How to choose an air purifier: the decision in order

Almost every bad air purifier purchase starts by comparing machines. It should start by naming a pollutant. "Clean air" is four different jobs wearing one label, and the machine that is excellent at one of them can be close to useless at another.

• Smog and PM2.5 — fine combustion particles, the winter problem in Delhi and in Beijing alike. This is a question of how much air a machine moves through a filter per hour, which is to say a question of CADR and room size, and almost nothing else.
• Dust and allergens — pollen, dust-mite debris, pet dander. Bigger particles, easier to catch, but they are stirred up constantly by people moving around, so the machine has to keep running rather than clear the room once.
• Formaldehyde and other VOCs — gases, not particles. A HEPA filter does not catch a gas at all. This needs activated carbon or a catalyst, and plenty of it.
• Pets and odours — also gas-phase, also a carbon job, but a far less demanding one than formaldehyde.

That last distinction is the one the catalogue makes most visibly. Every machine here filters particles. Only some of them do anything serious about gases, which is why each row carries a separate VOC-removal rating on a 1-to-3 scale. Five of the twelve air machines sit at 1 out of 3. Three sit at 3 out of 3. The rest are in between.

India and China are not the same problem

Smog matters in both markets. Formaldehyde mostly does not. In China a large share of buyers are moving into a new-build or freshly renovated flat, where boards, adhesives and paint release formaldehyde for months and sometimes years afterwards — which is why the Chinese machines in this catalogue carry real VOC ratings and why Chinese product pages quote a separate formaldehyde CADR under GB/T 18801 at all. In India the same buyer is far more often shopping for October-to-January particulate smog, plus year-round road dust.

The catalogue follows that split. Of the three air machines rated 3 out of 3 for VOC removal, two are sold only in China. Eight of the twelve air machines are available in India, eight in China, and four in both. If you are shopping in India for a new-build smell, your shortlist is genuinely shorter, and it is worth knowing that before you start. Formaldehyde and VOC removal is the long version of that argument.

This is the step that decides whether the purchase works, and it is the one most often skipped. The number that matters is CADR — clean air delivery rate, in cubic metres per hour, defined by the AHAM AC-1 standard. It is not a filter rating. It is a throughput: how much genuinely clean air the machine hands back to the room every hour.

Across the twelve air machines here CADR runs from 240 m³/h to 600 m³/h, with a median of 355 m³/h. Two machines tie at the top of that range. Stated coverage runs from 28 m² to 85 m², median 44 m².

Here is the part that catches people. A coverage figure is not a standard. AHAM AC-1 defines CADR; nothing defines the square metres printed next to it, and every brand picks its own conversion. In this catalogue most rows divide CADR by about 8.3 to get coverage — which, at a 2.5 m ceiling, works out at roughly 3.3 complete air changes per hour. For smog or for allergies the figure usually wanted is 4 to 5. So for those two jobs you should be sizing up from the stated coverage, not treating it as a ceiling.

It is worth being clear about why undersizing cannot be fixed later. A purifier is not clearing a fixed quantity of pollution once and finishing; it is competing continuously against outdoor air leaking in and against particles generated indoors. What you end up living in is a balance between those two rates. A machine too small for the room reaches a worse balance and then stays there, however long you leave it on. Running it for twelve hours instead of three does not improve the steady state — it only means you spend more of the day inside it.

The arithmetic behind that, worked all the way through, with the ceiling height and the air-change model spelled out, is in understanding CADR and room size . It is the single most useful twenty minutes in this set, because a machine that is one size too small never stops being one size too small.

Only after the size question does the filter grade start to matter — and it matters less than the marketing suggests, because the labels are a mess.

EN 1822 defines the H-classes precisely: H13 captures at least 99.95% of particles at the most-penetrating particle size, H14 at least 99.995%. Those are real, testable figures. What appears on boxes is not. The twelve air machines in this catalogue between them carry eight different filter descriptions: HEPA, H13, H13 HEPA, HEPA H13, HEPA H13 (sealed), HEPASilent, HyperHEPA H14 and Green True HEPA. Several of those are brand names with the word HEPA welded on. Only some of them name a class you can look up.

The practical reading: treat "H13" or "H14" as meaningful and treat everything else as a description of a product rather than a grade. Then ask the question that actually separates machines at the same grade — whether the filter is sealed into its housing, so that air cannot slip round the edge of it. HEPA grades explained covers the classes, the bypass problem and what the various in-house names are standing in for.

A purifier has two prices. The one on the shelf, and the one that arrives every six to twelve months in the form of a replacement filter. The second is the one people forget, and over five years it can be the larger of the two.

Every air row here carries a filter-cost level from 1 to 3. Three machines sit at 1 out of 3, two at 3 out of 3, and the remaining seven in the middle. It is worth reading that alongside the purchase price rather than instead of it.

Two cautions about that table. The Indian and Chinese prices are separate figures for separate markets, not converted versions of one another — never read one as the other. And the razor-and-blades pattern, cheap machine with expensive refills, does not really show up in this list: all three machines rated 1 out of 3 for filter cost sit below the median price in China, while both machines rated 3 out of 3 are the two most expensive on the list in both markets. That said, the ₹8,999 machine still sits at 2 out of 3, so cheap to buy does not automatically mean cheap to keep.

There is one more variable nobody prints on a box: how hard you run it. A machine on high in a smoggy January loads its filter far faster than the same machine idling in June. Air purifier running costs puts filter intervals, electricity and pre-filter maintenance into one annual figure.

For a machine in a living room, noise is a minor consideration. For a machine two metres from your head for eight hours a night, it is close to a veto.

Sleep-mode noise across this catalogue runs from 22 dB to 35 dB, with a median of 31 dB. That span sounds small and is not: decibels are logarithmic, so 13 dB of difference is comfortably more than a doubling in perceived loudness. The quietest machine here and the loudest are not in the same conversation for a light sleeper.

The catch is that the sleep-mode figure and the CADR figure describe two different machines. CADR is measured on the highest setting under AHAM AC-1, where the fan is loud. Sleep mode is the lowest setting, where clean-air output is a fraction of the headline number. You cannot have both at once, which is why bedroom sizing wants headroom: a machine comfortably bigger than the room can hit a useful air-change rate while still running slowly and quietly.

Where the machine physically stands matters roughly as much as which machine it is. A purifier wedged into a corner behind a sofa is recirculating its own output and cleaning a pocket of air rather than a room. Where to put an air purifier covers clearance, doors, bedroom versus living room when the budget only stretches to one, and whether to run it continuously.

Ten brands appear across the twelve air machines here, and seven of the twelve are smart — app control, auto mode driven by an onboard particle sensor, an air-quality readout. Five are not.

Smart is worth paying for in one specific case: auto mode. A machine that raises its own fan speed when the sensor sees particles and drops back when it does not will, in practice, run at a sensible speed far more often than a machine you keep meaning to adjust. If that is what you are buying, buy it for the sensor and the automation rather than the phone app.

Things that genuinely do not belong in a purchase decision: ionisers and "plasma" stages sold as the main event rather than an extra; UV lamps in a unit where the air passes in a fraction of a second; and any claim about preventing illness. What is well established is that these machines reduce indoor particulate concentrations, and that is measurable. What that reduction does for any particular person is a much weaker claim, and anyone stating it confidently is overreaching.

• A sealed filter path beats a higher grade of filter that air can bypass.
• A bigger machine run slowly beats a smaller machine run flat out — quieter, and the filter lasts longer.
• A second cheap machine in a second room usually beats one expensive machine trying to serve a whole flat through open doors.
• Ozone-generating devices sold as air purifiers are a different category of product. Do not buy one for a bedroom.

1. 1

   Name the pollutant

   Smog and PM2.5, dust and allergens, formaldehyde, or pets and odours. If it is formaldehyde, the VOC rating becomes the first filter on your shortlist rather than the last — and in India that shortlist is short.

2. 2

   Measure the room

   Floor area and ceiling height, of the room the machine will actually live in — not the flat. Then work out the CADR that gives you 4 to 5 air changes an hour rather than trusting the coverage figure on the box.

3. 3

   Check the filter grade and the seal

   H13 and H14 are EN 1822 classes and mean something. The other six descriptions in this catalogue are brand language. Ask whether the filter is sealed into its housing.

4. 4

   Price a year, not a day

   Machine price plus filters plus electricity. The filter-cost level in each row is the shorthand; the running-costs guide turns it into an annual figure.

5. 5

   Check the noise against where it stands

   Sleep-mode noise runs 22 dB to 35 dB here. If it goes in a bedroom, size up so it can do its work slowly.

6. 6

   Then, and only then, compare brands

   Auto mode is the smart feature worth money. Most of the rest is not.

If you would rather see the output before doing the work, the pages below are that method already run for a particular room and a particular worry. None will be exactly your flat; they are here to show which machine moves when one answer changes.