3 min read Data & Information

From Business Information Models to Data Products

Stop curating definitions. Start shipping data people can trust and use.

The Challenge

Your business doesn’t need another glossary. It needs data that leaders can bet decisions on this quarter—fast, reliable, owned, and auditable. If your “information model” can’t help a COO decide in 60 seconds or a CRO close an audit point, it’s theatre.

What’s broken with traditional business information models

Common problems

  • Definition paralysis: months arguing over terms; decisions still wait.
  • No owner, no SLA: who fixes freshness or accuracy when it breaks?
  • Pretty lineage, poor accountability: diagrams explain history, not responsibility.
  • Reports that don’t reconcile: MI packs conflict; no one knows which number wins.
  • Stale on arrival: models drift as products, vendors, and rules change.

Move to Data Products

  • Treat critical datasets as products with a purpose, a named owner, service levels (SLOs), controls, and a roadmap—just like any digital product.
  • Glossaries still matter—but they become by-products of running real data, not the main event.

What is a data product?

Data Product = a managed dataset (or event/API) built for a specific business job, delivered with guarantees.

Each data product has

  • Purpose: the decision it serves (“single, trusted customer-risk status for onboarding & reviews”).
  • Owner: one accountable person (business), one technical owner (delivery).
  • Access & Contract: where to get it, schema/meaning, policy.
  • SLOs: freshness, accuracy, availability, completeness.
  • Controls: checks/evidence for audit; exception queue with SLAs.
  • Roadmap & Cost: improvements, deprecations, and unit cost transparency.

Example: “Customer KYC Status”

  • Purpose: Provide a single, auditable KYC status per customer for onboarding decisions and periodic refresh.
  • Owners: Head of Financial Crime (business) + Data Platform Lead (tech).
  • SLOs: Accuracy ≥ 99.5%; Freshness ≤ 60 minutes; Availability ≥ 99.9%; Exceptions cleared < 24h.
  • Contract: events/kyc-status:v3 + table risk.kyc_status_v3 with field meanings.
  • Controls: source-to-target checks, rule versioning, evidence stored automatically.
  • Consumers: Onboarding workflow, Periodic Refresh engine, MI/Reg reporting.
  • Roadmap (90 days): retire legacy table v2; add adverse-media flag; de-dupe vendors.
  • Why business should care (real outcomes): -40% backlog, fewer QA fails, audit closures, run-rate savings.

Why a business architect should focus on data products?

For long Business Architects have been developing and maintaining Business Glossaries, Business Information Models.  These have helped build the knowledgebase, but are not readily used by the executives in decisions.  Data product solves this issue.  As with Information concepts, development of data products needs to be led by business architects. Below are key activities and collaborators:

A. Pick the fights worth winning (Weeks 1–2)

  • Activity: Identify the top 3 decisions slowed by bad/untrusted data.
  • Outputs: Decision statements; candidate data products; success metrics (money/speed/risk).
  • Collaborate with: COO/Operations, CRO/Risk, Product Owners, CDO/Data.

B. Define each Data Product (Weeks 2–3)

  • Activity: Write the Data Product Canvas (purpose, owners, SLOs, contract, controls, consumers, 90-day roadmap).
  • Outputs: 1-page canvas per product; owner acceptance.
  • Collaborate with: Business owner, Data Engineering, Platform, InfoSec, Compliance.

C. “Lineage on a Page” (1 workshop per product)

  • Activity: Map source → checks → rules → publish → consumers, with four health numbers (freshness, accuracy, open exceptions, last rules change).
  • Outputs: Executable lineage one-pager; named accountability at each hop.
  • Collaborate with: Data Eng, Source App Owners, Risk/Compliance, MI team.

D. Make quality and recon real (Ongoing)

  • Activity: Tie two critical DQ checks to each capability outcome; specify reconciliation at decision boundaries and automate evidence.
  • Outputs: DQ-to-SLO map; Recon spec (what, frequency, SLA, owner).
  • Collaborate with: Ops controllers, Finance, Risk, Audit, Data Eng.

E. Replace PDF packs with MI as a Product (Weeks 3–6)

  • Activity: Define an Exec Scorecard (1 page) + Explorable view; retire weekly email packs on a schedule.
  • Outputs: Live scorecard; decommission plan; hours saved → £ value.
  • Collaborate with: COO office, MI/BI, Product, Data Platform.

F. Governance & funding hooks (Monthly)

  • Activity: Add data products to portfolio reviews; tie funding to SLO attainment, duplicate-source retirement, audit closures, decision lead-time delta.
  • Outputs: Monthly “receipts” pack; tiered funding recommendations.
  • Collaborate with: CFO/Finance, Portfolio/PMO, Risk Committee.

Who should care (target stakeholders & your USP to each)

COO / Head of Operations

  • Value proposition: Faster, cleaner decisions; shorter queues; fewer manual workarounds; visible before/after metrics.
  • Soundbite: “One source, owned and measured, that cuts cycle time.”

CRO / Risk & Compliance

  • Value proposition: Controls by design; evidence captured automatically; exception queues with SLAs; audit findings closed.
  • Soundbite: “Prove it in 60 seconds—freshness, accuracy, last rules change.”

CFO / Finance

  • Value proposition: Run-rate savings from retiring duplicate sources; unit cost per metric; funding tied to tier-moving outcomes.
  • Soundbite: “Less spend, fewer reconciliations, clearer payback.”

CDO / Data Leadership

  • Value proposition: Shift from governance theatre to governed products; accountability at each hop; adoption that sticks.
  • Soundbite: “From policies to products with SLAs.”

Product & Engineering Leaders

  • Value proposition: Stable contracts; fewer rework loops; faster releases; clear ownership of upstream data.
  • Soundbite: “A contract you can code to—and trust.”

MI / BI / Analytics

  • Value proposition: Stop patching; consume reliable products; move from slide decks to live scorecards.
  • Soundbite: “Less wrangling, more signal.”

Internal Audit

  • Value proposition: Lineage that names accountable owners; automated evidence; faster closure of repeat issues.
  • Soundbite: “Traceable, testable, owned.”

Closing thoughts

This month, pick one decision that routinely stalls for want of “the right number.” Stand up one data product for it: name the owner, publish SLOs, draw Lineage on a Page, retire one duplicate source, and replace one PDF pack with a live scorecard. Report the before/after in 30 days.

If it doesn’t speed decisions, reduce waste, or close risk, stop. If it does, scale it. That’s the difference between information about the business—and data your business can use.

Share

Deepak Mohan