The reliable approach is a face-coded system that connects the approved tile manifest to artwork, production batches, storage bins, kitting positions, set-level verification, packing, inspection, and replacement records. Every finished retail set should pass the factory's defined complete-set control before it is sealed.
This guide explains how buyers and manufacturers can build that control for 152, 160, 166, or custom configurations.
Understand the Four Different Completeness Failures
Treat these as separate defect types:
- Missing tile: The set contains fewer physical tiles than specified or lacks one required face.
- Duplicate substitution: The total is correct, but an extra copy of one face replaces another.
- Wrong face: A tile carries artwork that does not belong to the approved configuration or SKU.
- Wrong variant: The face is conceptually correct but uses the wrong artwork revision, color, language, back, orientation, or process.
A simple count may detect the first failure. It will usually not detect the other three.
1. Create a Production-Ready Tile Manifest
The source of truth is the approved American Mahjong tile manifest. It should define:
- One stable code for every unique face
- Face name and category
- Approved artwork file or artboard
- Quantity per finished set
- Face process and color reference where needed
- Back specification
- Orientation
- Total tiles per set
- Revision, approval date, and applicable SKU
Avoid an instruction such as “standard 160-tile set.” Different products can use the same total with different jokers, blanks, spares, flowers, brand tiles, or other approved faces.
The manifest total should be generated by formula. Category subtotals help reviewers identify an error that a correct grand total can hide.
2. Use the Same Face Code Everywhere
The face code should appear on:
- Artwork file or artboard
- Digital proof
- Production program or setup sheet
- First-article record
- Work-in-process container
- Storage-bin label
- Kitting instruction
- Inspection checklist
- Replacement inventory record
If design files use JKR-A, production calls it Joker 1, and packing calls it red special, people must translate between systems. Every translation creates an opportunity for error.
Codes should remain stable across reorders. A revised visual should receive a revision identifier without silently reusing an old file under the same uncontrolled name.
3. Verify Every Unique Face at Production Setup
Before running quantity, compare one first-off piece from every unique face to the approved artwork index.
The check should include:
- Correct face code
- Correct artwork
- Correct scale and position
- Correct color assignment
- Correct engraving, fill, print, or approved process
- Correct material and tile construction
- Correct face-to-back orientation
- Correct revision
Arrange the first-off pieces in manifest order and photograph the complete board. This creates a visual map for production and inspection.
Do not approve only the most attractive or technically difficult faces. A simple symbol can still be mapped to the wrong production file.
4. Control Work-in-Process by Face
After production, keep different face codes physically separated and clearly identified. A good container label can include:
- Product SKU
- Face code
- Artwork revision
- Quantity in container
- Production date or batch
- Machine or line reference where used
- Inspection status
- Operator or checker
Avoid unlabeled open trays that rely on appearance. Similar suit faces, jokers, blanks, and spares are easy to confuse, especially after a tray is moved.
Use status controls such as awaiting inspection, approved for kitting, hold, rework, and scrap. Nonconforming pieces should never return to an approved bin without documented re-verification.
5. Reconcile Production Quantities Before Kitting
Calculate the required quantity for each face:
Required good quantity by face =
quantity per set x planned finished sets
+ separately approved replacement or process allowance
Keep process allowance outside the consumer-packed quantity. The system should distinguish:
- Required good pieces for retail sets
- Approved replacement inventory
- Setup and inspection pieces
- Rework quantity
- Scrap
- Remaining controlled stock
If one face has an unexplained shortage or surplus, investigate before packing. A surplus of one code can signal that another face was produced or labeled incorrectly.
6. Design the Kitting Station Around the Manifest
The packing work should guide the operator through a fixed sequence. Possible controls include:
- Bins arranged in manifest order
- One bin per face code
- Fixed tray positions
- Pick quantities displayed at each position
- Color-coded category dividers used only as a secondary cue
- Kitting board showing one approved face of each type
- Controlled replenishment from labeled reserve stock
- No unrelated SKU at the same station
The best layout depends on factory volume, tile configuration, packaging, and available automation. The principle is consistent: make the correct action visible and make a skipped or repeated action easy to detect.
Do not place nearly identical faces in unmarked adjacent bins without a verification step.
7. Use Fixed Pack Positions Where Practical
A retail tray, temporary kitting tray, row template, or count fixture can convert completeness from an abstract total into visible filled positions.
For example, the work instruction can define:
- Row or compartment number
- Face-code sequence
- Quantity in each position
- Orientation
- Accessory location
- Final tile total
An empty space reveals a missing piece. A wrong face can still occupy the space, so position control must be combined with face identification.
If the final retail package does not have fixed tile positions, a temporary production fixture can be used before transferring the verified set into the bag or case.
8. Add a Second, Independent Verification
The person who kits a set can unconsciously repeat the same error during self-check. Use an independent verification method suited to the process.
Options include:
- Second-person face-by-face comparison
- Fixed-sequence checklist with sign-off
- Image-assisted comparison
- Machine-vision or scanning system where face features support it
- Controlled tray comparison against an approved layout
- Electronic station prompts and completion records
Automation is not automatically reliable. It must be validated for clear, translucent, reflective, engraved, printed, and visually similar faces under actual production lighting and orientation.
Record who or what verified the set, when it occurred, and which manifest revision was used.
9. Treat Weight as a Screening Tool, Not Proof
Set weight can detect certain large shortages, extra components, or packaging errors. Establishing a weight range may help flag abnormal packs for recheck.
It cannot reliably detect:
- One face substituted for another of similar weight
- Wrong artwork on the correct tile body
- Wrong color or revision
- Missing lightweight insert replaced by extra packing material
- Wrong accessories with similar mass
Use weight after face-level kitting control as an additional signal. Do not let a passing weight release a set that failed composition verification.
10. Control Rework and Line Interruptions
Many mix-ups occur when the normal sequence is interrupted. Define what happens when:
- A tile is found scratched or incorrectly printed during kitting
- A bin becomes empty
- A set is opened after verification
- Packaging is damaged and replaced
- A face batch is held for investigation
- The line stops between shifts
- A customer-specific rush quantity is inserted
Use an isolated rework area and require complete re-verification after any sealed or verified set is reopened. Replacing one tile should not preserve the old verification status automatically.
At shift changes, reconcile open sets, bin quantities, hold material, and completed packs. Do not leave partially kitted sets unmarked at the station.
11. Separate SKUs, Colorways, and Revisions
Two sets may use identical faces but different backs, material colors, packaging, or joker artwork. Producing or packing them at the same time increases mix risk.
Use physical and system separation for:
- Different total configurations
- Retailer-specific editions
- Colorways
- Language versions
- Old and new logos
- Replacement batches
- Packaging revisions
Where possible, complete line clearance before introducing the next SKU. Account for leftover tiles, labels, inserts, cases, and carton marks.
12. Seal and Identify Verified Sets
After successful verification, apply the approved closure or status mark. The goal is to make it clear if a set has been reopened.
Depending on the packaging, controls may include:
- Tamper-evident production seal
- Verified-status label
- Controlled bag closure
- Lot code
- Packer/checker code
- Date or batch mark
The mark should not create an unsupported consumer claim. It is a traceability and process-status tool.
13. Audit the Process With Complete-Set Checks
Supervisors and third-party inspectors should open randomly selected finished sets and compare every face to the approved manifest.
The audit sample should span:
- Different master cartons
- Different packing times or shifts
- Different stations or lines
- Different material or face-production batches where relevant
If a duplicate or missing tile appears, contain the lot and investigate the kitting process. Do not simply correct the inspected set.
Acceptance sampling can support the shipment decision, but it does not prove every unopened set is complete. Review AQL and third-party inspection for custom Mahjong sets alongside the factory's complete-set process.
14. Reconcile Packed Output With Remaining Stock
At the end of packing, compare:
Good pieces produced by face
- pieces packed in finished sets
- approved replacement stock
- retained samples
- documented scrap or rework
= controlled remaining quantity
Unexpected face-specific balances reveal hidden errors. If one tile code has many extra pieces and another is short, some sets may contain substitutions even if finished-set totals look correct.
Reconciliation is stronger when quantities are recorded during each transfer rather than reconstructed after the line is cleared.
15. Keep a Retention Set From the Actual Lot
Select and identify one or more finished sets from bulk production according to the project's retention plan. Preserve:
- Complete tile composition
- Accessories
- Packaging
- Lot and carton traceability
- Manifest and artwork revisions
- Inspection status
The retention set helps investigate customer reports, compare reorders, and verify replacement tiles. It should come from the actual lot, not from a separately prepared showroom sample.
A Copyable Completeness Control Plan
| Stage | Risk | Required control | Evidence | Reaction to failure |
|---|---|---|---|---|
| Artwork release | Wrong face/revision | Face-code artwork index | Approved proof log | Stop release and correct revision |
| First-off production | Wrong mapping/orientation | Check every unique face | First-article board/photos | Hold affected face production |
| Face storage | Mixed tiles | Labeled, separated containers | Bin and status records | Quarantine and recount |
| Pre-kitting | Quantity imbalance | Face-level reconciliation | Inventory worksheet | Investigate shortage/surplus |
| Kitting | Missing/duplicate face | Fixed sequence/positions | Kitting record | Rebuild and re-verify set |
| Independent check | Repeated operator error | Second method/person | Verification sign-off | Hold since last good check |
| Pack closure | Reopened verified set | Controlled seal/status | Lot/packer code | Complete re-verification |
| Lot audit | Process failure | Random complete-set checks | Inspection report | Contain lot and corrective action |
| End reconciliation | Hidden substitution | Face-level balance | Final reconciliation | Investigate affected sets |
Assign owners and frequencies to every row. “Operator checks” is incomplete without a method and record.
Buyer Questions for a Mahjong Factory
Ask the supplier:
- Does every unique face have a stable production code?
- How is each code linked to approved artwork?
- Are different faces and revisions physically separated?
- How are quantities reconciled before packing?
- Does the kitting station use fixed positions or sequence?
- Is every finished set composition-verified before sealing?
- Is the verification independent from the original pick?
- How are reopened or reworked sets controlled?
- How are different SKUs cleared from the line?
- What records identify packer, checker, batch, and revision?
- How are random complete-set audits performed?
- How are replacement tiles and retained samples separated from retail inventory?
The factory should explain the actual workflow, not only state that workers are experienced.
Common Mistakes to Avoid
- Defining only the total tile count
- Using file names that do not match production bin codes
- Verifying total weight instead of face composition
- Mixing old and new artwork revisions
- Kitting several colorways at one uncontrolled station
- Returning reworked tiles to approved bins without verification
- Reopening verified sets without repeating the check
- Correcting only the set found by an inspector
- Keeping no face-level production reconciliation
- Preparing the retention sample separately from bulk output
- Assuming a third-party sampled inspection replaces full factory kitting control
Frequently Asked Questions
Can weighing every set prevent missing tiles?
Weight can identify some shortages or extras, but a duplicated tile can replace a missing tile with almost no weight change. Use face-by-face composition control.
Should every set be checked against the tile manifest?
The factory should have a defined complete-set process that verifies the approved composition before sealing. The exact method can be manual, fixture-based, image-assisted, or automated, but it must be validated and controlled.
Why are wrong tiles common after rework?
Rework interrupts the normal sequence. A replacement can come from the wrong bin or revision, and the set may retain an old verification status. Isolate rework and require complete re-verification.
Can third-party inspection guarantee every set is complete?
No. Inspectors normally evaluate a sample of finished sets. Their findings can reveal a process problem, but uninspected units still depend on the factory's production controls.
How should replacement tiles be identified?
Use the same face code, product revision, material/color batch, and SKU records used in production. This makes customer reports and replacement matching more reliable.
Review a Tile Manifest and Kitting Plan
Send Lucky Mahjong your intended tile configuration, artwork index, packaging layout, and replacement strategy. We can review the manifest structure and align it with sample, production, kitting, and inspection documents.
Discuss a Custom Mahjong Project or email [email protected] with the subject line Tile Completeness Control.
