A batch record is the written account of one production run: which product and which lot code, the ingredients and packaging that went in (with their lot numbers and amounts), the steps and checks along the way, how much came out, and the initials of whoever did it. It's filled in while the batch is made, not afterward. The term comes from the drug and dietary supplement rules, but the same page is what a food inspector, an organic certifier or a wholesale buyer means by "show me your batch records". There's a filled-in example below and a one-page template to download.
The short version
- One record per batch. It's the batch's biography.
- Five parts: what and when; what went in, with lots; what was done, with checks; what came out; who.
- Fill it in as you go. A record written at the end of the shift from memory isn't one.
- File it under the lot code, so "show me lot 260922A" is a page, not a search.
- The blank form is the master record. The filled-in copy is the batch record.
Where the term comes from
In FDA's drug rules, "batch production and control records" must document "that each significant step in the manufacture, processing, packing, or holding of the batch was accomplished", including dates, the equipment used, "weights and measures of components used in the course of processing", in-process results, labeling, and the "identification of the persons performing and directly supervising or checking each significant step" (21 CFR 211.188). The dietary supplement rule lists what its batch production record must include, item by item: the batch, lot or control number; the equipment and lines; "the identity and weight or measure of each component used"; "a statement of the actual yield and a statement of the percentage of theoretical yield"; monitoring results; the packaging and labels used; and a quality review (21 CFR 111.260).
Food makers under FDA's preventive controls rule don't have a section headed "batch record", but the rule's general requirement for records reads like a description of one. Records must be "accurate, indelible, and legible", "created concurrently with performance of the activity documented", and include the date and time, the signature or initials of the person, and "the identity of the product and the lot code, if any" (21 CFR 117.305). Acidified food processors keep processing records for each lot showing pH and the other critical factors, plus a log of any deviation and what was done about it, for three years (21 CFR 114.100).
So the page has a pedigree. Nobody handing you a form cares about the pedigree. They care whether you can prove what went into last Tuesday's batch.
This is general information, not legal advice. Your inspector, certifier or process authority decides what your records need to show.
Master batch record versus batch record
A master batch record is the blank: the recipe, the steps in order, the checks with their limits, and empty boxes. The batch record is a copy of the master filled in for one batch. You write the master once, and version it when the recipe changes. You fill in a batch record every time. That's the whole distinction, and if a customer's form uses both terms, master batch record vs batch record goes into it further.
A filled-in example
This is a batch record from Windward Roots, the sample business inside the TaroStack app. Ana Reyes made kūlolo (a taro and coconut dessert) on the bake line on Tuesday. The recipe per 8 oz tray is 170 g of taro paste, 120 ml of coconut milk and 60 g of sugar, and she made 120 trays' worth.
Header
| Product | Kūlolo, 8 oz tray (KULOLO-8) |
| Lot code | KUL-260922A |
| Date | Tue Sep 22, 2026 |
| Planned quantity | 120 trays |
| Master record | Kūlolo v3, June 2026 |
| Made by | A.R. |
What went in
| Ingredient or packaging | Amount | Lot | Where the lot came from |
|---|---|---|---|
| Taro paste | 20.4 kg | PASTE-2026-09-21 |
made here on Monday |
| Coconut milk | 14.4 L (36 cans) | CM-44017 |
supplier's lot, off the case |
| Cane sugar | 7.2 kg | SUG-0908 |
our code, received Sep 8 |
| Trays with lids, 8 oz | 122 | PO-1211 |
packaging, by purchase order |
| Labels | 122 | PO-1198 |
Check the arithmetic: 120 × 170 g is 20.4 kg, 120 × 120 ml is 14.4 L, which is 36 cans of 400 ml, and 120 × 60 g is 7.2 kg. Two spare trays and labels were drawn in case of damage.
What was done
| Step | Check | Limit | Result | Time | Initials |
|---|---|---|---|---|---|
| Mix paste, milk, sugar | Batch weight | 42.0 kg ± 0.5 | 42.1 kg | 7:40 | A.R. |
| Bake | Oven set | 325°F | 325°F | 7:55 | A.R. |
| Bake | Internal temperature at end | 190°F or above | 194°F | 10:20 | A.R. |
| Cool | Time to reach 70°F | 2 h or less | 1 h 40 min | 12:05 | A.R. |
| Fill and lid | Fill weight, three trays | 8.0 oz ± 0.2 | 8.1, 8.0, 8.1 | 12:30 | A.R. |
| Label | Lot and best-by on label | KUL-260922A, Oct 2 |
ok | 12:45 | M.T. |
The batch weight is 20.4 + 14.4 + 7.2 = 42.0 kg, so 42.1 is inside the limit. Kūlolo keeps ten days, so the best-by is October 2.
What came out
| Trays filled | 119 |
| Rejected after filling (one underweight, one split) | 2 |
| Good trays | 117 |
| Yield | 117 ÷ 120 = 97.5% |
| Trays used | 120 (119 filled, 1 damaged empty); 2 returned to packaging stock |
| Labels used | 119; 3 returned |
| Where it went | 117 trays to the walk-in, bin WALKIN-3 |
Sign-off
Reviewed by M.T. (Marcus Tanaka, operations), Sep 22, 13:10. Deviations: none. Note: "Paste lot was a day old, within its three days."
Every number can be checked against the one before it, which is the test of a good record. The two rejected trays are written down rather than quietly binned, because a yield that's always exactly 100% is the first thing an auditor stops believing.
The template
Download the batch record template. It's the five parts on one sheet, with the kūlolo example on a second tab, in Excel format that opens in Google Sheets and prints on one page. Fill it in on a tablet or print a stack and use a pen; both count. Change the checks to yours.
Make the checks yours
The "what was done" table is the part that differs by product. A hot sauce maker's checks are pH after cooking, fill temperature and the hot-fill hold time. A soap maker's are lye and oil weights and the temperature at trace. A roaster's are charge temperature, drop time and roast loss. For a cookie, it might be nothing more than oven temperature and a count.
Two rules hold for all of them. Every check has a limit written on the master before the batch, so that "result" means something. And if a result is outside the limit, the record says what you did about it, even if what you did was throw the batch away. A record with no deviations ever is either a very good kitchen or an incomplete record.
Where this stops working
Paper batch records are fine for one line and one person. They stop being fine when you need to search them.
"Which batches used CM-44017?" means turning every page since the delivery. They also can't add up: the record
says 36 cans went in, the storeroom says 35 came out, and nobody reconciles the two until a count a month later.
And the 4 a.m. problem: the person mixing dough with floury hands is the one who's supposed to be writing
ingredient lots down, and it's the step that slips. Keeping them is easy. How long
is usually two years or more, and the retention is rarely what fails.
How TaroStack does it
In TaroStack a batch record writes most of itself. When you record a batch against a recipe, the ingredients come out of the lots that expire first, and the record notes which lots it used and how much. The finished batch gets its lot code. Your own checklist fields go on the record, such as cook temperature, pH, fill weight or a sign-off, and each entry is stamped with who recorded it and when, so it was made at the time by definition. You can print the form blank for the line, or fill it in on a phone. Actual yield sits beside the recipe's expected yield, and the cost of what really went in sits beside the standard cost. Lots and recalls are part of every plan; recording batches against recipes, with checklists, is part of Standard, at $99 a month.
