If you’ve ever ended an event with a pallet of product that shouldn’t have been there, you know the feeling. Forecasting is one of the hardest parts of running a roadshow business — and bad forecasts have real costs: too much freight, too much product, and staff spending energy managing inventory instead of selling.
We’ve been running the same forecasting engine since ZenShows launched. It was solid — it worked from historical averages, weighted by warehouse rank, day of the week, and the time of year. But it had a consistent habit of over-predicting. Across every vendor we tested, it was calling for 20 to 34 percent more product than actually sold.
We built something better.
What the new engine does differently
The new AI forecasting model learns from your actual sales history — not just averages, but the specific patterns in your data. Which of your warehouses consistently outperform their Costco rank? How do your sales shift across a five-day event versus a two-day one? Which products spike on weekends and which hold steady? The model figures this out from your event records and bakes it into every forecast.
It also works at a finer level of detail. Rather than predicting a single number for an entire event, it gives you a forecast for each day — so you know what to expect on a Wednesday versus a Saturday, and you can plan staffing and replenishment timing accordingly.
The proof
Before releasing this, we ran both engines head-to-head on real historical data — nearly 3,000 events across three active vendors over the past year. The results were consistent:
| Old Forecast | New AI Forecast | |
|---|---|---|
| Average units off per product/day | 2.6 units | 1.6 units |
| Over-prediction (Vendor A) | 34% | 3.5% |
| Over-prediction (Vendor B) | 20% | 4.3% |
| Over-prediction (Vendor C) | 26% | 6.8% |
The old engine over-predicted by an average of 26%. The new one is within 5% — across different vendors, products, and warehouse types. That’s not a minor calibration. That’s a fundamentally more accurate model.
What this means for your operation
Less wasted freight. When your forecast is 26% high, you’re shipping product that doesn’t need to be there. Cut that down to 5% and your logistics costs follow.
Better staffing targets. Per-day forecasts mean your daily goals are set to what’s actually likely to happen that day — not a flat average spread across the event.
A system that gets smarter. The model retrains every night on your latest sales data. The more events you run through ZenShows, the more it knows about your specific patterns.
How to try it
The new engine is available now. Go to Forecast Configure in your ZenShows dashboard, find the Algorithm column next to your retailer, and switch to Enhanced Forecast v2. Hit Rebuild All and you’ll get a fresh forecast run with the updated model. Your locked event goals are untouched — only unlocked events get new predictions.
If you want to compare them side by side before committing, you can switch back anytime. One click.