Nielsen said share was down 1.2 points. Internal sales said billing was up 8%. Both were right. Nobody could explain why.
The company operates across oral health, pain relief, digestive health, respiratory, and vitamins. Each category has different competitors, channels, seasonality, and data sources. A single portfolio, but six separate intelligence requirements.
Nielsen IQ measures market share monthly across GT and MT, but undercounts the pharmacy channel where pain relief and respiratory products move. Kantar tracks household penetration on a different cadence with a different geography taxonomy. Internal primary sales shows distributor billing. Secondary sales from the field is patchy in newly expanded rural territories.
A category head sees oral health share dip 1.2 points in Nielsen. She asks sales. They say primary billing is up 8%. Both are correct: the company shipped more, but a competitor launched a Rs 10 sachet that captured trial in rural outlets where Nielsen coverage is thin. By the time someone manually assembles the picture, the competitor has been in-market for 6 weeks.
Nielsen in one portal. Kantar in another. Internal sales in SAP. Pharmacy data in a quarterly spreadsheet. Brand health in a research vendor's platform. Each technically accessible. None connected.
Different taxonomies, no entity resolution. Nielsen uses "Urban North." Internal sales uses state-level RSM territories. Kantar uses SEC classifications by city tier. A question like "is our share loss concentrated where we recently expanded distribution?" required manual joins across 3 naming systems.
No cross-domain diagnosis. Nielsen shows share is down. Internal data shows billing is up. Neither can tell you that a competitor launched a low-unit-price SKU in pharmacy outlets that Nielsen undercounts. That connection requires crossing system boundaries. No dashboard did this.
Pharmacy channel blind spot. Chemist outlets drive 30-40% of sales in pain relief and respiratory. Nielsen's pharmacy coverage is thin. HCP recommendation patterns were tracked in a separate IQVIA system. An entire channel operated without integrated intelligence.
The Intelligence Warehouse models the business as a single graph: brands, categories, geographies, channels, competitors, and the external measurement systems that track them. When share moves, the graph traces why across system boundaries.
The graph was populated through structured sessions with stakeholders across brand, sales, and analytics using Morrie. Not "what reports do you use?" but "when Nielsen says share is down and sales says billing is up, how do you figure out who's right?" 18 sessions across 12 personas.
18 sessions · 46 entities, 28 metrics, 15 decision rules · Every node traceable to a specific conversation
Cross-domain diagnosis across Nielsen, Kantar, and internal data. When share moves, trace whether it's distribution, pricing, competition, or channel mix. Automated anomaly detection and root-cause decomposition across all six arenas.
The first use case encoded the business model across all six arenas. Market Pulse rode on that same foundation, adding demand signal and competitive intelligence layers at a fraction of the original cost.
| Use Case | BKG Reuse | Accuracy | Time to Live | Cost |
|---|---|---|---|---|
Sales Intelligence Cross-domain diagnosis, share decomposition, competitive response detection | Baseline | 96.1% | 6 weeks | 100% |
Market Pulse Symptom-led social signals, Google Trends demand sensing, competitive creative detection, share-of-search | 72% | 95.2% | 13 days | 10% |
Market Pulse adds social listening, symptom-led search signals, and competitive creative detection entities to the existing brand-category-geography-channel graph. The ontology from Sales Intelligence carries directly because the brand, competitor, channel, and geography model is shared. Only the new signal sources and their associated decision rules need to be built.
Versus the 46 entities, 28 metrics, and 15 decision rules already in the graph.
Two use cases. One knowledge foundation. Six arenas, finally connected.