by the database, and checked by the self-test. |
| Acquisition to activation funnel | Program KPI | Contracted aggregate. |
| Spend per active cardholder, redemption rate | Program KPI | Contracted aggregate. |
| Settlement accuracy and reconciliation time | Operational | Revenue-share reconciliation built into the settlement product. |
- Days from signed agreement to first governed report10 business days
Target. The differentiator against a bespoke build.
- Partner metrics traceable to a contract clause100%
Target. Every tile names its clause.
- Cardholder-level rows delivered to the partner0
Enforced by the database, and checked by the self-test.
- Acquisition to activation funnelProgram KPI
Contracted aggregate.
- Spend per active cardholder, redemption rateProgram KPI
Contracted aggregate.
- Settlement accuracy and reconciliation timeOperational
Revenue-share reconciliation built into the settlement product.
Why this matters at issuer scale
The largest US issuers run many co-brand programs at once. Each program is a partner that wants its data.
JPMorgan Chase credit card purchase volume in 2024, the largest of any US issuer (Nilson Report).
American Express, second. Citi was third at $616.1B.
Card network, issuer and brand. The brand is typically paid on card spend beyond its own stores or flights, so it wants to see how the program performs.
The credit function earns most of card profitability (about 80%), fees about 15%, and the transaction function roughly breaks even or loses money once rewards are paid (Federal Reserve, 2022).
Programs turn on the cardholder base and pre-bought loyalty currency, so moving a program between issuers is expensive.
Marriott runs co-brand cards with both Chase and American Express; a partner can compare issuers on how well each serves its data needs.
Sources and method
- 01Nilson Report ranking of US credit card issuers - 2024 purchase volume: JPMorgan Chase $1.344T, American Express $1.168T, Citi $616.10B.www.globenewswire.com/news-release/2025/03/06/3038338/0/en/jp-morgan-tops-nilson-report-ranking-of-us-credit-card-issuers.html · as of published March 2025
- 02Federal Reserve, FEDS Notes: Credit Card Profitability - The credit function accounts for roughly 80% of card profitability and fees about 15%;www.federalreserve.gov/econres/notes/feds-notes/credit-card-profitability-20220909.html · as of September 2022
- 03Chase co-branded card catalogue - Current Chase co-brand programs, including Amazon, United, Southwest, Marriott Bonvoy, Hyatt, IHG, Disney, British Airways, Aer Lingus, Iberia and Aeroplan.creditcards.chase.com · as of September 2026
- 04JPMorgan Chase technology blog: the evolution of data mesh architecture - JPMorgan Chase's public description of its data mesh on AWS, with data shared in place rather than copied.www.jpmorganchase.com/about/technology/blog/evolution-of-data-mesh-architecture