How we choose what to recommend

Last updated: August 2026

The honest starting point

We do not test products by hand. No robot vacuum here has been run across a rug, no microphone has been recorded in a treated room. Anything you read on this site is desk research, and the recommendations should be judged on that basis.

That rules out some claims entirely. We will never tell you a vacuum "sounded quieter in use" or that a preamp "had more headroom than expected", because we would be making it up. What research can do well is narrow a category down to the few models that plausibly fit your situation, and explain the trade-off you are actually choosing between. That is the job this site takes on.

The evidence used

Manufacturer specifications. Suction in pascals, polar pattern, sample rate, tank capacity, filter class. Where a number appears on a page, it comes from here, and where two sources disagree the page says so rather than picking one.
Current retail listings. Availability, configuration, what is actually in the box. Listings change constantly and are the single biggest source of a page going stale.
The pattern in owner reports. Not the star average, which mostly measures delivery and packaging. What matters is the recurring specific complaint — the app that drops connection, the filter that is expensive to replace, the mount that does not fit a standard arm. One angry review is noise; the same fault named a hundred times is a specification the manufacturer did not publish.
Category context. What a segment costs, and where the price stops buying anything a normal person notices.

What gets a product excluded

It is no longer listed, or is routinely out of stock. A recommendation you cannot act on is not a recommendation.
The specification is missing or contradictory across the manufacturer's own materials.
Owner reports converge on a fault that defeats the product's main purpose.
It only wins on price. Cheapest is a fact, not a recommendation, and it gets its own line rather than the top slot.
It is functionally identical to something already listed. Two versions of the same machine in one comparison waste your time.

How "best for" is decided

Every comparison is organised by situation rather than rank, because the honest answer to "what is the best robot vacuum" is that it depends on your floors, your pets and your patience. A page will tell you which model suits a small flat with hard floors and which suits a large house with a shedding dog, and it will say plainly when the difference between two options is too small to matter.

Where a product is a poor fit, the page says so. "Who should avoid this" is usually more useful than another paragraph of praise, and it is the part most affiliate pages omit because it costs them clicks.

Prices and how they age

Prices are not published on these pages. They move too quickly for a static page to be trustworthy, and a stale price is worse than none. Where price matters to a decision it is described in ranges and relative terms; the live figure is on the retailer's own listing.

How often pages are rechecked

Links are audited automatically, and pages are revisited when a product is discontinued, when a category gets a significant new release, or when a reader reports a problem. The date at the top of each page reflects its last substantive change, not an automated timestamp bump.

Where the money comes from

Amazon pays a commission on qualifying purchases made through links on this site. The commission rate does not vary by product in a way that would make one recommendation more profitable than another, and the ordering of recommendations is not influenced by it. The full disclosure is here.

Tell us when this fails

If a page overstates what is known, cites a specification that turns out to be wrong, or recommends something that has been discontinued, email kaptenayov@gmail.com. See about for who is behind the site.