Factory Audit Checklist for Display Suppliers: 10 Checks
Almost every display factory visit follows the same script: reception, a company video in the meeting room, a walk past a few machines, coffee, and a quotation the next morning. Buyers leave impressed and no better informed. The tour answers “do they have equipment” — a question that was never in doubt. It does not answer the question that decides whether the next three shipments arrive identical to the last one.
An audit is not a tour. It is a short, structured attempt to settle three things: what standard this plant uses to call a unit good or bad, whether the line can hold that standard without supervision, and whether a failed unit can be traced back to a date, a line and an incoming batch. Machine count, showroom samples and framed certificates are supporting evidence at best — often they are decoration.
What follows is the checklist we would want a buyer to run on us: three documents to request before booking a flight, five stations worth walking, a 10-point score sheet, a remote version for when travel is not practical, and a 90-minute route for when it is. If you are still at the earlier question of whether the company is a manufacturer at all, settle that first — it is a different exercise with different evidence.
What a Factory Audit Should Answer
Three questions, in order of importance. Everything you see during a visit should be read as an answer to one of them.
- Is the standard written down? Ask for the inspection standard for the exact model you intend to buy. If the dead-pixel limit, brightness range, colour tolerance, touch test method and appearance criteria are not on paper, “quality control” is one inspector’s judgement on the day.
- Can the line hold that standard without being watched? A written standard plus records is evidence. A written standard with no reject rate, no rework tickets and no calibration dates is a document, not a process.
- Can a failed unit be traced back? Pick any carton in finished goods and ask which work order, line, shift and incoming batch produced it. If the answer takes more than a few minutes, traceability is a sales story.
A plant can be large and still fail all three. Capacity tells you how much it can make, not how consistently it makes it — and for most buyers, consistency is what the money is actually buying. The same logic applies when you compare OEM and aftermarket screens: the label on the box matters far less than the process behind it.
Three Documents to Request Before You Book Anything
Ask for these before scheduling a visit, and ask for them for the model you actually intend to order — not for the company in general. A supplier who can produce all three within a day is organised. One who sends a generic quality policy instead has already told you something useful.
| Document | What It Proves | What Good Looks Like |
|---|---|---|
| Inspection standard (SIP) | Whether “good” was defined before production rather than after a complaint. | Model-specific limits per item — pixel defects, brightness range, colour difference, touch method, appearance — with a revision number and date. |
| Outgoing inspection record (OQC) | Whether the standard is applied to production, or only quoted to customers. | Batch or work order number, sample size, items checked, defects found, pass or fail, inspector and date. |
| Nonconforming material record | What happens when the line fails — the part of quality systems buyers rarely think to ask about. | Failure mode, quantity, disposition (rework, scrap or concession), root cause, corrective action and sign-off. |
Read them together, not one at a time. A standard with no nonconformance history usually means failures are not recorded, not that they do not happen. An OQC report that has never failed anything is a report written to be shown.
The Five Stations to Walk
Walk the line backwards: start at the reject bin and the packing table, then move upstream to the stations that produced those rejects. Forward tours are choreographed. Reverse tours are not.
| Station | What to Look At | Evidence to Ask For | How It Gets Faked |
|---|---|---|---|
| Incoming inspection | Whether incoming panels are batched separately and sampled against a stated plan. | Incoming batch numbers, sampling record, a marked isolation area for rejects. | A bench with calipers on it and no records anywhere in the room. |
| Lamination and assembly | Whether bonding is genuinely in-house, how bubbles are removed, how the room is controlled. | Work order for the batch you picked, equipment parameter sheets, cleanroom class and humidity logs. | “We have our own bonding line” — while the work order number does not exist in their system. |
| Aging and test | Whether the racks are actually running, for how long, and under what conditions. | Aging logs with start and end times, test conditions, a failed sample you can hold. | Empty racks and a claim that “every batch is aged” with nothing written down. |
| Final inspection | How a unit is judged: dark-room lighting, grey levels, touch response, appearance. | Physical limit samples, the defect decision record, the name of the person who signs release. | A “PASS” sticker and no limit sample box in the room. |
| Packing and traceability | Protection in transit, retention samples, and whether labels link back to a batch. | The sample store, the label rule, and a live trace of one carton. | A sample cabinet that is empty, or holds only competitor products kept for benchmarking. |
Two details are worth more than everything else on this page. The first is the limit sample box at final inspection: if it exists, pass and fail are defined by physical objects everyone can point at. If it does not, inspection is verbal. The second is the reject bin — ask what the units in it failed on, and whether that failure was written down. Nobody stages a reject bin.
The 10-Point Score Sheet
Score each item 0, 1 or 2. Two points means the evidence exists and is current; one means partial, verbal or out of date; zero means absent. Ten items, twenty points, one number you can compare across suppliers.
| # | Item | 2 Points | 0 Points |
|---|---|---|---|
| 1 | Written standard aligned to your specification | Model-level SIP with revision number and date. | “We follow industry standards.” |
| 2 | Physical limit samples | Dark room, dated limit samples, signed decisions. | Judgement depends on which inspector is on shift. |
| 3 | Key processes in-house | Bonding and assembly carry an in-house work order. | Outsourced, and no work order can be produced. |
| 4 | Verifiable test coverage | Per-unit test records, or a documented sampling plan. | “100% inspected” stated verbally, nothing recorded. |
| 5 | Traceability granularity | Traceable to work order, line, shift and incoming batch. | Traceable only to the shipping month. |
| 6 | Nonconforming material control | Isolated area, written disposition, signature. | Rejects stacked beside the line for anyone to pick up. |
| 7 | Equipment calibration | Labels valid, calibration register available. | No labels, or expiry dates already passed. |
| 8 | Controlled environment | Stated cleanroom class plus temperature and humidity logs. | “It is a clean room” with no monitoring record. |
| 9 | Changeover and first-article check | First-article record, named approver, retained copy. | No first-article concept at all. |
| 10 | Complaint closure | 8D or equivalent, with verified corrective action. | Replacement sent, cause never investigated. |
A total in the 14–20 range means you are buying process capability, not just product. Seven to thirteen is the normal result for a competent mid-sized plant, and it is enough to start a defined trial. Zero to six does not automatically mean the supplier is dishonest — it means you cannot rely on their internal control, so the inspection has to move to your side of the shipment. The practical version of that trade-off is covered in our guide to buying replacement screens in bulk.
Running the Same Audit Without Flying
A remote audit is weaker than a visit and far stronger than a video call with a salesperson. Six rules make the difference between evidence and performance.
- You choose the units. Ask for a list of finished-goods batches and pick two yourself. Never let the supplier select the “representative” one.
- One continuous take per station. Unedited video with the phone’s clock visible. Cuts are where staging happens.
- Follow a single unit. Have the camera track one module from incoming inspection to packing. Following an object exposes waiting time, re-handling and re-inspection that a station-by-station tour hides.
- Show the reject bin and the rework table. These are the two places nobody prepares in advance.
- Original documents, photographed at the desk. A re-typed PDF is a marketing document; a photo of a signed sheet with a batch number on it is evidence.
- Ask live questions while walking. Have the person holding the camera read out the batch number on the ticket in front of them. Hesitation is data.
Then compare what you saw with what you were sent. The most common finding is not fraud. It is drift: a standard written for a customer who never placed an order, an OQC report copied from a different model, an aging rack that runs only when someone remembers. None of these are visible in a quotation.
A 90-Minute On-Site Route
Time-box the visit. Long tours drift into hospitality; ninety minutes with a fixed route keeps everyone honest and gives you a comparable record across suppliers.
| Minutes | Where | What to Do |
|---|---|---|
| 0–10 | Meeting room | Ask the three closing questions up front. Note who answers and who defers. |
| 10–30 | Incoming inspection and store | Check batch separation, sampling plan and the reject shelf. Pick your carton here. |
| 30–55 | Lamination and assembly | Ask for the work order of the batch you picked, then watch one unit move through. |
| 55–70 | Aging and testing | Ask what fails most often and have them show you one failed unit. |
| 70–85 | Final inspection, packing, sample store | Open the limit sample box. Read the last three OQC reports, not the best one. |
| 85–90 | Meeting room | Ask the same three questions again. Compare the answers with the first round. |
The three questions, deliberately chosen so that a salesperson cannot answer them from memory:
- What was the reject rate on this model last quarter, and what were the top two failure modes? Numbers, not adjectives.
- If my shipment fails your outgoing inspection, what happens next, and who signs the release?
- When my model goes onto the line, who confirms the first article, and where is that record kept?
Consistent answers in both rounds mean the process is understood by more than one person. Different answers mean you have been talking to the sales department, and the plant has not been briefed.
Turning the Score Into a Decision
An audit produces a decision, not a report. Three outcomes are useful, and each has a next step attached.
- Score 0–6 — do not buy on the strength of their process control. Either source through a channel that inspects on your behalf, or plan full incoming inspection and price that labour into the landed cost.
- Score 7–13 — run a defined trial. One model, one small lot, the defect definition written into the purchase order, full inspection on arrival, and sealed samples kept by both sides. Define “pass” in the PO, not in the email thread.
- Score 14–20 — scale, with two conditions. A change-notification requirement, so that any change of panel source, adhesive or process is declared before shipment, and a re-audit interval so the file stays current.
Whatever the score, put the defect definition in the purchase order. Most disputes we see are not really about quality — they are about two parties working to different definitions of a dead pixel, and discovering it after the container has arrived.
Five Ways Audits Get Misread
1. The showroom is not the line
Display samples are selected, cleaned and often made on a different line from the one your order will run on. Ask where a specific sample was built and on which work order.
2. A certificate is not a process
A quality management certificate says a system exists somewhere in the organisation. It does not say the line holds your tolerance. Check the scope statement and the certificate holder’s name, then verify process evidence separately — the two are not substitutes.
3. “We have our own bonding line” is a claim until you see a work order
Bonding is the step most often outsourced and most often claimed. One work order number for the batch in front of you settles it in a way that a walk past a machine cannot.
4. “100% inspection” needs a takt time that supports it
If the claimed inspection rate cannot physically fit the line’s output, it is sampling with a different name. Ask how many inspectors are on the shift and the cycle time per unit; the arithmetic settles the question faster than any argument.
5. Passing an audit is a moment, not a state
Panels, adhesives, fixtures and staff all change. Without a change-notification clause and a re-audit interval, last year’s pass describes last year’s plant. This is the same reason lifecycle cost planning has to include supplier continuity, not only unit price.
How STMAX Is Audited
The same sheet applies to us, so here is what you would find on the walk.
- Space and environment. Around 20,000 square metres of production space, split into 12,000 square metres of Class 100 and 8,000 square metres of Class 1000 cleanroom, with monitoring records kept per area.
- Capacity. Monthly output in the region of 2KK displays, 1KK touch panels and 1KK bonded assemblies. We quote this as a scheduling fact, not as a quality claim — the two are separate questions.
- Inspection. Modules are tested one by one rather than by lot sampling; the ONE BY ONE TESTED mark on our packaging refers to that step. The method behind it is described in our replacement screen quality checklist.
- Feedback loop. Our functional after-sales rate runs at around 100 PPM, and confirmed failures go to failure analysis rather than straight to replacement — which is the only version of after-sales that improves the next batch.
- Lab and process. A full reliability testing laboratory runs environmental and mechanical tests before a design is released, alongside the production chain described in our production process overview.
- Certificates. ISO 9001, ISO 14001, ISO 45001 and IATF 16949 for automotive-related programmes, plus RoHS and REACH declarations. Ask for the scope and the holder name, exactly as you would with any other supplier.
- Range. TFT and AMOLED modules from 1.54″ to 32″, standard and custom, across phone replacement assemblies and industrial panels. The list is on the full product index; industrial selection criteria are set out in our industrial TFT module guide.
If you want to run the ten points on us, ask for the inspection standard and last month’s outgoing inspection report for the model you are considering, then book a live walkthrough of the cleanroom. We will show you the reject bin and the rework table, because those are the two places where an audit actually happens.


Ask Us for the Audit Pack
Send the model you are qualifying and the defect limits you work to. We will return the inspection standard for that model, a recent outgoing inspection report and a time for a live line walkthrough — including the parts of the process that are less flattering.
Request the Audit Pack →Frequently Asked Questions
1. How long should a display factory audit take?
Ninety minutes is enough for a focused audit if the route is fixed in advance: ten minutes of questions, twenty at incoming inspection, twenty-five on the bonding and assembly line, fifteen on aging and test, fifteen at final inspection and packing, five to close. A full-day audit is only worth it when you are qualifying a new process rather than a new supplier.
2. Can a factory audit be done remotely?
Yes, with conditions. You choose the batches, you require continuous unedited video with a visible clock, the camera follows one unit through the line, and you see the reject bin and rework table. That covers most of what a short visit covers; what it cannot replace is walking the floor unannounced.
3. What is the single most useful document to request?
The model-specific inspection standard. It tells you whether pass and fail were defined before production, and whether the definition matches the one you will use on arrival. A generic quality policy is not a substitute, however well formatted it is.
4. Should I audit before or after the sample order?
Before the first bulk order, and ideally before the sample order. Samples are often built with extra attention or on a pilot line, so a good sample is weak evidence about mass production. Use the sample to confirm specification and the audit to judge repeatability.
5. What if a supplier refuses a factory visit?
A refusal is information, not necessarily a disqualification. Some legitimate suppliers protect other customers’ lines and will offer a video walkthrough or a third-party inspection instead. Accept that offer, run the remote route, and treat a flat refusal with no alternative as a reason to look elsewhere.
Send Your Model List and Defect Limits
Tell us which models you are qualifying and how you judge a dead pixel, a colour shift or a touch failure. You will get a line-item answer from the people who run the line — and a straight no on anything we cannot meet.
Talk to the Engineering Team →