Editorial

Editorial standards

This category produced a federal enforcement action over inflated claims. We take the reliability of what we publish seriously as a result.

Our rules

1. We do not estimate figures

   Every price, limit and certification comes from the vendor's own published documentation. Where a vendor does not publish something, we write "not disclosed" rather than guessing. A missing number is scored, not invented.

2. Every factual claim is dated

   Pricing in this market changes monthly. Each review lists its sources with the date we read them, so you can judge how stale a figure might be.

3. Regulatory claims cite the regulator

   When we report enforcement action, we link the primary filing — not a secondary article about it. Our Air AI entry cites the FTC case file and press release directly.

4. We do not repeat vendor performance claims as facts

   Marketing figures like "4x more answers" or "300% more contacts" are attributed as vendor claims, never presented as measurements. We do not currently run independent latency or connect-rate benchmarks, and we say so on our methodology page rather than implying we do.

5. We publish unfavourable findings about big names

   Several heavily-marketed platforms score poorly here, largely for refusing to publish a price. We rank them where the rubric puts them regardless of commercial relationship.

6. We state what we do not know

   Gaps in our own method are published alongside the method. If we have not tested something, we will not imply that we have.

Corrections policy

If we get something wrong we correct it on the page and note what changed. Vendors are welcome to flag inaccuracies. Being a customer or partner does not accelerate a correction, and not being one does not delay it.