Run your building block connection fit inspection against a sealed physical golden sample, not a photo and not a laboratory tolerance you cannot source. Record what you actually observe — colour, fit, flash, function — label every entry as a non-laboratory observation, and defer any AQL number to the agreed inspection plan written on the PO.
That single discipline decides whether your second lot matches your first. Buyers who approve only a photo pay for it on the second lot; buyers who freeze a sealed sample, a tool number and a revision card have something to argue with when a container arrives loose, tight, or off-colour.
This page gives you the repeatable log: the fields, the checkpoints, the questions to put to a supplier, and the limits you must state honestly. It applies to ABS and plastic construction sets — brick sets, marble-run and track sets, stack-and-build kits — where connection feel is the whole product.
What buyers should lock down before the first lot ships
- A sealed, dated golden sample stored at a named location is the reference. Emails do not settle a colour or fit dispute; the physical piece on the QC shelf does.
- Log the tool number and revision card, and refuse a trade-fair booth photo as the production standard. Fair lighting lies; the revision card does not.
- Every connection-fit entry is a non-laboratory observation. If no supplier document or PO states a tolerance, write "non-lab observation" and stop there — do not convert a feel into a number.
- Major and minor defect limits must be stated per lot or inspection plan; a factory's default percentages are not automatically your acceptance criteria.
- Package dimensions and pieces per inner or carton should come from measured production packaging, not only a preliminary artwork file, and carton CBM must be recalculated when carton size or pack quantity changes.
Why does connection fit need its own log instead of a line in the QC report?
Because connection fit is the one characteristic a general QC report tends to describe in adjectives. Fit inspection should check that parts assemble without excessive force, looseness, interference, sharp exposure or unintended release — that is a set of observable conditions, not a measurement, and it deserves its own columns.
A toy quality plan should map each critical characteristic to an inspection or test method, frequency, acceptance limit and reaction plan. Connection fit belongs in that map as a characteristic with a named method (compare to golden sample), a frequency (per lot, per sampled unit), an acceptance statement (assembles and holds without excessive force or unintended release) and a reaction plan (isolate, record defect class, rework, release).
The trap is escalation. A buyer sees a loose stud, asks the factory for "the tolerance", receives a number in an email, and pastes it into a spec sheet as if it were a laboratory limit. It is not. It is a supplier statement, and it belongs in the supplier-statement column, not the measured-observation column.
For construction sets specifically, piece-size, connection-force, edge and age-grade controls all matter, and a large box does not make every individual piece safe for every age. Keep the age-grade question separate from the fit question so neither one hides the other.
How to build the repeatable sample log, step by step
Anchor the log to a physical sealed sample. Record the golden sample ID, the seal date, and the storage location. If the sample lives in a drawer nobody can name, you do not have a reference — you have a rumour.
Freeze the revision at the source. Write the tool number and the revision card reference into the header of every log sheet, and keep a photo of the card rather than a photo of the booth. When a mould is re-cut or a texture is changed, the revision changes and the log header must change with it.
Record fit as an observation against the sample. For each sampled unit, note whether parts assemble with normal hand force, whether the joint holds when the assembly is lifted and gently shaken, whether there is visible flash at the mating faces, and whether any part releases when it should not. Mark the entry "non-lab observation" and leave the tolerance field blank rather than guessing.
Separate defect handling from sampling numbers. When a lot fails, isolate it, record the defect class, and release only after rework. If you want an AQL figure, take it from the agreed inspection plan on the PO — not from a catalogue sentence and not from this page.
Tie the sample to packing and parts records. Reference the pieces-per-carton figure and the packing list, and confirm that package dimensions came from measured production packaging. If the carton size or pack quantity changes, recalculate CBM rather than carrying the old figure forward.
Log tooling terms as a document. Fee, ownership, storage period, and what happens if the tool sits idle should be written down. A buyer once asked who owned the tool after paying a fee; the contract had not frozen storage years or release conditions. That is a paperwork failure, not a moulding failure.
Date and attribute any supplier visit. A factory walk-through that covers the raw-material warehouse to the finished-goods warehouse, with random worker interviews on hours, wage payment and overtime, produces a rating and a follow-up date — record both, and record who did the walk.
Flag supplier-stated reach as unverified. If a supplier says it serves clients on almost all continents, that is a supplier statement. Write it in that column.
Capture the commercial rationale without inventing market figures. If a low-cost line becomes the reorder driver because it fits small shelf space and moves fast, note the rationale in plain words. No percentages, no market size.
Close the loop with a three-column sheet: measured observation, supplier statement, recommendation. Every row lands in exactly one column. That separation is what makes the log defensible when a dispute arrives eight weeks later.
Sample log checkpoints and the red flags next to them
| Golden sample | Check: sealed, dated, stored at a named location, ID written in the log header. Red flag: reference is a photo, a chat image, or "the one we sent last time". |
|---|---|
| Revision control | Check: tool number and revision card referenced on every sheet. Red flag: trade-fair booth photo offered as the production standard. |
| Connection fit entry | Check: assemble force, joint hold, flash at mating faces, unintended release, each marked non-lab observation. Red flag: a tolerance number appears with no document behind it. |
| Defect handling | Check: failed lot isolated, defect class recorded, release only after rework. Red flag: a default defect percentage quoted as your acceptance criteria. |
| Sampling plan | Check: major and minor limits stated per lot or inspection plan on the PO. Red flag: the number comes from a catalogue, a blog, or a verbal promise. |
| Packing and parts | Check: pieces per carton and packing list referenced; package dimensions from measured production packaging. Red flag: dimensions taken only from a preliminary artwork file. |
| Carton CBM | Check: derived from measured outer-carton dimensions and recalculated when carton size or pack quantity changes. Red flag: CBM carried over from a previous revision. |
| Tooling terms | Check: fee, ownership, storage period and idle-tool outcome written into the contract. Red flag: ownership settled by handshake after the fee is paid. |
| Supplier visit | Check: visit date, scope, rating, follow-up date, and who performed the walk. Red flag: an undated "we visited the factory" note. |
| Supplier-stated reach | Check: recorded in the supplier-statement column and marked unverified. Red flag: a reach claim pasted into the measured-observation column. |
Worked example (illustrative, not a real shipment)
This scenario is invented to show where a log earns its keep. It is not a real order, a real factory, or a real inspection result.
A buyer places one lot of roughly 1,200 units across two construction-set SKUs — a track-and-marble style set and a stack-and-build kit — split between two destinations, one requiring retail-ready inner packs and the other bulk cartons. The supplier's packing list gives pieces per carton for the bulk destination; the retail-ready destination uses a different inner pack, so the carton count and CBM differ between the two.
The log header carries the golden sample ID, seal date, storage location, tool number and revision card. The buyer samples units at the start, middle and end of the run and records, per unit: assemble force (normal hand force or not), joint hold after a gentle shake, visible flash at mating faces, and any unintended release.
Two units show a joint that releases under a gentle shake. The buyer does not ask for a tolerance. The entry is written as a non-lab observation against the golden sample, the defect class is recorded, and the affected units are isolated. The lot is released only after rework, and the release note references the same golden sample ID.
The step that would have held the goods: the packing check. Because the retail-ready destination uses a different inner pack, the pieces-per-carton figure and the carton CBM in the log had to be recalculated against measured production packaging. Had the buyer carried the bulk-destination CBM into the retail-ready shipment, the container plan would have been wrong before the fit question was ever raised.
A log with a measured-observation column, a supplier-statement column and a recommendation column tells you which problem you actually have — and stops a loose joint from being answered with an invented number.
What should you ask suppliers before approving the sample log?
Ask for documents, not adjectives. The questions below are the RFQ checklist for this decision; each answer lands in the supplier-statement column until you can verify it against a physical piece or a signed PO.
Can you provide the sealed golden sample ID, seal date and current storage location for this SKU?
What is the tool number and revision on the revision card, and can you send a photo of the card rather than a trade-fair booth photo?
When you compare bulk to the golden sample, which attributes do you check — colour, fit, flash, function — and how is each one recorded?
What is your defect-class list, and what is your isolate, rework and release procedure for a failed lot?
What AQL and inspection plan are written on the PO for this order, and can you send the inspection plan rather than a default percentage?
Which package dimensions and pieces-per-carton figures come from measured production packaging, and when were they last recalculated?
If the carton size or pack quantity changes, who recalculates the CBM and where is the revised figure recorded?
What are the tooling fee, ownership, storage period and idle-tool terms, and are they written into the contract?
Which parts of this set are subject to piece-size, connection-force, edge and age-grade controls, and how are those controls documented separately from the fit log?
If we visit, what will we be shown, who will we speak to, and will the visit produce a dated rating with a follow-up date?
Where does this log sit against compliance paperwork?
It sits beside it, not inside it. A sample log records your observations against a reference; it does not certify a product and it does not replace a test report or a declaration of conformity.
Keep the two visibly separate. If a supplier sends a compliance document, file it under compliance and note the document reference in the log — do not paraphrase its contents into a fit entry. Where a regulatory timeline is involved, state only what the official text supports: whether a date is the entry into force, the date the rule actually applies to this product category, or a date the official text does not confirm. A year in a secondary article headline is not an application deadline.
One short box is enough: compliance documents belong in a compliance file with their own references, and the fit log should point to that file rather than absorb it.
Which construction sets make this log worth the effort?
Any set where the joint is the play experience. Track and marble-run sets, stack-and-build kits, and multi-model vehicle sets all live or die on whether pieces hold and release predictably.
If you are building the assortment, browse the brick building sets range at /products/brick-sets/ for stud-and-tube construction, and the large building blocks range at /products/large-building-blocks/ where piece size and age grade shift the fit question toward edge and size controls rather than connection force.
The log template does not change between those two families. What changes is which characteristic you sample most heavily: connection force and unintended release for small-piece construction, piece size and edge condition for large-piece sets.
FAQ
Can I write a numeric tolerance for connection fit if my supplier gives me one?
Only if the number is tied to a document you can point to, such as an agreed inspection plan or a supplier specification sheet. Otherwise record it as a supplier statement and keep the measured-observation column free of numbers. A verbal figure in a chat message is not a tolerance.
What AQL should I put on the PO for a building block order?
This page cannot give you a number, and neither should a catalogue. Major and minor defect limits must be stated per lot or inspection plan, and a factory's default percentages are not automatically your acceptance criteria. Set the limits with your inspection provider and write them into the PO.
How many units should I sample for a fit check?
The count comes from the agreed inspection plan, not from a rule of thumb. What matters for the log is coverage across the run — start, middle and end — so a drift in the mould or the material shows up as a pattern rather than a single unlucky unit.
Is a trade-fair sample good enough as a production reference?
No. A booth sample is often a hero unit, and fair lighting flatters colour. Label the tool number and revision on the table, keep the revision card, and treat the sealed golden sample as the production standard.
What if the carton size changes after I approve the sample?
Recalculate. Carton CBM is derived from measured outer-carton dimensions and should be recalculated when carton size or pack quantity changes. Package dimensions and quantity per inner or carton should be based on measured production packaging, not only a preliminary artwork file.
Does this log replace a compliance document or test report?
No. The log records your own observations against a reference and states its non-laboratory limits. Compliance documents belong in a separate file with their own references, and the log should point to that file rather than restate its contents.
Sources
Related Products & Categories
Put the log to work on your next construction-set order
Take the ten checkpoints above into your next supplier conversation. Ask for the golden sample ID, the revision card, the inspection plan on the PO, and the measured packing figures — then write down which column each answer belongs in.
If you are assembling the range at the same time, start with the brick building sets and large building blocks ranges and apply the same log header to every SKU before the first lot ships.