How to Forecast Food Production Without Overproducing or Stocking Out

Somewhere right now, a production manager is throwing away product they didn’t need to make. Another is calling an account to explain a shortage they didn’t see coming. A third is staring at a blank production schedule because the one person who knew how to fill it just quit.

Of course, I’m speculating. But if you’ve been running production for more than a year, I’m sure you’ve lived at least one of these scenarios.

The instinct is to get better at forecasting, and that’s not wrong. But a production forecast only helps if that number connects to what comes next:

• whether the materials are there,
• whether the schedule can absorb it,
• whether the floor knows what to make.

Otherwise, it’s just a number in a spreadsheet.

In this article, we’ll look at why overproduction and stockouts are so hard to avoid. Then I’ll walk you through the six stages that take a demand signal all the way to a finished production run, with actual costs and traceability included.

TL;DR

• Most manufacturers treat forecasting as the whole job. It’s only Stage 1 of your production cycle.
• A forecast is just an empty number until it triggers a materials check, a schedule, floor instructions, and a cost calculation.
• Overproduction and stockouts are usually symptoms of a disconnected production cycle, not bad forecasting.
• Production is a six-stage cycle. The data from the last stage is what makes the next forecast more accurate.
• Craft Cannery cut 4 hours from every production planning cycle and can now run a mock recall in 5 seconds. Both came from the same change: connecting production planning stages in their ERP.

Why Most Food Manufacturers Overproduce (or Run Short)

Every production run starts with a question: what do we make, and how much of it?

The honest answer, for most small and mid-size food manufacturers, is: one person checks what sold last week, adjusts by gut feel, adds a little extra just in case, and calls it a plan. What it actually costs shows up in the accounting report, sometime next month. Here’s why.

When One Person Holds the Whole Plan

Behind most production plans is usually one person, a “super planner.” Food & beverage technology expert Marcel Koks describes the archetype well:

Your super planner might not have 30 years under their belt, but you get the type. They’re the person who ends up knowing which customer needs early delivery, which account doubles their order before a holiday, and which recipe takes longer than the system says.

And here’s the irony: the better your super planner is, the less anyone notices a problem — how much production relies on them — until they leave or are out sick during a critical week. That’s when the team discovers the production numbers were based on that one person’s best guess. The one that connected the scattered sales and production data about sales orders all over Excel and paper notes.

When There’s Always a Buffer

Even the best super planner is working from experience, not data. Their estimate is better than anyone else’s in the building, but it still is an estimate. So they do what any reasonable person would: build in a buffer. A little extra, just in case. Running out feels worse than throwing away, right?

But here’s when things go wrong. In food manufacturing, buffers also have an expiry date. A furniture manufacturer can stack the surplus and try again next week. You have 24–48 hours before it’s worthless. And you can’t make exactly the right quantity anyway: minimum batch sizes mean rounding up to fill the mixer regardless.

And when ingredient prices shift on top of that? A Pittsburgh bakery saw the price of eggs jump 345% in a single year. Every unit they overproduced that week cost more than they planned when they decided to make it.

Cherry on top: most manufacturers won’t know how much production costs them until accounting reconciles next month.

When You Don’t See The Actual Cost

Between the production run ending and accounting closing the books, a lot can go wrong.

Your supplier raised prices between deliveries. A batch ran longer than planned. Someone on the floor used more than the recipe called for. And if something goes wrong (a contaminated batch, a supplier issue, a customer complaint), someone has to dig in paper records and figure out which lot went where, making an accurate cost-per-unit impossible to calculate.

Yet, the recipe still says the product costs what it cost last quarter.

Somewhere along the way, you realize you don’t know what your product costs to make anymore. Take Alcides Lopez, Production Manager at Roggenart, a Maryland bakery bistro with nine locations:

That’s what static food costing looks like in practice: a number that was right only when you wrote down the recipe.

Better Forecasting Doesn’t Fix a Disconnected Production Cycle

At this point, your first instinct might be “We need better forecasting”. Chances are you’re picturing some kind of AI-powered demand intelligence platform, and I get it. It’s hard to avoid that conclusion when half the industry seems to be heading that way.

The truth is, the forecast itself isn’t usually the problem. Most manufacturers can get close enough with a few years of sales history.

The problem is what happens after.

A forecast sitting in a spreadsheet is just a number. Knowing you need 4,800 units of chocolate chip cookies by December 1 doesn’t tell you whether the butter is on hand, whether the schedule has room, or what the run will actually cost when it’s done. The forecast needs somewhere to go, and in most operations, it doesn’t have this room.

That’s the gap the next six stages close. It doesn’t need some “cutting-edge” tech, but only a chain of steps in your ERP.

6 Production Stages That Connect Forecast to a Finished Run

At FlexiBake, we work with food and beverage manufacturers every day. The production cycle we see (regardless of what they make or how big they are) consistently breaks down into six stages. Each one feeds the next, and the last feeds back into the first.

Most ERPs will let you run some version of this cycle. I’ll walk you through it on the example of our own software, following a packaged cookie manufacturer preparing for their holiday season.

Stage #1: Demand Forecasting and Sales Planning

→ This is where you get your production quantities: how much of each product to make, and when.

Most manufacturers arrive at that number the same way: someone checks recent orders, tweaks it by gut feel, and that becomes the plan. It works until demand shifts (a holiday runs hotter than last year, a new account changes the pattern, and the like). Suddenly, you’re short by 2,000 units or stuck with excess stock you can’t move.

The problem here is that production never gets a specific number to plan from.

A proper forecasting tool should do three things:

• Pull historical sales data,
• Let you adjust for what the data can’t see,
• Send the result directly into production.

Imagine our cookie manufacturer preparing for the holiday season. Glenn, the production manager, steers the wheel. In FlexiBake, he opens the Forecast Production tool, sets the date range for the holiday period, and selects their cookie product categories. FlexiBake pulls three years of sales history for those SKUs as the baseline.

Then comes the adjustment. For example, this year they signed a new grocery chain, so Glenn applies a 10% growth percentage on top of the historical numbers.

Once the numbers look right, they get a forecast that becomes the production number.

Stage #2: Raw Material Planning and Inventory Check

→ This is where the forecast meets reality (and where most production plans go awry).

Glenn has a number. He knows they need 4,800 units of chocolate chip cookies by December 1. What he doesn’t know yet — unless he physically walks the warehouse — is whether they have enough butter, chocolate chips, and packaging to make them. In most operations, that check happens informally: a purchasing manager eyeballs the shelf, sends a few supplier emails, and hopes nothing arrives late. When something does arrive late, well, you know the rest.

This is when a raw material check should happen. It answers three questions:

What do I need? What do I already have? What do I need to order — and by when?

In FlexiBake, Glenn can:

• open the Raw Materials Required for Open Orders report,
• set the date range for the holiday window,
• and select ingredients.

The system traces every open order back through each product’s recipe and adds up every ingredient and packaging component needed across their entire holiday range.

Before, it’d mean walking the warehouse with a clipboard and cross-referencing recipes one by one. But this way, it takes a few minutes. When any ingredient falls below a pre-set reorder level, FlexiBake automatically generates a purchase order.

Case in point: When Craft Cannery, a sauce and dressing contract manufacturer in Rochester, NY, doubled production in their first year, Paul Guglielmo, the owner, was running around the warehouse before every production cycle, paper in hand, counting ingredients by hand. That’s over now. One raw material report in FlexiBake saves them 4 hours per each production planning cycle.

Now Glenn knows what they have and what they need to order. Stage 3 is where they decide when to make it.

Stage #3: Master Scheduling

→ This is where a list of quantities becomes a production plan assigned to dates, facilities, shifts, and teams.

Without a scheduling tool, the plan lives on a whiteboard, in someone’s head, or in a shared spreadsheet that three people edit simultaneously. But changes don’t propagate. The floor team works from one version while the production manager has another. For manufacturers running more than one facility, it’s even worse because there’s no single view of what’s happening where.

This is where a production schedule comes into play. It needs to answer what gets made, where, by whom, and in what order across every day of the production window.

Here’s how this process looks in Flexibake. Glenn drags their holiday SKUs onto a visual calendar: classic line on Monday and Tuesday, seasonal gift box SKUs the following week, dough prep scheduled two days ahead of baking.

Two weeks of production are scheduled across facilities, shifts, and product lines.

Each item gets assigned to a production facility and a shift, with everything visible in one calendar and production worksheets printable directly from it.

Stage #4: Production Preparation (Work Orders and Sequences)

→ This is where the schedule gets translated into instructions the floor can follow.

A schedule says what to make and when. It doesn’t tell the mixing team how much of each ingredient to pull, in what order to run the products, or what the baking parameters are for the seasonal line that only runs once a year.

Without written instructions generated from the actual schedule, that knowledge lives in the head of whoever has been doing this the longest. When that person is out, the run slows down or stops.

Once Glenn confirms the production quantities, the Baker’s Sheets, Work Orders, and Detailed Production Sheets are all ready to generate:

• The Baker’s Sheet tells each department exactly what they need to make that day.
• The Detailed Production Sheet breaks down every recipe step.
• The Work Orders list ingredient quantities and instructions, organized by department and recipe sequence.

All are ready to print, export, or email to the floor.

Production worksheets and work orders generated from confirmed quantities.

For real teams, this is what replaces the morning of chasing numbers before anyone can start mixing. Case in point: Velvet Hall, Production & Sourcing Coordinator at H&F Bread Co., noted that production reporting now takes them less than 10 minutes, instead of hours.

Now, the floor has everything it needs to run the day. Stage 5 is where they take over.

Stage #5: Execution and Production Entry

→ This is where the plan meets the floor, and actual production quantities are recorded.

In most operations, what actually happens during a production run doesn’t get captured anywhere useful. Someone writes quantities on a clipboard. Someone adjusts the batch size on the fly and doesn’t tell anyone. The run produces 188 units instead of the planned 200, and no one tracks why. By the time the next cycle starts, that information is gone, along with any accurate read on what the run cost is. And a 6% yield gap that repeats every run adds up to hundreds of cases lost across a holiday season.

The fix is simple in principle: record what was actually produced, against what was planned, in the same system the schedule came from.

In FlexiBake, Glenn opens the day’s production run with quantities already filled in, pulled directly from open sales orders. Where the order requires 1,400 units and the mixer runs in batches of 500, the system rounds up to 1,500 and flags the difference.

Before he can close out the day and add finished goods to inventory, Glenn needs to confirm that what’s produced matches what the system expected to consume. If the numbers don’t add up (say, chocolate chips came up 10kg short), Glenn sees exactly what’s missing. He either locates the stock that wasn’t logged, or adjusts the actual quantity produced. Either way, he resolves it before the run closes.

The confirmed production number (what was planned vs what was actually made) is what Stage 6 runs on.

Stage #6: Post-Production, Traceability, and Analysis

→ Aka the most neglected stage, and the one that makes every preceding stage more accurate next time.

In most operations, the run ends and the team moves on. The accountant reconciles costs at month-end. Lot numbers go on a whiteboard. When a customer calls two weeks later about a contamination issue, somebody spends an afternoon digging in paper records trying to trace which batch used which supplier’s ingredients. The information exists somewhere, but nobody can find it fast enough.

A good post-production routine closes three gaps:

• It finalizes what was made,
• Captures what it actually costs, and
• Creates the traceability record that enables quick recall.

Here’s how that works in FlexiBake. Glenn creates the day’s run. Raw materials are consumed from inventory using FIFO — oldest stock first, in the sequence set in Stage 4. Lot numbers are assigned to finished goods. Products are added to inventory and available for fulfillment.

After posting, he checks two things.

First, did the run cost what it was supposed to? The Production Costing Report shows products manufactured, scheduled vs produced quantity , and unit cost — so if butter came in 15% over recipe spec, it shows up here.

Second, was the run on time? The Expected vs. Actual Times analysis shows exactly which recipes ran slower than planned and by how much. If the holiday chocolate chip batch consistently takes 20 minutes longer than expected, that’s a scheduling assumption that needs fixing before next November.

The actual cost per unit that appears here is what the run cost with this week’s butter price and yesterday’s yield (that’s dynamic food costing in action, which we have a dedicated article on).

The traceability story is worth its own moment, as well. Case in point: Craft Cannery runs a mock recall in 5 seconds only. Every time they create a production run in FlexiBake, the system automatically records which supplier lots were used to make each finished product. When a customer calls with a quality issue, that chain (from supplier ingredient to finished case) is queryable in seconds.

This is also where the cycle closes. The actual yield, actual cost, and actual time per batch recorded here are what make next season’s forecast in Stage 1 less of a guess and more of a plan.

What a Connected Production Cycle Looks Like

And there’s that. Our cookie manufacturer started October with a forecast, a new grocery account, and a holiday spike. By the time Stage 6 closed, Glenn knew what they made, what it actually cost, and where every lot went. Next November, that data is his starting point.

When the stages of production aren’t connected to each other, most operations are stuck with products going in the bin, a client on the phone, or a Monday plan that depends entirely on one person showing up.

When they are, something much more pleasant happens:

• The forecast connects to the schedule.
• The schedule connects to the floor.
• The floor connects to the actual cost.

It may not feel like a great transformation. But your operation finally has a memory. That’s worth more than it sounds.