How a Nashville Restaurant Group Cut Food Waste 35% in One Quarter

Prism Hospitality operates four full-service restaurants in the Nashville metro - two in Midtown, one in East Nashville, and a newer concept in Franklin. Combined, they were doing about $9.4 million in annual revenue heading into Q3 last year. Their food cost was running at 31.8%, which their director of operations, Todd Menchaca, knew was about 2.5 points higher than it should be.

"We knew waste was a problem," Todd told us. "We just didn't know where it was happening. We thought it was portioning. We thought it was the East Nashville location because it's the highest volume. Turns out we were wrong about almost everything."

The Starting Point: Everything Manual

Before Q3, Prism's inventory process was the same one Todd had used for 15 years in the industry. Weekly paper count sheets, transferred to an Excel spreadsheet, compared against POS sales to calculate usage. Waste wasn't really tracked - it was estimated as whatever was left unaccounted for after the math.

The problem with that method is obvious in retrospect: you can't fix what you can't see. Estimated waste is just a rounding error in a spreadsheet. It has no location, no reason, no item attribution. You know waste is happening, but you don't know if it's happening at lunch service or dinner, in prep or on the line, with proteins or with produce.

When Prism started actively logging waste - item, quantity, reason, location, shift - the first two weeks of data were uncomfortable. Todd described it as "a fire drill when you find out the fire has been burning for two years."

What the Data Actually Showed

Three findings stood out from the first month of structured waste tracking.

Over-prep was the biggest driver, not portioning. The team expected portioning to be the issue - it usually is in high-volume kitchens where line cooks are working fast and eyeballing instead of portioning by weight. But at Prism, over-prep at the start of service was generating 60% of their waste. Prep quantities were based on manual estimates from kitchen managers, not on data-driven projections tied to reservation counts and historical cover patterns by day of week.

At the Midtown location on Monday and Tuesday dinners, prep quantities were calibrated for Thursday-Friday volumes. The kitchen was prepping for 140 covers and serving 85. That excess went in the garbage at close.

Produce was the highest-waste category, not protein. Again, the assumption was proteins. Proteins are expensive, so waste feels worse. But the actual waste by item count was concentrated in produce - lettuces, herbs, sliced citrus, cut vegetables. These items have short shelf lives, are often prepped in batches for consistency, and are frequently over-prepped because they're cheap enough that nobody thinks twice about throwing away a half-full pan of brunoise carrots.

But when you add up produce waste across four locations over 90 days, the number gets uncomfortable fast. Prism calculated it at approximately $14,800 in the quarter before intervention - just from produce.

The Franklin location had the best waste performance, not the worst. The newest, smallest location, which the team assumed was the least efficient, was actually running tighter than the established Midtown spots. The reason, once they looked: the Franklin location had a smaller menu, shorter prep lists, and a kitchen manager who had started counting and logging waste as a personal habit before the formal process was introduced.

The Changes Prism Made

The intervention wasn't complicated. It was disciplined, and it was consistent.

They introduced prep quantity guides tied to projected covers - not as hard limits, but as benchmarks. If the guide said 6 portions of housemade gnocchi and a kitchen manager wanted to make 9, they could, but they needed to note the deviation and track whether it sold through or was wasted. Over eight weeks, the prep guides tightened as the data accumulated. Kitchen managers started trusting the numbers because the numbers kept being right.

They created a waste log requirement at the end of each shift. Two minutes per service - item, quantity, reason. The reasons were categorized: over-prep, spoilage, drop, re-fire, quality reject. That last category revealed a separate problem: at the high-volume Midtown location, quality rejects were running nearly double the other locations, suggesting a line execution issue rather than a purchasing or prep issue.

They changed produce ordering frequency. Instead of two produce orders per week, they moved to three - smaller quantities, more frequent, with less shelf time. The increased delivery cost was more than offset by reduced spoilage. At current produce pricing, they calculated break-even at about 8% spoilage reduction. They hit 22%.

The Numbers After One Quarter

Total measured food waste dropped 35% across all four locations in Q3 compared to Q2. Food cost percentage moved from 31.8% to 29.4% - a 2.4-point improvement. On $9.4 million in revenue, that's roughly $226,000 annualized.

Todd's read on it: "The technology helped us see the problem clearly. But the change was behavioral - we just got disciplined about logging and reacting. The data told us where to look. The team did the actual work."

That framing is worth holding onto. Systems don't reduce waste by themselves. They make the causes visible. The restaurants that close the gap between visible and fixed are the ones that treat operational data as a daily management tool, not a monthly accounting artifact.

What's Replicable Here

Prism's situation isn't unusual. Most multi-location operators running manual inventory processes have similar patterns hidden in their cost structures. The scale of opportunity varies, but the mechanism - waste that's untracked because the logging infrastructure doesn't exist - is consistent.

The minimum viable version of what Prism did doesn't require software. A paper waste log and a weekly review meeting will surface a lot of the same information. What software adds is aggregation across locations, pattern detection across time, and integration with prep guides and purchasing data so the feedback loop closes faster.