Quick Answer
A sample that performs well is evidence. It is not an approval, and it does not transfer to the next production batch by itself.
Three things decide whether a trial becomes a usable approval:
| What people usually do | What actually makes the result usable |
|---|---|
| Apply the sample and judge by feel | Write the substrate, the method and the acceptance criteria before the tube is opened |
| Approve the product after a good result | Approve a product + substrate + method combination, and record which batch was tested |
| Treat the supplier’s sample as representative | Confirm the sample came from the same formulation and production line as the order |
The gap between the left and right column is where most avoidable disputes live. A buyer who ran an informal trial has a memory; a buyer who ran a defined trial has a file.
1. Decide What the Trial Is For Before Ordering Samples
The same tube can be used for four different questions, and they need different procedures:
- Feasibility — is this chemistry even a candidate for the substrate and the movement?
- Application — can the crew actually gun, tool and finish it at the specified rate?
- Adhesion — does it bond to the real surface, including the coating?
- Comparison — which of two or three candidates behaves better under identical conditions?
Mixing these into one informal trial is the most common reason a sample result cannot be used later. A crew that liked the way a product tooled has not established adhesion; a peel test on a clean lab panel has not established that the same product works on a chalky powder coat.
Write the purpose in one sentence before requesting anything. If the sentence contains “and also”, split it into two trials.
2. Make the Substrate the Subject, Not the Product
The single most useful correction in sample evaluation is to stop treating the product as the variable.
A product that passes on anodised aluminium may fail on mill-finish aluminium, on a coated sheet, or on a substrate that was cleaned with the wrong solvent. The surface is part of the system, and it has to be described precisely enough that someone else could reproduce it.
Describe each substrate with four attributes:
| Attribute | Why it changes the result |
|---|---|
| Base material | Different chemistry families bond differently to metal, glass, stone and plastic |
| Surface finish | Anodised, mill, powder-coated and PVDF behave nothing alike |
| Condition | New, weathered, chalky, contaminated, previously sealed |
| Preparation | Cleaner, primer, abrasion, and how long before application |
If the buyer cannot fill those four lines, the trial will produce a result that cannot be repeated — which is the same as producing no result.
Adhesion failures are frequently a substrate story rather than a product story. The checks that separate the two are set out in what to check across glass, aluminium and stone.
3. Fix the Method Before the Sample Arrives
A trial method only needs to be as formal as the decision it supports, but it must be written down first.
For most construction adhesive and sealant evaluations, five decisions cover it:
- Specimen — what is bonded to what, at what dimensions, with which bead or bond geometry.
- Preparation — cleaning, abrasion, primer, and the interval between preparation and application.
- Conditions — ambient temperature, surface temperature, humidity, and whether the test is done indoors or on site.
- Cure — how long, and at what conditions, before the joint is assessed.
- Assessment — what is looked at, and what counts as pass or fail.
The fifth is the one that gets skipped. “Looks fine” is not an assessment. A usable assessment names the observation: cohesive failure within the bead, adhesive failure at the substrate, no visible change, or a measured value against a stated limit.
Where a quantitative property matters, the test method comes from the product documentation or the project specification — not from improvisation. If the TDS does not name the standard, ask which one applies before designing the trial.
4. Record the Batch, Not Just the Product Name
Complaints that cannot be resolved usually share one root cause: nobody wrote down which batch was approved.
Samples and production often come from different lots. A sample approval that references only a model number leaves the buyer unable to answer the first question any serious investigation will ask — was the material on site the material that was approved?
Record, at minimum:
- Product model and the supplier’s sample reference.
- Batch or lot number of the sampled material, where the supplier provides one.
- Date sampled, and date applied.
- Who applied it.
- Where the retained sample is kept.
Keep a retained sample. It costs nothing at the time and it is the only thing that can settle a dispute six months later, when the same question arrives as a complaint. How that evidence should be collected once a problem is reported is covered in handling a quality complaint with traceable evidence.
5. Define Pass and Fail Before You Look
Deciding the criteria after seeing the result is not evaluation; it is justification with extra steps.
Acceptance criteria should come from three places, in this order:
- The project specification, if the material is for a named project.
- The product documentation, for properties the manufacturer states and tests.
- The buyer’s own operational requirement, for things like tooling time or finish appearance that matter to the crew but are not in any standard.
The third category is legitimate but must be labelled as such. “Our installers need to tool a 20 mm joint before the skin forms” is a real requirement; it is not a performance standard, and it should not be written as one.
A failure should trigger investigation rather than rejection of an entire product family. Check whether the correct product was supplied, whether the method was followed, and whether the substrate was what it was thought to be. A trial that is changed mid-way without a record stops being comparable to anything.
6. Separate the Three Decisions
Sample evaluation produces three distinct outcomes, and they are often conflated:
| Decision | Question answered | Who owns it |
|---|---|---|
| Technically suitable | Does it work on this substrate and joint? | Technical / engineering |
| Commercially acceptable | Are price, MOQ, lead time and documents workable? | Procurement |
| Approved for this use | May we use it here, on this project? | Project authority, where one exists |
A distributor can make the first two decisions. The third frequently belongs to someone else — a consultant, a client, or a project engineer — and no amount of internal sample testing substitutes for that approval. Where a substitute product is being proposed for a specified one, the formal route is covered in how a substitution gets approved on a project.
Writing these three lines into the sample record prevents the most expensive misunderstanding in the whole process: assuming that a good trial equals permission to install.
7. A Record That Survives the Project
The record does not need software. A single page per evaluation is enough, provided it answers these nine questions:
| Field | Purpose |
|---|---|
| Product and supplier reference | Identifies what was tested |
| Batch or lot | Connects the result to physical material |
| Intended application | Prevents later use of the result elsewhere |
| Substrates and preparation | Makes the trial reproducible |
| Method and conditions | Distinguishes a field trial from a test |
| Observations | Records what was actually seen |
| Acceptance criteria and source | Shows the basis of the decision |
| Decision | Approved, conditional, rejected, or more testing |
| Retained sample location | Enables later verification |
If a distributor runs this process repeatedly, the record becomes a qualification file. The second customer asking about the same application does not need a second trial — only a check that the substrate and exposure still match.
8. What to Send the Supplier
A sample request that produces a useful reply contains the application, not a product preference.
Send the substrate pair with finishes, the joint or bond dimensions, the exposure, the target market, and the property that matters most with the method you intend to assess it by. Ask explicitly which document states that property, and whether the sampled material came from the same formulation that will be supplied in production.
A supplier who cannot answer the last question is telling you something useful about how change control works on their side. How to verify that the documentation actually matches the product is a separate exercise, covered in verifying TDS, SDS and test reports.
Conclusion
A sample is a single data point. What makes it usable is everything written around it: the substrate description, the method, the criteria, the batch, and the decision.
Buyers who fix those five things before opening the tube get a result they can defend. Buyers who do not get an opinion, and opinions do not survive the first dispute.
