Editorial standards
Every directory holds the same product names. What differs is the method, so ours is written down here in full, including what it refuses to publish.
The process
01
Every product is a file in a public repository, so a change to what this site says about a tool is a diff someone can read, review and revert. A database row can be edited silently; a commit cannot. It also means the catalogue can be checked by machine before it is published, which is where the next two steps happen.
02
Researchers work from a fixed brief: alternatives first, then the questions a buyer actually asks, then concrete limitations, then integrations, then company history. The brief bans vague criticism, because "steep learning curve" is true of everything and checkable on nothing. What is wanted instead is the specific thing that is hard: no Linux build, SSO on the top tier only, an API capped at 100 requests a minute.
03
Before anything is published, an automated gate removes claims citing an aggregator or a bare social profile, price and compliance claims with no source, banned vague phrases, and alternatives pointing at products that do not exist. It runs on the same rules the audit reports, so what passes review is exactly what ships.
04
Each proposed pairing is judged twice, by reviewers who cannot see each other's reasoning, on one question: would a buyer evaluating this product seriously consider that one instead. Only pairings that survive both rounds are published. The second round exists because the first over-rejected, deleting real competitors, and a comparison deleted in error is a page a reader needed.
The refusals
These are the ones worth stating, because each of them costs us something a competitor takes for free.
Softwr hosts no reviews. Ratings aggregated from other sites were removed from the catalogue rather than displayed, because publishing a number we did not collect, next to our own analysis, invites you to read it as our assessment. What replaces it is what a tool costs, where it runs, and what it is concretely bad at.
Anything that could go stale carries the page it came from: prices, plan limits, funding amounts, acquisition dates. Where a claim could not be traced to a primary source it was deleted rather than softened. Across the research passes so far, 406 individual claims were removed for citing an aggregator, a personal blog, or nothing at all.
A comparison page only exists where a reviewer confirmed two products genuinely compete. 225 of 1,841 candidate pairings were deleted for failing that test, including a font inspector matched to a dark mode extension and a code formatter matched to a package manager. A wrong comparison is worse than a missing one, because it asserts a rivalry that does not exist.
Being listed cannot be bought, and no filter result can be bought. What a vendor can buy is a numbered place in the results on the home page, on a category listing or on the catalogue, and every paid place carries a Sponsored label. It does not move anybody else: the editorial order flows around a sold position rather than being reordered by it, and a product that buys a place appears once rather than twice. Comparison and alternatives pages sell nothing at all. Where a link earns a commission it is disclosed on the page carrying it.
Where we are
Research has been completed on a few hundred of the products listed here. The rest carry the basics, pricing, platforms and a description, and no researched limitations or verified alternatives yet. Those pages say less because there is less to say, rather than being padded to look finished.
Company history is the thinnest field of all, and deliberately the last priority: it matters least to somebody choosing a tool this afternoon. Where an event could not be traced to a primary source it was dropped rather than approximated.
If something here is wrong, the fastest way to get it fixed is to tell us which page and what is wrong with it. Corrections are made against the source, not the complaint.
Pricing moves without announcement and it is the field most likely to be wrong on any given day, so treat every figure here as the starting point for a check rather than the end of one. Each price carries the vendor page it was read from, which is the fastest way to confirm it yourself.
Capability claims move far more slowly. Whether a tool runs offline, exposes an API or supports SSO is usually true for years at a time, and those are the claims the research prioritises for that reason.
A correction does not require you to be polite or detailed. Name the page and the thing that is wrong, and it gets checked against the source rather than against the complaint. If the source disagrees with us, we change; if it disagrees with you, we will say so and show you why.