Tile Buying in China: Water Absorption, Calibre and the Report That Says "Conforms"

I once spent three days arguing with a Foshan tile factory about a 0.4 mm size difference on a 600x600 porcelain tile. My client's tiler had already started laying, the grout joints were visibly uneven, and the whole batch of 1,400 m² was sitting in a container on site. The factory's quality manager kept pointing at a test report that said "conforms." He was right, technically. The report conformed to the batch tolerance, and the batch tolerance was wider than the joint I had specified.

That is the thing people learn late about tiles: almost every claim on the box is true and almost none of it is what you actually needed. Here is what I check now, in the order I check it, and why.

Water absorption is a class, not a number

ISO 13006 splits tiles into tiles with water absorption below 0.5% (BIa), between 0.5 and 3% (BIb), and then increasingly porous classes up to BIII. Suppliers love to say "0.1% absorption, very dense." Ask for the actual test result on a specific production lot, and ask which laboratory. A 0.4% and a 0.05% tile are both BIa, but the 0.4% one will stain more in a wet area and will behave differently if you cut and polish the edge.

The practical split most buyers need: interior floors and anything in a bathroom or a commercial kitchen should be BIa, and I want to see under 0.1% for polished surfaces because polishing opens the surface. Exterior walls, some cladding, and low-cost wall tile can be BIb without drama. Anything that is going to freeze and thaw (outdoor paving, a rooftop terrace) needs to be the low end of BIa plus a frost resistance test on the same lot, not on last year's sample.

Ask for the report to name the lot number. A report without a lot number tells you what the factory can make, not what they shipped you.

PEI wear rating and R-value slip rating get confused constantly

PEI (now folded into ISO 10545-7) is abrasion resistance of the glaze, rated I to V. It says nothing about how slippery the tile is. R-values (R9 to R13, from DIN 51130) are measured on a ramp with oil and say nothing about durability. I have seen R11 tiles with PEI III that buyers rejected because traffic made them look dull in eighteen months, and PEI V tiles in a hotel bathroom that failed a slip inspection outright.

For a residential kitchen or living room, PEI III is plenty and PEI IV is what I spec for anything commercial. For the wet parts, R10 is a reasonable floor minimum indoors, R11 for commercial wet areas, and R12 or R13 for ramps, kitchens with oil, or changing rooms. The barefoot rating (A, B, C per DIN 51097) matters for pool decks and is a separate test; a tile can be R11 and still fail the barefoot test because the barefoot method uses water, not oil, and tests a different surface condition.

Calibre and tone: the two numbers nobody quotes

Every tile is fired, so every tile shrinks a bit, and kilns do not shrink perfectly evenly. That is why factories sort output into calibres (size groups) and tones (shade groups). A "600x600" tile might be a 600.2 in one calibre and a 599.3 in another. Both are the same nominal product. Mixed calibres in a box or in a mixed pallet cause exactly the crooked grout line problem I described at the top.

What I write into the order now: all tiles in one project must be the same calibre and same tone, calibre stated on each carton, and a maximum delta of 0.3 mm between the largest and smallest tile in the shipment, measured on three tiles from three cartons pulled at random. That last number is my own, not a standard, and factories push back on it. They usually accept it once they understand I am going to measure.

Rectified edges (cut after firing, so the edge is a true square) are worth the small premium if you are using narrow joints or doing large-format work. A non-rectified tile in a 2 mm joint will show every calibre difference and every firing warp. If you must use non-rectified, go to 3 mm joints and stop fighting physics.

Large format is a logistics problem before it is a design problem

Anything 900x1800 or bigger, and certainly 1200x2400 slabs, changes the whole equation. Breakage rate in transit goes from around 1-2% on 600x600 to genuinely meaningful numbers, and the packaging has to be A-frame standing on edge, never flat stacked. I have seen a 750-piece 1200x2400 order arrive with 11% broken because the factory stacked them flat in a 40HQ and the container flexed at sea.

Ask how they pack before you ask for the price. Standing A-frame, foam at every contact point, steel or plywood A-frame uprights, and a written loading pattern that the factory has actually used on export. Then put an agreed breakage allowance in the contract, and specify who eats it. My default is that the factory covers anything above 2% breakage on arrival with replacement in the next order, documented with photos at the unloading point. It is not perfect but it forces the packing conversation.

What I ask for before the first bulk order

The current ISO 10545 report covering the specific product and lot; the calibre/tone sorting procedure; a 1 m² dry-laid photo of the actual production batch on a level surface with a raking light, so I can see the warp; a packing drawing; and a reference from one existing customer in my region whose tile I can go and look at, ideally installed at least a year.

That last one kills more bad deals than all the certificates combined. A tile that has been down for a year in a climate similar to mine tells me about staining, scratching, efflorescence at the joints, and whether the surface finish holds up under cleaning chemicals. None of that shows up in a lab report.

Foshan remains the right place to buy most of this. The cluster's depth in glazes, big-format pressing, and finishing is still unmatched. But it is also the place where the gap between "the certificate is valid" and "the tile is right for your job" is widest, because there are so many sellers competing on price with products that pass the paperwork and nothing more. Spend the extra hour on calibre, tone and one installed reference. It is cheaper than 1,400 m² of regret.