Five stages in series. The cheapest one is first, the dearest is last.

Every stage starves every stage after it

A five-stage system is a series circuit, not a set of parallel filters. Water has to pass through the sediment cartridge, then the carbon block, then the pump, then the membrane, then the post-filter, to reach the tap. That has one consequence which explains this whole page: anything that restricts flow anywhere in the chain reduces the flow everywhere downstream of it. So "the flow is slow" tells you nothing about which stage is responsible. You have to find the pressure drop, and the only way to do that cheaply is to work from the free checks toward the expensive parts.

And the ordering is unfortunate, because the chain is built the wrong way round for diagnosis: the cheapest part is first and the most expensive part is last. A sediment cartridge costs a few dollars. A membrane costs twenty times that. So the natural instinct — replace the filter — is aimed at the wrong end of the price list before a single free test has been done. Repair data from service call-outs is consistently blunt about this: most visits are for filters that were simply overdue, and a large share could have been closed by the homeowner with a pressure gauge and a bicycle pump.

Which brings us to the part almost nobody checks. An RO system stores its water in a bladder tank with air behind it — typically 7 to 10 psi of pre-charge with the tank empty. That air is what pushes the water out of the tap. When it leaks down, the tank still fills, the membrane still works, the filters are still fine, and the tap dribbles. Nothing is broken. You add air with a bicycle pump and the flow comes back. It is the single most common cause of "slow flow" that is not a filter, and it costs nothing to check.

Then there is the trap in the carbon stage. Sediment cartridges fail honestly: they go brown, then black, and you can see it. Carbon does not. Activated carbon works by adsorption, which is a chemical process — it exhausts its capacity long before it looks dirty, and a saturated carbon block passes chlorine straight through to the membrane. Chlorine is what degrades RO membranes. So a carbon cartridge that “looks fine” is the mechanism by which a membrane that should have lasted three years dies in eighteen months. Replace carbon on the calendar, not on appearance.

Three faults, and the order to check them in

• Inlet pressure or a restriction. Free to check, and the most common cause that is not a cartridge. Open a normal tap elsewhere in the house: if that has also slowed, the problem is the building’s supply, not the filter — high floors and peak hours do this. Then look at the inlet tee valve, which is very often left half open after installation, and run your hand along the tubing looking for a sharp bend where it was pushed back behind the cupboard. RO needs at least about 40 psi at the inlet, and 50 to 80 is where it is happy.
• The tank pre-charge or the inlet solenoid. The signature is distinctive: the flow starts strong for a few seconds and then dies, and you have to wait for the system to "catch up". That is a tank with no air in it — the water goes out until the pressure equalises and then stops. Measure with a tyre gauge with the tank drained, add air in short bursts, and if it will not hold pressure the bladder has failed and the tank is scrap. While you are there, listen for the solenoid: a faint hum means it is working, silence plus weak flow means it is clogged.
• The cartridges, in order. Sediment first, because it is first in the chain and clogs fastest — three to six months in most municipal water. Then carbon at six to twelve months. Membrane last, at two to three years, and only when the numbers say so: a TDS meter reading rejection below about 85 to 90 per cent, or product water above 15 per cent of the inlet TDS. Change the membrane on a hunch and you have spent the most money on the least likely cause.

The one number that is not a guess

A TDS meter costs less than one cartridge and turns the last decision into arithmetic: (inlet − product) ÷ inlet. A healthy membrane rejects 90 to 95 per cent. Below 85 per cent it is done. And note what it does not tell you: high product TDS with normal flow is not a worn membrane — it is usually a membrane housing O-ring that is not seated, a membrane fitted the wrong way round, or a counterfeit membrane. That is a reinstall, not a replacement, and it is worth checking before you buy anything.