What the test actually does
In a Martindale test the fabric is held on a specimen holder and rubbed against a standard wool abradant under a fixed load, in a figure-of-eight path, until either the fabric shows a defined degree of damage or a set number of rubs has been reached. The contract standard is 12 kPa of head pressure, run to EN ISO 12947-2, with the end point defined as two broken yarns or a specified change in appearance. Four specimens are normally tested and the result is the lowest, not the average.
That last point matters more than buyers expect. A swatch quoted at 40,000 cycles is a statement that the weakest of four specimens survived 40,000 rubs, but the flat specimen in a rig is not a seat cushion under a person. The rig rubs the whole face evenly; a diner abrades a patch roughly 100 mm across at the front edge of the seat, plus two small pressure points at the hips. Same fabric, very different load distribution.
So read a rub count as one input, not as a durability score. Two fabrics at 40,000 cycles can behave completely differently in year two if one has a loose weave that snags on a zip and the other has a tight, high-twist construction that resists it.
The three bands you will actually be offered
Around 20,000 cycles is the entry band for contract work. It is honest for a low-turn room, a private dining area used twice a week, or a banquette back that nobody sits against with a belt buckle. It is not enough for an all-day seat, and a supplier who offers it for a 200 cover restaurant is either assuming you will replace in two years or has not asked about your covers. Most restaurant chairs bought at this band in a room turning more than four times a day are recovered before the second winter is out.
Around 40,000 cycles is the working standard for general restaurant seating and the band we quote most. It covers a room running four to six turns a day with normal cleaning, and it holds up on the seat rather than just on the back. If your specification has one number in it, this is the most defensible one.
Around 80,000 cycles is severe contract, and it earns its price in specific situations: bench seating in a high-turn room, seats at a counter where people perch and swivel, and any surface that takes a bag or a coat hook as well as a person. Beyond that, the gain flattens. Above 80,000 the limiting factor stops being abrasion and becomes soiling, seam failure or foam collapse, and paying for 100,000 cycles to solve a staining problem buys nothing.
Turning cycles into years of service
Do the arithmetic for your own room before you accept any band. Sixty seats at three turns a day over 300 trading days is 54,000 sittings a year, which is about 90 sittings per seat per month. If one sitting abrades the front edge of the seat in a way roughly comparable to a few hundred even rubs on a flat specimen, a fabric in the 20,000 band is on borrowed time at the highest-wear seats in the room.
The seats that fail first are predictable: the ones nearest the door, the ones at the end of a run where people slide in, and any seat a guest uses to put a bag down before sitting. Rotating the loose chairs in a room every six months spreads the wear across the set and is the cheapest life extension available to an operator, and it costs nothing on a fixed banquette, which is why banquette fronts should be specified one band higher than the loose chairs beside them.
Our own rule of thumb for a full-service room is 40,000 on loose chairs and 60,000 to 80,000 on fixed seating, with the front edge of the seat treated as the deciding surface. That is a working number from our own project history, not a laboratory claim, and it is the number we argue with buyers about least.
Abrasion is not stain resistance
On restaurant dining chairs the cover is the first thing a guest judges and the first thing an operator notices when it goes wrong, which is why the rub count gets quoted in every proposal and why it gets misread in most of them.
These are separate tests and separate failure modes, and the confusion between them causes more warranty arguments than any other single item. Martindale measures how the fibres resist being rubbed away. It says nothing about whether a fabric holds a wine ring, a sauce spill or a line of oil at the seat front, and a high rub count can even work against you: tightly woven, high-abrasion fabrics are often the ones that hold liquid at the surface where it wicks into the yarn.
Stain performance is a different set of properties: fibre type, whether the yarn is solution-dyed, whether there is a fluorocarbon or silicone finish, and how the backing is constructed. Ask for the stain test rather than the rub count when staining is your problem, and ask what cleaning agents the finish tolerates, because a finish that fails at the first quat-based sanitiser is worse than no finish at all.
Fire performance is a third, separate question. In most markets the requirement is a cigarette and match test on the composite of fabric plus foam, not on the fabric alone, so a certificate for the face fabric does not satisfy the requirement. Ask for the composite report, with the foam density and the barrier if one is used, because the same fabric on two different foams can give two different results.
Pilling, snagging and the seams nobody tests
Pilling is graded 1 to 5 under EN ISO 12945-2, and a 4 after 2,000 rubs is a working minimum for a seat that gets used daily. A 2 or 3 looks acceptable on a swatch and turns into a grey, fuzzy seat within a season, which reads to a guest as dirt rather than as wear. Wool blends and loose-spun chenille are the usual offenders.
Snagging is a construction problem rather than a fibre problem. A loose weave catches on a bag buckle or a zip and pulls a float; once one float is out, the seat has a visible line that no amount of cleaning removes. Tight, high-twist weaves and a bonded backing both help, and on a bench seat where people slide along, the backing is doing as much work as the face.
Seams are where upholstery fails structurally while the face is still perfect. Six to eight stitches per 25 mm in a suitable thread, a 10 mm seam allowance and a double-stitched or taped seam at the seat front will outlast a 6 stitch seam by years. Ask for the seam construction in writing; it is one line on the specification and it is not something you can inspect from the outside of a finished chair.
Writing the upholstery line in the purchase order
A usable line reads like this: face fabric, composition, Martindale 40,000 minimum to EN ISO 12947-2, pilling grade 4 minimum, stain finish specified, composite fire test to the destination market, seams at 6 to 8 stitches per 25 mm. Any item missing from that list is a decision the factory will make for you, and factories make it in favour of the cheaper option roughly every time.
Then tie the fabric to the model rather than to the order. A JP-C023 upholstered dining chair and a JP-C007 wood dining chair with a padded seat take the same fabric very differently, because one has a full upholstered back that takes abrasion from shoulders and the other only has a seat pad. The test data has to be read against the shape it is wrapped around.
On upholstered restaurant chairs we hold the fabric report per article, together with the composite fire report and the foam density, and we re-test when a mill changes its finish. It is the same discipline we apply to frame construction in the restaurant furniture specification we issue with each project, because a specification that covers the frame and leaves the cover open is only half a document.
If you are comparing swatches from three suppliers, send the same brief to all three: covers per day, seat count, cleaning regime and expected replacement cycle. A request a restaurant furniture quote returned against that brief is comparable; three quotes against nothing but a photograph are not, and the numbers in them will not mean the same thing.