Guides
Specialty food KPIs: what to check when a number moves
Specialty food KPIs read as pairs: which monthly number moved first, which record to open, and how to tell a production, price, demand or records problem.
What to take away
- Read the sheet as pairs. One number moving says little, and two moving together usually name the cause.
- Start with yield, because it moves first, and work down the sheet until the story stops.
- Contribution has four moving partsprice, deductions, unit cost and units. Find which one changed before you change anything.
- Every check ends at a recordthe batch record, the sales ledger, the count or the drill. If the record is wrong, so is the number.
- Some problems never show as a number. They arrive late, as reorders, days to payment and yield.
Pairs of numbers and what they point to
A number that moves is a question, not an answer. The pair it moves with narrows the question to production, price, demand or records, and each pair has one record to open before you decide anything.
| Moving together | Look at first | Record to open |
|---|---|---|
| Yield down, unit cost up | Production: ingredient, machine, person or process | Batch record |
| Contribution down, yield steady | A deduction or a price | Sales ledger |
| Stock up, reorders per sample down | Demand, not production | Finished goods count and sample log |
| Variance up, time to trace up | The records themselves | Finished goods count and drill record |
| Days to payment up for one buyer | That buyer's terms, and whether it is your largest account | Sales ledger |
| Contribution up in one channel | Whether the operation can carry more volume | Sales ledger |
Start with the number that moved first
Open the batch record first. Yield moves before the other numbers, because it compares what the batch delivered with what the recipe expected. If yield held, production did not change, and you can look further down the sheet.
Unit cost is total batch cost divided by sellable units, so fewer sellable units raise it even when ingredients cost the same. That is why yield and unit cost move as a pair, and why the batch record answers both.
A yield that fell after you bought or changed a machine is an equipment question. The station-by-station view in the equipment and setup guide shows where the new machine changed the batch.
When contribution falls, check four parts in turn
Contribution in a channel can fall for only four reasons: price, deductions, unit cost or units. Named variables make each one traceable to a record.
For a product in a month, define U as sellable units per batch, E as the units the recipe expected, and C as the batch's total cost from invoices and hours at a real wage. Then p is price received in a channel, d the deductions there, and n the units sold there.
Named Variables
- Usellable units per batch
- Eunits recipe expected
- Cbatch total cost
- pprice received in channel
Contribution in a channel is (p minus d minus c) times n, with c the unit cost. Work through the four parts for the channel that dropped:
- Price received pdid the buyer pay what you invoiced, or less?
- Deductions ddid a new allowance, fee or return come off the price?
- Unit cost copen the batch record, and check whether yield fell with it.
- Units nlook in the sales ledger for a missing reorder.
If yield held steady while contribution fell, the cause is a deduction or a price, and steps three and four can wait.
C's labor is the line to be honest about. A C that carries your hours at zero produces a contribution figure that vanishes the day you hire. If your labor rate is still a guess, the steps in the specialty food pricing guide show how to set it from batch cost.
Stock rising: is it production or demand?
Stock against shelf life rising while reorders per sample falls is a demand problem, not a production one. Compare stock in weeks with shelf life minus the remaining life the buyer requires, and look at the product that fails that test.
Stock rising while reorders per sample holds is a batch-size question. It was set at the fourth stage of starting the business, and it is revisited whenever demand moves.
Reorders per sample is reorders in the month divided by samples given, per channel. A fall that lasts two months in one channel is the cue to act, but the number cannot tell you that the samples went to the wrong buyer. The habits behind a standing reorder are in how to win buyers.
When the count and the drill disagree
Variance up and time to trace up point at the records rather than the product. Variance is opening stock plus produced, minus sold, samples, dumps and closing stock, and the same sign three months running is the signal.
The number cannot say whether the cause is a record gap or a loss. Only the count and the batch records can, so lay them side by side before you blame a person or a supplier.
Time to trace comes from the quarterly drill: the minutes from picking a lot to listing every unit and buyer. A longer time than last time means the system slowed. A drill done by the one person who knows the system hides how slow it really is.
Protect the records the diagnosis rests on
Every check above ends at a record, and a lost batch record loses the sheet and the trace at once. The sheet is only as true as the batch record and the count, wherever they live. The software and KPI guide covers the five records behind these numbers.
Keeping Records Safe
- Batch record with invoices attached
- Tested backups
- Separate logins
- Plan for a day without the system
The records are also the business's tax records. The IRS says you may choose any recordkeeping system suited to your business that clearly shows your income and expenses, and batch records with invoices attached belong in that picture.
CISA notes that small businesses hold information cyber criminals seek and often have fewer resources dedicated to cybersecurity. For an operation this size, tested backups, separate logins and a plan for a day without the system are the minimum.
Days to payment: the number that closes businesses
Days to payment sits last because it is the one that closes businesses. A rise for any buyer, or any buyer above your terms, is the cue to act, even when contribution looks healthy.
Read it per buyer, not as a blend. The number cannot show that the slow payer is your largest account, and a total can look calm while that one account drifts.
What the sheet cannot diagnose
Whether the product is good, whether a buyer is about to leave and whether the founder is exhausted do not appear as numbers. They show on the sheet late, as reorders, days to payment and yield.
The sheet is a monthly instrument. The weekly conversation with buyers and the daily read of yield are where the early signs live.
Compare with your own history
Compare each number with your own last five batches, last quarter or last year. A figure quoted on a page as a target for any specialty food business knows nothing about your shelf life, channel mix or fixed cost.
Your Own History
12 months
of your own sheet is the only benchmark about you
Compare with your last five batches, last quarter or last year
A generic target knows nothing about your shelf life or channel mix
Acting on a generic target has you fixing things that are not broken. Twelve months of your own sheet is the only benchmark about you.
Where each cause sends you next
When a Number Changes
Yield falling on a recently changed machine?
equipment question, see equipment guide
check batch size at fourth stage
Contribution rising in one channel while the calendar fills is the signal that the operation can carry more volume. At that point the routes and their cost belong to the expansion and market guide.
Common questions
Which number do I check first when a month looks wrong?
Yield. It moves first, so it tells you whether production changed before you look at price, stock or cash.
What if yield held but contribution fell?
Look at a deduction or a price. Open the sales ledger for the channel that dropped and compare what was invoiced with what was paid.
Why do two numbers moving together matter?
A single number can have several causes. A pair such as stock up and reorders per sample down names demand, not production.
Should I use a published target for any of these numbers?
No. A published target knows nothing about your product, channels or shelf life. Compare with your own last five batches, last quarter or last year.
How long before the pairs are useful?
Three months for the direction of each number, twelve for the seasonal pattern. Keep every month; the value is in the comparison.
