Which Robot Pool Cleaner Fits Your Pool? A Sizing Guide

Four machines, and they are not degrees of the same thing

“Robot” gets attached to anything that crawls around a pool on its own, which buries the distinction that actually decides what you should buy: where the machine gets its power, and where the dirt ends up. There are four classes, and the differences between them are mechanical.

A suction-side cleaner is a mouth on a hose. It has no motor of its own. You plug it into the skimmer or a dedicated suction port, your pool pump does the pulling, and everything it collects travels back through the pump basket and into your filter. It is the cheapest way to stop pushing a vacuum head around yourself, and the trade is that your filter now eats every leaf the cleaner finds.

A pressure-side cleaner runs the other way round. It hangs off a return line, usually driven by a dedicated booster pump, and the water pushed through it both moves the machine and lifts debris into a bag it carries. The bag is the whole argument for the class: on a pool under trees, leaves never reach your filter, which is the difference between backwashing weekly and backwashing when you think of it. You pay for that with a second pump running while it cleans.

Both robotic classes are self-contained — their own motor, their own filter basket, no interest in what your pump is doing. That independence is the fact that shows up on your electricity bill later. A corded robot takes mains power through a low-voltage supply that sits on the deck. A cordless robot carries a battery and comes out of the water to charge. Everything else about the pair is the argument you have already had about vacuum cleaners: a cable that reaches everywhere and never runs down, against a machine you can drop in the water without hunting for a socket.

Start with the surface, not the sales page

Every question further down — how long a cycle needs to be, how much cable, whether one charge is enough — is a question about area. Volume is what the chemistry runs on; area is what a cleaner has to physically cross. Listings are quoted in gallons because gallons are what people know about their pool, and gallons tell a cleaner nothing.

There are two areas worth knowing, not one. The floor is what any cleaner will cover. The floor plus the walls is what a wall-climbing machine signs up for, and it is a much bigger number than people expect: on the 24 × 12 ft, 3 ft to 6 ft pool below, the floor is 27 m² and the walls add another 30 — more surface above the floor than on it. Deciding you do not need wall coverage is a legitimate way to buy a smaller machine. Assuming the walls are a rounding error is not.

If your pool is not one of these four, the pool volume calculator takes the same dimensions and runs the same code that produced this table — shape, length, width, shallow and deep depth — and gives you the volume in gallons and litres. The area arithmetic below is that same geometry, one step further on.

What that surface asks for in run time

A cleaner covers a certain amount of surface per minute, and it is far less than a single pass would suggest. Crawling at around 12 m a minute with a 30 cm intake, a machine sweeps something like 3.6 m² a minute — if it never crossed its own path. It does. Patterns overlap, walls and waterline are separate work, and most cycles go over the same ground more than once, which drops effective coverage to somewhere around 0.6 to 1.2 m² a minute: 3 to 6 times slower than the pass. Real cycles on the 27 m² floor in the table below run 20–45 minutes at that rate, and a normal pool lands inside the 1.5 to 3 hour cycles manufacturers actually quote.

Both halves of that are estimates rather than measurements: the crawl speed and intake width are what these machines look like in the water, not a figure off a spec sheet, and the coverage band was chosen so a normal pool lands inside the cycle lengths the market sells. The 3-to-6 gap between the two follows from those numbers rather than standing on its own.

Use the slow end when you are asking “will one run finish my pool”, because that is the end that matters when the answer is no.

The cable number is the one people get wrong

Cable length has three segments and they are all measurable. The run from the power supply to the waterline, which this table assumes is 2 m and yours might not be. The drop down the deep-end wall. And the longest path along the bottom, which is the pool’s length plus its width — the furthest apart any two points in the pool can be for a machine driving in straight legs. A taut cable needs less than that, which is the point: the sum is a guarantee rather than a hope.

For the 40 × 20 ft, 4 ft to 8 ft pool that comes to 23 m or 75 ft. Cable length is printed on every spec sheet, and it is worth checking against that figure rather than against the pool’s length: 18 m is a common mid-range cable, and 18 is less than 23. Nothing about that is a defect — it is a machine sized for a smaller pool — but it means the far corner never gets cleaned, and running it twice does not fix a cable that does not reach. Measure your own deck run first; if your supply sits 5 meters back, add 3 m (10 ft) to every figure in that column, which prints both units.

Which class fits which pool

The grid below is judgment rather than arithmetic, and every verdict in it rests on something structural: above-ground plumbing has no return line for a booster pump; a long suction hose loads your filter with everything it picks up; steep walls and a waterline are where hose-driven cleaners stop; a battery cycle is finite, and one of these pools asks for more of it than one charge holds.

Above-ground round. Flat floor, soft wall, no deep end — the easiest pool in this list to clean and the hardest to justify spending on. A suction-side unit does it. Pressure-side is out, because there is no booster line to hang it from. A robot works fine and is mostly buying you time back rather than a cleaner pool; if you go that way, cordless sidesteps the real problem, which is that above-ground decks rarely have a socket within 18 m of the far side.

Small in-ground. Everything works here, which makes this the row where price and convenience decide rather than capability. 57 m² of floor and walls is 45 to 95 minutes of work, inside most cycle lengths on the market. The pressure-side caveat is economic rather than mechanical: running a second pump to clean a pool this size is a lot of machinery for the job.

Large in-ground. 141 m² and 120 to 235 minutes changes the answer. Suction-side still works, but it is slow and it sends every leaf into your filter, so you buy the cleaner and inherit a backwashing habit. Pressure-side comes into its own — the debris bag is the right answer at this volume. A corded robot is the other right answer, provided its cable clears 23 m. Cordless is the one to check carefully, because the numbers above run past what a single charge is usually sold as.

Freeform with a deep end. Curves and a deep end are where hose-driven cleaners get stuck: a suction unit that navigates by bumping into things spends a lot of its life in the same corner, and the waterline stays yours regardless. This is the shape that genuinely justifies a corded robot, because a programmed path and tracked drive handle what a hose cannot. Cordless is capable here too, with one number worth staring at: 90 m² works out at 75 to 150 minutes, and 150 is exactly the top of the range these are sold with. That is not a margin.

What actually decides whether you like it

Sizing gets you a machine that can finish. Whether you are still happy with it in year two comes down to four things, and none of them is on the front of the box.

• What your pool collects. Fine silt and pollen ask for a fine filter; leaves and twigs ask for capacity. A cleaner optimized for one is mediocre at the other, and a fine cartridge under an oak tree clogs in a single cycle.
• Whether the waterline is included. The scum line at the water’s edge is the part everyone can see, and it is the part most classes do not touch. If nobody is going to scrub it, this is the feature to pay for.
• How you empty it. You will do this after every single clean, wet, probably barefoot. A basket that lifts out of the top is a different life from a bag you unclip underneath, and no spec sheet ranks them.
• Cable tangle against remembering to charge. The real cordless-versus-corded question. A cable in a pool twists; a swivel helps and does not eliminate it. A battery does not tangle and does need planning, and a cleaner you forgot to charge is a cleaner that did not run.

What it does to your running costs

A suction or pressure-side cleaner cannot work unless something else is running: your pump for the first, usually a booster pump for the second. Cleaning time and filtration time are welded together, so wanting a cleaner floor means running the pump longer.

A robot breaks that link. It filters the water it moves through its own basket, independently of your circulation, so the hours it spends cleaning are not hours your pump has to spend as well — which for a lot of pools means the pump schedule can come down to whatever turnover and chemistry actually need. That is the honest shape of the saving. Anyone quoting you a figure in kilowatt-hours is quoting their pump, their tariff and their season, not yours.

When not to buy a robot

A robot is the wrong purchase in three common situations, and in two of them the cheaper machine is also the better one.

• A small above-ground pool with a flat floor. There is not enough work here for the machine to earn back the difference. A suction cleaner off the skimmer is the right tool.
• A heavy leaf load. A robot’s basket fills and then it is a machine pushing leaves around. Under trees, an inline leaf canister or a pressure-side cleaner with a real bag is the first purchase; a robot, if you want one, is the second.
• Water that is already green or cloudy. A cleaner picks up what has settled. It is not a treatment, and running one in an algae bloom mostly clogs the filter. Fix the chemistry first — the pool chemical calculator works out the doses — then let the cleaner deal with what falls out of suspension.

Which numbers here are computed, and which are estimates

Computed. The areas and volumes are arithmetic on the dimensions printed beside them, run through the same pool volume code the calculator uses. Change a dimension and they move with it, because they are generated at build time rather than typed in. Estimated terms do sit inside that arithmetic, and here is where: the deck-run margin is inside every cable figure, the coverage band divides every run-time figure, and the freeform row’s area and perimeter are both approximations.

Estimated. The rest, each named again here so none of it hides behind the computed figures. All of it should lose to the spec sheet of a specific machine the moment you have one.

• The coverage band, 0.6–1.2 m² per minute, and the single-pass figure it is argued down from — 12 m a minute across a 30 cm intake, so 3.6 m² — and with it the 3-to-6 gap between the two. Every run time on this page is an area divided by that band.
• The 0.45 factor in the freeform pool’s area, 0.45 × (width A + width B) × length. Rectangles, circles and ellipses are exact geometry; a kidney is not, and this approximation off manufacturer volume worksheets is what every freeform figure on this page — water, floor, walls, run time — is built on. It lands slightly under a true double-ellipse.
• The freeform perimeter, and therefore its wall area. It is taken as the ellipse through the pool’s bounding box, and a real kidney’s edge is longer than that because the concave notch adds length. So that row is a lower bound rather than a measurement — the smallest it could honestly be.
• The cycle ranges the two robotic classes are sold with, which is a reading of the market rather than a measurement.
• The 2 m deck run in the cable column. It is a placeholder for your yard, not a fact about it — measure the distance from your supply to the water and adjust.
• 18 m as a common mid-range cable length. A claim about what is on sale, and the thing most likely to be out of date first. The cable figure it is compared against is computed; this one is not.

Judgment. The fit grid. Each verdict is tied to a mechanical reason stated in the paragraph beneath it, so you can disagree with the reasoning rather than just the conclusion — which is also why there are no model names or prices anywhere on this page. Those change every season; the classes and the geometry do not.