How a QSR Bakery Location Cut COGS Margin by Nearly 3 Percentage Points

A QSR bakery location operating under a multi-unit franchise portfolio saw its COGS as a percentage of sales drop from 20.6% to 17.7% in under one year while using weekly inventory counts and PAR-level adjustments. This occurred despite rising supplier costs and no changes to its menu or product mix.

If you have ever stood in a storage room with a clipboard and a bad feeling, you already know the problem. The numbers on the page do not match what is on the shelf. You count the same item three times because the number you are seeing does not make sense. You wonder if the order last week was too big, or if someone is taking product, or if the usage data you have been trusting is just wrong.

This is the story of one QSR bakery location that was operating at a 20.6% COGS ratio. That is high. Not dangerously high, but high enough that every monthly review felt like a conversation about why the number was not improving. The answer was always the same: we think we are over-ordering, but we cannot prove it without better data.

What was happening before weekly counts?

The location was ordering from memory and habit. The manager knew the items, knew the delivery schedule, and had a general sense of what sold. But "general sense" is not data. It is a feeling, and feelings do not catch the difference between a slow week and a trend.

Without a weekly count, there was no usage data. No rolling average. No PAR level to compare against. Just a standing order that grew over time because it felt safer to order more than to run out. Running out is visible. Over-ordering is invisible until the monthly COGS report shows up.

How did weekly counting change the picture?

The location adopted a simple weekly counting routine. Every week, the manager walks the storage room with a phone, counts every item, and the app does the rest. It shows usage averages, PAR levels, and a simple color-coded indicator for each item: green means you are fine, yellow means watch it, red means order now.

The count takes 30 to 90 minutes. Before this, the manager was spending 2 to 3 hours on inventory, most of it wasted on manual spreadsheet entry and second-guessing. The weekly count replaced all of that with one structured session.

What the data showed once counting started

Within the first month, the weekly counts revealed the real problem. Several items were being ordered at PAR levels that had been set years ago and never adjusted. The location's sales volume had shifted, but the ordering had not. The result was systematic over-ordering on roughly 20% of the product list.

Here is what the COGS trend looked like once PAR levels were adjusted to match actual usage:

Period	COGS as % of Sales	Total COGS	Total Sales
Prior Year (full year-to-date)	20.61%	$56,989	$276,554
Current Year (full year-to-date)	17.66%	$47,353	$268,079
Change	-2.94 percentage points	-$9,636	-$8,475

Sales actually declined slightly year over year, which makes the COGS improvement even more significant. Most locations see COGS as a percentage of sales increase when sales drop, because fixed ordering habits do not adjust fast enough. This location did the opposite. Sales went down, COGS went down faster, and the margin improved.

How PAR level adjustments corresponded with the reduction

Once the weekly count generated real usage data, the app's auto-PAR feature recalculated reorder points based on a rolling 3-week average. Items that were being over-ordered saw their PAR levels come down. Items that were genuinely running low saw their PAR levels go up.

This is not a one-time fix. PAR levels shift with seasons, holidays, and sales trends. The rolling average catches these shifts gradually, so the location is always ordering based on what is actually happening, not what happened last year.

What this means for operators with high COGS

If your COGS as a percentage of sales is sitting above 18% and you cannot explain why, the answer is almost always in the data you do not have. Without a weekly count, you are ordering from memory. Memory is optimistic. It remembers the stockouts and forgets the overstock.

The location in this case study did not change its menu, its suppliers, or its pricing. It changed one thing: it started counting every item, every week, and adjusted PAR levels based on what the data said instead of what habit said.

$9,636 in COGS reduction was observed in less than a year at a single location. Across a multi-location portfolio, that compounds quickly.

Do locations using weekly counting see lower COGS without changing suppliers or pricing?

The mechanism is straightforward. When you count every item every week, you generate real usage data. That data feeds PAR levels. PAR levels inform order quantities. Better PAR levels mean you order what you need, not what you think you might need. The waste disappears from the order before it ever hits the shelf.

How long before weekly counting shows measurable COGS changes?

This location saw measurable changes within the first month of consistent weekly counting. The rolling 3-week average needs a few weeks of data to stabilize, but once it does, the PAR corrections start immediately. Full-year impact depends on how far off the original PAR levels were.

What if sales are growing? Do PAR levels keep up?

Yes. The rolling average adjusts upward as sales increase, so PAR levels rise to meet growing demand. The key is that they adjust gradually, not all at once. This prevents the common mistake of over-ordering during a growth spike and then sitting on excess product when sales normalize.