Editorial Policy

Corrections

Mistakes here are inevitable. Pricing pages change without notice, vendors rename plans, and a spec typed into a table can be typed wrong.

The policy:

• Report it to [email protected]. Include the page and what’s wrong. A link to the vendor’s own page settles most disputes in one exchange.
• Factual errors are fixed on sight. Wrong price, wrong limit, wrong plan name, wrong date. Target turnaround is 72 hours from a verifiable report.
• Corrections that change a score or ranking are logged in public. A dated correction note goes at the foot of the affected page saying what was wrong, what it is now, and whether the SquadScore™moved. Silent edits to a ranked list are how review sites lose the plot, so we don’t do them.
• Small fixes are stamped, not annotated.A typo or a broken link updates the page’s review date without a note. Anything a reader could have acted on gets the note.
• We don’t unpublish to fix things. Pages get corrected in place, with the history visible. Deleting a page to make an error disappear is not a correction.

Independence

Right now this is simple, and the simplicity is temporary, so both states get described.

Today, August 2026: the site earns nothing. No affiliate links, no buy buttons, no display advertising, no sponsored content, no paid placement, no vendor sponsorship. Nobody has paid us anything, and nothing on the site is monetised.

If that changes, these constraints come with it, and they’re binding:

• Monetisation never touches SquadScore™. The formula, weights and bounds are computed from specs. No editorial override, no manual adjustment, no exception for a partner.
• Any commercial relationship is disclosed on the page it affects, not only in a policy document you’d have to go looking for.
• No paid placement in a ranked list. Ever. If a slot in a “best of” list is ever purchasable, this site has failed at its only real job.
• Sponsored content, if it ever exists, gets labelled as sponsored at the top of the page and is excluded from rankings and scores.

Robert MacKelfresh runs other businesses that use software covered here as ordinary customers. Where a tool under review is one those businesses pay for or hold any interest in beyond a normal subscription, the post says so.

Vendor relationships

Vendors are welcome to get in touch. Here’s the shape of it.

What we accept: corrections with evidence, spec clarifications, product documentation, briefings, and access to a product for the purpose of building the artifact in a post.

What we don’t accept: payment for coverage, payment for placement, payment for a better score, review copy conditional on approval, or embargo terms that require us to submit anything for sign-off before publication.

Pre-publication review: never for the article. Vendors may be asked to check a spec table for factual accuracy, and their correction is a data point rather than an instruction. Assessments, criticism and rankings are not shown to vendors in advance.

Free access and trial extensionsget disclosed on the page where they were used. They don’t buy anything beyond the ability to see the product.

Vendor pressure gets published. If a company attempts to condition access on favourable coverage, that attempt becomes part of the coverage.

AI disclosure

This site is about AI tools. Pretending we don’t use them would be absurd, so here’s the exact division.

Data tables and scores are programmatic output. Spec tables, SquadScore™ scores, ranked lists, comparison grids and filter results are generated by code from a structured dataset. The formula is published on the methodology page. No language model decides a ranking, and no ranking is written by hand.

Prose is human-edited, and a named human is accountable for it. Drafting assistance is used. Every assessment, recommendation, criticism and judgement call on this site has been read, checked and signed off by Robert MacKelfresh, who is accountable for the result regardless of how the first draft was produced.

What never gets generated: a spec, a price, a benchmark, a certification status or a quote. Facts come from vendor-published sources or they get recorded as unknown. A model that fills in a plausible price is the single most dangerous thing you can point at a dataset like this one, and it is never allowed near ours.

No fabricated first-hand testing.You won’t read “when I tested” on this site describing something that didn’t happen. Where a claim rests on general use rather than a controlled test, the wording says so, and how we test sets out the limits in detail.

Last reviewed August 2026. Material changes to this policy are dated here.