GSM tells you the weight, not the quality
GSM — grams per square metre — is the number every quotation leads with, and it is genuinely useful. It tracks how much fibre is in the cloth, which in turn drives absorbency, how substantial the cloth feels, and how long it survives washing. For general household retail, 220–260 GSM is where most successful programmes sit: lint-free on glass, convincing in the hand, and priced where volume happens. Above 300 GSM you are into premium and automotive territory, where extra absorbency earns the higher shelf price. Below 200 GSM the cloth feels thin and shoppers notice on the first use.
What GSM does not tell you is whether the fibre works. That is decided by a property you cannot photograph.
Split fibre is the specification that matters most
Microfiber only outperforms cotton because each filament is mechanically split into wedge-shaped strands, multiplying surface area and creating capillary channels that pull grease and fine dust off a surface without chemicals. Unsplit microfiber is smooth filament. It pushes dust around and smears grease.
Two problems follow. First, unsplit yarn is meaningfully cheaper. Second, at retail distance the two are indistinguishable — same colour, same weave, same GSM on the label, same photograph. The field test is simple: draw a dry cloth across a clean pane of glass or the palm of your hand. Split fibre grabs and drags slightly. Unsplit fibre glides.
Ask your supplier to quote split and unsplit as two separate line items with two separate prices. A supplier who cannot or will not separate them is telling you something.
80/20 versus 90/10 is a real difference, not a rounding error
The blend is polyester and polyamide. Polyester does the scrubbing and provides structure; polyamide does the absorbing and gives the cloth its softness and density. An 80/20 blend absorbs more, feels denser and lasts longer through wash cycles. A 90/10 blend costs less and is entirely adequate for dry dusting or promotional giveaways, but it is noticeably less absorbent when wet.
Neither is wrong. Choosing 90/10 to hit a price point is a legitimate decision. Being quoted 90/10 while believing you bought 80/20 is not, and the two are impossible to tell apart without a burn test or lab analysis. Put the ratio in the purchase order.
Edges: where cheap cloths fail and expensive ones get complaints
There are three treatments. Overlocked edges are stitched, cheapest, and perfectly fine for kitchen and general-purpose cloths. Satin-bound edges wrap the border in a ribbon, which looks premium and protects against fraying. Laser-cut edgeless construction has no border at all.
Edge choice becomes critical for anything used on paint or a screen. A stitched border is harder than the pile around it, and it is the part that leaves a mark on clear coat. If the cloth will touch a car or a display, specify edgeless or satin-bound. For kitchen cloths, overlocked is the sensible economy.
The two traps
Trap one: sampling from a different run than you order from. The sample is often pulled from premium stock while production runs on whatever yarn is on the floor that week. The defence is a retained reference sample — one physical piece from the approved sample, held by both sides and referenced by order number, so a later dispute is settled by comparison rather than by argument. Ask whether your supplier retains references. Many do not.
Trap two: buying wash life on trust. "300 washes" appears on nearly every specification sheet and is almost never tested. If wash durability is part of your product claim, ask for the test report or run 20 domestic cycles on the sample yourself before the purchase order. Twenty cycles is enough to reveal pile collapse and edge fraying.
A specification you can send today
Everything above reduces to one paragraph in a purchase order. Fill in the brackets:
Microfiber cleaning cloth, [240] GSM, [80% polyester / 20% polyamide], split fibre (confirmed), [warp-knitted pearl weave], [30×40] cm finished size, [overlocked] edges, colour [Pantone / stock reference], packed [3 pcs per printed sleeve]. Retained reference sample required. AQL 2.5 pre-shipment inspection.
That paragraph removes most of the ways this category goes wrong.