.png)
Sushi at 5 p.m., yesterday’s salad, a sandwich from yesterday’s lunch: For highly perishable fresh products, there’s no second day of sales. Whatever doesn’t sell today is lost—and that includes the full cost of goods plus labor.
At the same time, the penalty for having too little stock is just as severe: an empty sushi shelf at 12:30 p.m. drives away exactly the customers who pay the most—the lunch rush. No other product category penalizes planning errors so directly in both directions.
Why Fresh Produce Is the Ultimate Challenge in Planning
The sales window is incredibly short. When it comes to sushi, poke bowls, salads, and sandwiches, it’s not the day that counts, but the hour: the majority of sales happen during the lunch rush. The amount produced in the morning determines sales for the entire day.
Demand is driven by the weather and the day of the week. Salads and cold bowls are popular when it's sunny, while soups and hot snacks are in demand when it's cold—the same location can have a completely different demand profile from one day to the next.
The margin depends on the shrinkage rate. Fresh products generate good margins—but every instance of shrinkage eats into those margins many times over. Just a few percentage points less in shrinkage can make or break the profitability of the entire category.
How Forecasts Twist the Problem
The daily question—how many units per product and location?—is a classic use case for machine learning: GoNina uses sales data to learn how factors such as weather forecasts, the day of the week, holidays, and location drive demand, and provides specific daily production volumes for the next seven days.
The difference compared to estimates based on experience becomes apparent at the extremes: on atypical days—the first warm day of spring, a long weekend, or the end of a vacation. That’s exactly where the biggest markdowns and the most frustrating clearance sales occur—and that’s exactly where the data advantage is most valuable. Businesses reduce their excess inventory by up to 52% and increase sales by up to 6% with GoNina .
Three Practical Guidelines for Fresh Produce Assortments
1. Plan by product, not by category. “Salads” isn’t a planning unit—the Caesar salad and the poke bowl have different sales trends. Forecasting by item shows which products thrive in certain weather conditions and which ones sell consistently.
2. Track the time when items sell out. A shelf that is empty every day at 1:00 p.m. consistently loses afternoon sales. For fresh products, the time when items sell out is the most important metric alongside the write-off rate.
3. Plan for a second production wave. Where the workflow allows, intentionally setting aside a post-production buffer in the late morning is more effective than any large safety margin in the morning—the forecast shows on which days the second wave is worthwhile.
Frequently Asked Questions
Do forecasts work even for products with low sales volumes?
Yes—especially with small quantities, every error carries relatively significant weight, and the model uses patterns from similar products and locations to reliably forecast even low-volume items.
We outsource production and place orders daily—does that still help?
Yes, in that case, the forecast serves as the basis for ordering rather than as a production target—the effect on markdowns and availability is the same.
Conclusion
Highly perishable fresh products are the most challenging category to plan for in the foodservice industry—and at the same time, the most rewarding when it comes to data-driven forecasting: No other category translates precision so directly into profit margins. Those who forecast sushi, salads, and sandwiches on a per-product, per-day basis sell more and throw away less.
To learn how AI forecasting works in the restaurant industry as a whole, check out the complete guide for the restaurant industry.
