The rules
Editorial standards
These are the rules every page on Automation Atlas is written and maintained under. They apply identically to every listing, whether or not the vendor has ever contacted us.
Voice and neutrality
Content is encyclopedic, neutral, and third-person. We state facts, not enthusiasm: a claim like "the most popular" appears only with a cited metric. Analytical conclusions, including those in comparisons and rankings, are stated with the evidence that supports them. First-person material is confined to clearly marked Editor's Notes, which draw on real consulting work and always pair a specific figure with an honest caveat.
Sourcing and citations
Factual claims are checked against primary sources: vendor pricing pages, official documentation, release notes, press releases, regulatory filings, and repository data. Where a claim exists only as a vendor statement, it is attributed as one. Third-party estimates are named, not laundered into unattributed figures. Sources are recorded in a citations database, and individual facts carry their source and the date they were checked.
Dating
Numbers are anchored to dates. "Supports 9,000+ apps as of July 2026" stays accurate as a historical snapshot; an undated number silently rots. Pricing tables show the date they were last verified, and superseded prices are replaced rather than left to mislead.
Verification and refresh cycles
Facts are classed by how fast they decay, and each class carries a maximum permitted age: staleness has a hard ceiling, and every fact shows its as-of date on the page, so freshness is verifiable rather than promised.
Pricing, integration counts, customer counts, quotas — refreshed routinely and never more than 90 days old.
Funding totals, headcount, certifications — never more than 180 days old.
Founding year, licence, parent company — re-checked the moment something changes, not on a calendar.
Contested or high-stakes claims are adversarially re-checked before publication: an independent pass attempts to refute the claim against primary sources, and claims that cannot be settled are published with attribution or dropped. Every content revision is recorded in an internal audit log with a timestamp and a note of what changed.
Scores and rankings
Each ranking publishes its own methodology: the criteria, their weights, and how positions were assigned. Review scores follow a consistent rubric and are revisited when the underlying facts change materially. Rankings are re-evaluated at least every 90 days. Positions and scores are never for sale, and a low score cannot be paid away; see how funded evaluations work.
Independence and funded evaluations
Automation Atlas is published by ShadowGen, an automation consultancy, and that relationship is disclosed on every page. Thorough coverage takes editorial hours, and we allocate them on our own schedule. Vendors who want broader or earlier coverage than that schedule would reach can fund the additional editorial time through ShadowGen. Funding buys time, never outcomes: the same sourcing, the same dating, the same right to publish unfavourable findings, and no influence on scores or rankings. Nothing on this site is a paid placement, and no commercial relationship changes what a page says.
Corrections
Errors are fixed, not defended. Report one via the submission form with the page URL, the claim, and ideally a source; we re-verify against primary sources and update the page, its dated facts, and any other page carrying the same claim. The revision is recorded in the audit log. Where we got something materially wrong, the corrected claim carries its new verification date.
AI access
We welcome machine readers. AI crawlers are permitted in robots.txt, the full content corpus is exported at /llms-full.txt, a summary index lives at /llms.txt, a machine-readable JSON API is exposed at /api/v1/tools.json, and structured data is embedded on every page. Assistants citing this site get the same dated, sourced facts human readers do, which is the point.