-
1. Write the specs down in numbers, not adjectives
-
2. Verify vendor references — not vendor brochures
-
3. Request the documentation — and actually read it
-
4. Run the FAT with your bottles, your product, your operators
-
5. Document cleaning/CIP cycles during the FAT
-
6. Inspect before shipment — in person if you can
-
7. Agree on commissioning responsibility
-
What goes wrong most often
If you're sourcing a bottling line — beer, wine, honey, or contract-packed beverages — this is the checklist I wish someone had handed me before my first FAT. I spent six years doing quality acceptance for a mid-size co-packer. We commissioned new lines roughly every quarter. Projects that ran this list end to end took about three weeks to commission. Projects that skipped steps four and six took nine weeks and one recall.
One thing up front: this is not a negotiating tool. It's a checklist to keep you from accepting a machine that runs fine on paper and never quite calibrates in your plant.
1. Write the specs down in numbers, not adjectives
"High-speed," "industrial-grade," "compatible with multiple bottle formats." Those words mean whatever the vendor wants them to mean in month twelve.
Write speed in bottles per hour, with the bottle format attached. "12,000 BPH on 500ml glass" is a spec. "High speed" is a marketing line.
For honey bottling lines, this matters more than most people realize. Honey viscosity shifts with temperature — a line that fills 40 jars/minute at 20°C may only do 30 at 15°C, because the pumps and filler valves are pushing syrup instead of water. If you don't spec the temperature range into the contract, the vendor tests at their lab temperature, which is never your warehouse temperature.
People assume higher-priced vendors deliver better quality. The causation usually runs the other way — vendors who consistently deliver within tolerance can charge more because they know they'll hit the number.
Checkpoint: Every spec in the contract should have a number, a unit, and a test condition attached.
2. Verify vendor references — not vendor brochures
Every supplier shows you their best customer. You want a reference in your product category — same packaging format if possible, similar line speed.
Call three of them. Ask two specific questions: "How long did calibration take after delivery?" And "How many times did the line go down in the first six months?"
Vendors who take a week or more to commission, or whose line goes down monthly — walk away. Politely.
A trap here: assuming a newer supplier will automatically try harder. I went back and forth between our established vendor and a new one for about two weeks in 2023. Established offered reliability; new one offered 20% savings on a honey line. Picked savings. The new vendor's machine was theoretically fine, but their commissioning engineer — I'm not exaggerating — had never commissioned that specific model before. That 20% was eaten by downtime within six weeks. Now every contract we sign includes a clause: commissioning engineer must have 5+ years on the specific model, or the vendor submits a replacement and eats the delay.
Checkpoint: References must match the exact model and application. Ask about downtime, not satisfaction scores.
3. Request the documentation — and actually read it
This is the step most people skip because it's tedious. The vendor sends a folder of PDFs, the buyer forwards it to QA, QA scans it and signs.
What to actually read:
- Material certificates for food-contact parts (EC 1935/2004 or FDA 21 CFR depending on market)
- CE declaration for the electrical panel and control cabinet
- 3-A sanitary standard certification where dairy/adjacent products are involved
- Weld certificates if tanks or piping are fabricated on site
- Calibration certificates for the pumps and valves (relevant for honey and syrups)
Receiving the file isn't the same as reading it. In 2022 I caught a tank certificate listing an alloy grade that didn't match what was actually installed — a "clerical error" on the vendor's side. Left alone, those tanks would have corroded faster than spec'd. A quiet, expensive failure.
Checkpoint: Every serial number on the certificates should trace to a physical part or nameplate.
4. Run the FAT with your bottles, your product, your operators
This is the big one. Most people skip it in the wrong way.
A factory acceptance test (FAT) should run with the unit assembled in front of you. Video is not a FAT. Empty-bottle runs are not a FAT.
Bring your actual bottles — not the vendor-supplied sample. Bring your actual product — not water or a substitute. If you can, bring your operator, or at minimum watch the operators they provide.
For honey lines especially, this is where it matters. Viscosity, crystallization tendencies, and filling temperature tolerance will hide under lab conditions. We've seen honey filling nozzles that "passed" at low volume drift out of fill accuracy once running at production speed — which only became obvious after 30 minutes under load.
Checkpoint: FAT at the stated production speed, with your product. Refuse water or empty-container tests.
5. Document cleaning/CIP cycles during the FAT
CIP (clean-in-place) parameters are the last thing anyone wants to talk about. They're also where the line fails its first real pasteurization audit.
Write it down. Log it. Cycle time, water temperature, chemical concentration, and verification sampling points need to be in the FAT protocol.
I reviewed a brewery's CIP records once where they'd assumed 30-minute cycles on the tanks. Turns out the piping connecting to the filler had dead legs that required manual rinsing. Nobody had documented that, because nobody had tested it. First shipment was rejected on microbial count.
Checkpoint: CIP/CIP-verification logs as part of the FAT protocol. Changes to parameters logged and signed.
6. Inspect before shipment — in person if you can
The vendor wants to ship. They'll send you some phone videos and a set of photos and call it an inspection.
Fine. You don't have to accept that.
A pre-shipment inspection for a large automated bottling line is worth the trip. It usually takes a day. Confirm component packing, verify the spare parts list, check for damage in transit, verify the last open action items are closed out.
Pre-shipment inspection also gives you a glimpse at how the factory actually operates — will plant management know the project details, or will every follow-up go through a sales contact who's already moved on?
Don't skip this one. In one 2023 project the PLC module arrived installed backward. That error would've been caught in an hour at the vendor's facility. It cost us three days of chasing on arrival, because we'd trusted a video inspection.
Checkpoint: Signed punch list. Photos and video don't substitute.
7. Agree on commissioning responsibility
Commissioning is where 30-50% of project time typically goes, and it's the part of the contract most people skim.
Nail down:
- Who provides utilities (compressed air, steam, power, water)
- Who staffs the start-up runs
- Guaranteed ramp-up to rated speed (how fast, held for how long)
- Remediation if it doesn't hit that number
- Spare parts and wear items priced and available after acceptance test
Most contracts say "vendor responsible for commissioning" without specifying where that responsibility ends and who defines when it's over.
Checkpoint: Commissioning plan with milestones, sign-offs, and exit criteria. If it's not spelled out, it isn't owned by anyone.
What goes wrong most often
Ranked by frequency, from what I've seen on our own projects and others' — these are the mistakes that recur:
- Specs written in adjectives instead of numbers
- References checked via brochure instead of phone call
- FAT treated as a formality, not a test
- CIP verification deferred until after install
- Pre-shipment video trusted in place of an inspection
- Commissioning treated as boilerplate instead of project scope
Honestly, if you only get one thing right from this list, get steps 5 through 7 written into the contract. The rest can usually be negotiated after the fact. If those three aren't nailed down, you'll eventually be paying for a machine that had a fixable problem you didn't catch early enough.
One more thing, and it's a fairly strong opinion: if a vendor won't let you run a FAT with your own product, that tells you something. No savings on equipment will make up for a launch that never quite finishes.