A 180-vessel-equivalent NVOCC operating across Asia–MEA trade lanes cut empty container repositioning by 12% in a single quarter. Here's how.
Forecast-driven empty flow rebalancing across Asia–MEA lanes, delivered in 8 weeks on top of existing bookings data.
The customer was already mature — good TMS, clean booking data, disciplined yard ops. Their pain wasn't data quality. It was forecasting. The empty-reposition plan was a monthly spreadsheet, updated weekly, never predictive.
When a lane surprised them (Q4 Red Sea diversions, e.g.) the buffer stock was wrong in the wrong ports. The cost wasn't the diversion itself — it was 3-week lag before the reposition plan caught up.
We stood up the AI Demand Forecaster on 24 months of booking history. Output: 13-week forecasts per lane + suggested reposition flows per port cluster. Runs weekly. Results feed into their existing equipment planning module as recommendations, not mandates.
The key design decision: recommendations, not auto-execution. Planners approve or reject each batch. The AI learns from rejections. After 6 weeks the reject rate dropped from 28% to under 9%.
Empty repositioning: −12% over the first full quarter (baseline: prior 4-quarter average). Equipment utilisation: +6%. Detention exposure at receiver ports: −18%. Gross margin impact on affected lanes: +$540k / quarter at their volumes.
The build was 8 weeks. Go-live was a soft launch on one trade lane, then a rolling rollout to three more over the following 6 weeks.
One email a fortnight. Field notes + case studies + early access to webinars.