Executive Summary

A major sports venue needed to evaluate four international catering suppliers for a £15-18M annual contract, with 1,200+ pages of complex proposals and just 14 days to decide. We deployed AI-powered analysis that completed the evaluation in 48 hours, uncovering £200K+ in hidden costs and delivering board-ready presentations with complete audit trails.

The Challenge

Each supplier submitted 300+ pages in different formats — Excel spreadsheets with embedded formulas, PDFs with complex tables, Word documents mixing narrative and financial data. There was no standardised baseline for comparison.

Food cost percentages ranged from 28% to 35%. Growth projections varied from 3% to 15%. Labour allocation methodologies were completely different. The finance director estimated three weeks minimum for proper manual analysis — time they didn't have.

What We Built

Phase 1: Intelligence Context (30 minutes)

Before touching a single document, we built an industry context foundation — typical stadium catering margins, standard staffing ratios, realistic post-pandemic growth rates, and common contract structures. This context is what made anomaly detection possible.

Phase 2: Multi-Format Processing (45 minutes)

The AI processed 1,200+ pages across all formats, extracting 500+ data points with complete source attribution. Every figure was traceable back to its source document, page, and section.

Phase 3: Multi-Lens Analysis (60 minutes)

Three-dimensional evaluation across base case performance, growth potential, and risk analysis — each supplier assessed against the same standardised framework.

Critical Discoveries

The £200K Staffing Gap

One supplier looked cheapest on paper — until our AI flagged they'd proposed 40% fewer staff than industry standard. Buried deep in the assumptions: the venue would provide sales staff. That's £200K annually hidden from the headline figures.

Unrealistic Growth Projections

Supplier growth projections ranged from 3% to 15% annually. Industry reality post-pandemic: 5-7%. Without context-aware benchmarking, the board might have accepted projections that would create budget shortfalls within 18 months.

Cost Structure Inconsistencies

Food cost percentages varied by 25% across suppliers due to different allocation methodologies. Standardising the comparison revealed that the apparent cheapest option was actually the most expensive.

Delivered Outputs

The finance director received a complete executive decision suite: a 15-slide board presentation with supplier comparison matrices and 3-year projections, a 25-page financial analysis with hidden cost identification, an implementation roadmap with negotiation strategy, and supporting documentation including audit trails, assumption registers, and risk registers.

Every figure traceable. Every assumption documented. Every anomaly flagged.

Results and Business Impact

Beyond the immediate savings, the framework is now reusable for all major procurement decisions — contract renewals, vendor assessments, and partnership evaluations.

Why It Worked

Complex procurement isn't about comparing numbers — it's about understanding the context behind them. When AI understands industry benchmarks, staffing norms, and realistic growth rates, it catches what humans miss in the data avalanche.

The AI didn't replace expert judgement. The finance director's expertise guided the analysis whilst the AI handled the heavy lifting of extraction, standardisation, and anomaly detection.

Key Takeaway: AI-powered procurement analysis doesn't just save time — it fundamentally improves decision quality by catching the anomalies buried in complexity.