Case Study

Automated Category Prioritisation

A business needed to allocate marketing budget across hundreds of categories and multiple markets. Every decision depended on performance data, external constraints, and opportunity size — but the process was manual, inconsistent, and had no way to check whether the underlying conditions could actually support the investment.

The Framework

An automated decision process that checks conditions before it recommends action.

How it decides
01
Check market readiness
Can the market absorb more investment? Capacity, quality, and output trends are evaluated. If conditions aren't right, the framework blocks the recommendation.
02
Evaluate performance
What's the return on existing spend? Is there enough data to be confident? Evaluation windows are standardised with fallbacks when recent signals are thin.
03
Size the opportunity
What's the incremental potential? Is it driven by a broad base or concentrated in a few large accounts?
04
Recommend and rank
Every category gets one action — Activate, Scale, Grow, Test, Hold, or Pause — and a priority score. One ranked list across all markets.
What it catches
Capacity constrained — do not invest
Demand surging while quality declines. More spend makes things worse.
Strong return, healthy conditions — invest now
Proven performance with room to grow. Allocate budget.
Revenue concentration risk
Most revenue from a few large accounts. Scalable opportunity is smaller than the headline.
Constraint lifts in two months — time the investment
Month-level tracking means the team plans around windows of opportunity.
What It Delivers

Feed it data. Get a prioritised, documented output.

The entire framework runs as an automated pipeline. New data in, fresh recommendations out — every category reclassified, every signal recomputed, every recommendation traceable to the data that produced it.

The team opens one ranked list instead of cross-referencing spreadsheets. Any stakeholder can audit how each recommendation was made.

A single prioritised list Every category ranked by a composite score balancing opportunity, performance, and constraints.
Self-documenting output Every workbook embeds the decision logic. No external documentation needed to understand the recommendations.
Constraint calendars Month-level visibility into when conditions change — plan investments ahead rather than react.
Fully automated From data ingestion to final output. No manual steps, no copy-paste, no spreadsheet cross-referencing.
Adaptability

Same automated pipeline. Different data realities.

Rich data environments

Pre-built constraint models with month-level granularity, historical performance data, and incremental revenue projections. The pipeline reads existing models and classifies with high confidence.

In practice: month-level liquidity models provided constraint calendars showing exactly when each marketplace category opens for investment.
Month-level precision · 200+ categories

Sparse data environments

No pre-built models, limited history, variable depth. The pipeline computes its own signals from raw data — year-on-year trends in capacity, quality, and output — with stricter thresholds to compensate.

In practice: across 8 emerging markets, supply-side signals were computed directly from trading data where no constraint models existed.
Computed signals · 8 markets · 500+ categories
Impact

From gut feel to defensible decisions.

Before

  • Investment decisions based on gut feel
  • No check on whether conditions could support it
  • Budget allocated by familiarity, not data
  • High revenue mistaken for high opportunity
  • Framework logic lived in someone's head

After

  • Systematic prioritisation, fully automated
  • Every recommendation is condition-gated
  • Single ranked list by composite score
  • Concentration risk separated and flagged
  • Every output embeds the decision logic
What This Means For You

Every investment decision should be defensible.

We build automated prioritisation frameworks that balance opportunity against constraint — so your team invests where it matters and stops spending where it doesn't.

Adapts to data maturity Blocks constrained investments Every recommendation is traceable

Delivered and running in production across multiple markets.