Multi-Style, Multi-Color, Multi-Size Clothing Orders from 1688: Cluster Sampling QC for South African Importers | $169
When South African importers order T-shirts in 6 colors and 5 sizes from a single 1688 factory, the risk of quality drift across styles is high — each color has a separate dye lot, each size has its own pattern grading, and any mismatch means inconsistent products on your shelves.
When South African importers order T-shirts in 6 colors and 5 sizes from a single 1688 factory, the risk of quality drift across styles is high — each color has a separate dye lot, each size has its own pattern grading, and any mismatch means inconsistent products on your shelves. Standard AQL sampling across the full batch won't catch color-specific defects. You need cluster sampling — one sample set per style-color-size combination.
The Multi-Variation Challenge on 1688
1688 suppliers encourage single-SKU orders but most SA importers need variety — 3 T-shirt styles in 4 colors each and 5 sizes = 60 unique SKUs from one order. Here's what can go wrong when colors and sizes mix:
| Defect Type | Impact | Frequency |
|---|---|---|
| Cross-color dye lot mismatch | One red T-shirt is bright, another is burgundy | High |
| Size grading error | "Large" fits like "Medium" on one color | Medium |
| Fabric substitution | Blue batch uses thinner cotton than black batch | High |
| Print flaw isolated to one color | Screen print peeling only on white batch | Medium |
| Label sewn in wrong language | One batch has Chinese labels, others have English | Low but critical |
How Cluster Sampling Works for Multi-Variation Orders
Standard AQL sampling draws randomly from the entire batch. But when you have 60 SKUs, a random sample across the warehouse may miss the one defective color entirely. Cluster sampling solves this:
Step 1: Group by Variation
Divide the batch by color first (all red together, all blue together, etc.). Then within each color, divide by size. Each color-size combination becomes a "cluster."
Step 2: Sample Per Cluster
For each cluster, apply AQL 2.5 to determine sample size. A cluster of 100 red size-L T-shirts requires 20 samples. A cluster of 80 blue size-M T-shirts requires 13 samples. This ensures every color and every size is tested, not just the batch average.
Step 3: Measure Each Size Grade
Lay out garments from each size cluster and measure: chest width (across chest, 2.5cm below armhole), body length (HPS to hem), sleeve length (shoulder to cuff), and sweep (bottom hem width). Compare against the spec sheet. Common 1688 sizing drift: the "Large" that was 112cm chest at sample stage becomes 108cm in production.
Step 4: Color Fastness Per Dye Lot
AQ 2.0 rating minimum for dry rubbing and AQ 3.0 for wet rubbing. Test each color cluster separately. Factory dyes may be consistent for black but bleed on red. CloudSpects runs a full crockmeter test on each color cluster.
Identifying Fabric Substitution Across Variations
A common 1688 cost-saving trick: the supplier uses better fabric on the first production run (sample approval) and switches to thinner/lower-GSM fabric on later runs. This shows up as measurable differences between colors ordered at different times.
Inspector checks: GSM scale measurement of 5 samples per color cluster. Burn test for fiber composition (cotton vs polyester blend verification). If the black T-shirt weighs 180 GSM and the white weighs 150 GSM, you have a substitution issue — even if the AQL sample pass rate is 98%.
SA-Specific Considerations for Multi-Variation Orders
Care Label Compliance Per Style
Every style in your multi-variation order must have the correct care label sewn in English — fiber content, country of origin, washing instructions per SANS 556. Inspectors check every cluster's care label. A batch of 60 SKUs may have 3-4 styles with different care instructions (e.g., pure cotton T-shirt vs polyester-cotton blend hoodie — different wash temperatures).
Sizing Conversion Across Clusters
Asian sizing differs from South African sizing. Your spec sheet should specify SA-equivalent measurements. Inspectors check each size cluster against the agreed measurement table. Common mismatch: a "Large" made for Asian market is labeled "Large" for SA but is actually a South African "Small."
Packaging Consistency
Each color and size cluster should have consistent poly bag size, bag thickness (min 0.025mm with suffocation warning), and individual vs bulk packing. If your supplier packs 50 red T-shirts in one bag and 50 blue in individual poly bags, that's a packing inconsistency that affects your JHB warehouse receiving process.
CloudSpects Multi-Variation Inspection Workflow
CloudSpects handles multi-SKU orders from 1688 every week. Our cluster-inspection workflow covers:
- Pre-inspection: you send the full SKU matrix (style × color × size), we build a sampling plan per cluster
- On-site: one inspector handles measurement verification per cluster, a second runs color fastness and GSM testing
- Reporting: defect matrix organized by cluster — you see exactly which color or size has the problem
- Photo evidence: 50+ photos per inspection covering each cluster's critical areas
Frequently Asked Questions
Does cluster sampling cost more than standard inspection?
Most multi-variation orders require 1.5-2 inspector-days at the factory due to cluster separation. From $169/man-day, a typical 60-SKU T-shirt order costs $250-340.
I only have 3 colors but 10 sizes per color — do I still need cluster sampling?
Yes — if any color or size has a quality defect that's unique to it, standard random sampling may miss it. Cluster sampling guarantees every color-size intersection is tested.
What happens if one cluster fails AQL but others pass?
You can reject only the failed cluster. The factory reworks that color or size while the rest ships. CloudSpects inspects the reworked batch before release.
Don't Let 60 SKUs Become 60 Headaches
Multi-variation 1688 clothing orders can grow in complexity faster than your profit margin. Let CloudSpects design a cluster-sampling plan for your next order and catch the color-specific defects before they reach Cape Town. Same-day quote from $169/man-day.
Frequently asked questions
Step 1: Group by Variation Divide the batch by color first (all red together, all blue together, etc.). Then within each color, divide by size. Each color-size combination becomes a "cluster." Step 2: Sample Per Cluster For each cluster, apply AQL 2.5 to determine sample size. A cluster of 100 red size-L T-shirts requires 20 samples. A cluster of 80 blue size-M T-shirts requires 13 samples. This ensures every color and every size is tested, not just the batch average. Step 3: Measure Each Size Grade Lay out garments from each size cluster and measure: chest width (across chest, 2.5cm below armhole), body length (HPS to hem), sleeve length (shoulder to cuff), and sweep (bottom hem width). Compare against the spec sheet. Common 1688 sizing drift: the "Large" that was 112cm chest at sample stage becomes 108cm in production. Step 4: Color Fastness Per Dye Lot AQ 2.0 rating minimum for dry rubbing and AQ 3.0 for wet rubbing. Test each color cluster separately. Factory dyes may be consistent for black but bleed on red. CloudSpects runs a full crockmeter test on each color cluster. Identifying Fabric Substitution Across Variations A common 1688 cost-saving trick: the supplier uses better fabric on the first production run (sample approval) and switches to thinner/lower-GSM fabric on later runs. This shows up as measurable differences between colors ordered at different times. Inspector checks: GSM scale measurement of 5 samples per color cluster. Burn test for fiber composition (cotton vs polyester blend verification). If the black T-shirt weighs 180 GSM and the white weighs 150 GSM, you have a substitution issue — even if the AQL sample pass rate is 98%. SA-Specific Considerations for Multi-Variation Orders Care Label Compliance Per Style Every style in your multi-variation order must have the correct care label sewn in English — fiber content, country of origin, washing instructions per SANS 556. Inspectors check every cluster's care label. A batch of 60 SKUs may have 3-4 styles with different care instructions (e.g., pure cotton T-shirt vs polyester-cotton blend hoodie — different wash temperatures). Sizing Conversion Across Clusters Asian sizing differs from South African sizing. Your spec sheet should specify SA-equivalent measurements. Inspectors check each size cluster against the agreed measurement table. Common mismatch: a "Large" made for Asian market is labeled "Large" for SA but is actually a South African "Small." Packaging Consistency Each color and size cluster should have consistent poly bag size, bag thickness (min 0.025mm with suffocation warning), and individual vs bulk packing. If your supplier packs 50 red T-shirts in one bag and 50 blue in individual poly bags, that's a packing inconsistency that affects your JHB warehouse receiving process. CloudSpects Multi-Variation Inspection Workflow CloudSpects handles multi-SKU orders from 1688 every week. Our cluster-inspection workflow covers: Pre-inspection: you send the full SKU matrix (style × color × size), we build a sampling plan per cluster On-site: one inspector handles measurement verification per cluster, a second runs color fastness and GSM testing Reporting: defect matrix organized by cluster — you see exactly which color or size has the problem Photo evidence: 50+ photos per inspection covering each cluster's critical areas Frequently Asked Questions Does cluster sampling cost more than standard inspection?
Most multi-variation orders require 1.5-2 inspector-days at the factory due to cluster separation. From $169/man-day, a typical 60-SKU T-shirt order costs $250-340.
I only have 3 colors but 10 sizes per color — do I still need cluster sampling?
Yes — if any color or size has a quality defect that's unique to it, standard random sampling may miss it. Cluster sampling guarantees every color-size intersection is tested.
What happens if one cluster fails AQL but others pass?
You can reject only the failed cluster. The factory reworks that color or size while the rest ships. CloudSpects inspects the reworked batch before release.