Decision AI for Marketing Excellence

The Customer

Global Top 10 Pharma Manufacturer

A global pharmaceutical conglomerate with 50+ brands across Rx and OTC in Emerging Markets, where media is the dominant lever in the marketing mix.

Close to 70% of marketing spend in the Russia region goes to media across TV and Digital. With rising inflation and an ambition to launch newer brands in both OTC and Rx, the company needed the most efficient way to optimise that spend.

At this scale, even small improvements in media efficiency translate into material P&L impact. That is what made media effectiveness a key priority rather than a channel team concern.

The Ground Reality

The Real World Complexity

Every market had its own media rules, its own metrics and its own language, and the know how sat with agencies and country teams rather than in any system.

Each country has different media dynamics across TV and Digital, and the nuanced context usually sits with local agencies. That makes geo aware intervention levers extremely hard to build. In Russia, for example, the NRA group drives a monopoly on rates and constraints around buying TV spots annually.

Performance metrics were non standardised and meant different things in different countries, with files written in local language and usually in messy formats. And because pharma is regulated, every answer needed full provenance and explainability. The margin for a false positive is nil.

People were used to working with BI and xls reports. The AI native way required upskilling and reimagining the workflow, so that agencies and teams could trust AI as capacity augmentation rather than replacement.

The Decisioning Challenge

Decisioning Loop was Slow, Siloed and Broken

Marketing, analytics, agency and data teams spent massive time stitching one loop: what changed, why it changed, and where to intervene.

Tracking sat across multiple siloed SaaS tools, BI and agency reports, while context lived in SOPs, decks and tribal knowledge. Teams spent weeks collating and analysing data, context handoff took months, and blindspots on why, and the next best action, were hard to catch manually.

By the time planning came round, the window to intervene was shorter, and the intelligence to optimise stayed unidirectionally with the agency. That left the in house media team with limited levers to negotiate the annual media plan.

Earlier AI attempts hit a harder wall. AI hallucinates without context and has no pharma domain intelligence, existing intelligence does not self learn or compound, and bringing the data to be AI ready looked like a 12 month project at this scale.

The Architecture

How the DecisionX platform works?

On the way in, the brain reads what the marketing organisation already has. GBQ data covering sales, IQVIA and competitor activity. Agency data including PPT plans, weekly reports and notes. External signals from market inputs and 3rd party data. And the tribal knowledge held by the agency and the team, which had never been anywhere a system could read it.

On the way out, decisions arrive in the flow of work: AI reasoning and decisioning, blindspots surfaced 24x7, AI tracking boards, and reports teams collaborate on directly. Every response carries an auditable trail, governed by the Data and AI teams themselves.

One connected, self learning marketing brain across domain, data and process context.

The Operating Loop

How the Digital Marketing CoE Uses It

DecisionX AI continuously runs the loop of tracking, learning, identifying blindspots and suggesting optimisations.

It tracks around the clock and surfaces blindspots into mail and chat. It reasons across three decisions the CoE owns: annual and quarterly media planning, sharper targeting of the audience, and competition aware placements and campaign recommendations.

Recommended interventions and optimisations land where the team already works, and every outcome feeds back in. What the loop learned this cycle is still there the next one.

TRACK

LEARN

OPTIMISE

The Outcome

Business Impact at Scale

An 11% bottom line gain on a 19 mil $ annual TV and digital spend base, and a team that now spends its time deciding.

Time to hypothesis moved from a week of sprints and context hand off to a 60 second conversation, so the marketing team spends more time on decisioning and action than on analysing across 300+ dashboards. For the data team, a 12 month ontology building journey across structured and unstructured data was compressed to 45 days, producing one fully governed ontology to ground future marketing agentic applications.