DOC 002 · METHOD
How we evaluate AI email tools
This site covers two layers of the same job: the AI writing assistants that draft or coach an email, and the sending tools that deliver it. The method below applies to both. It is deliberately modest, because the editor reads pages and documentation. The editor does not run vendor accounts or campaigns.
What we actually do
For every tool we cover, the editor opens the vendor's own pricing page, product page and help documentation on a given date, and writes down what they say. Prices, plan names, usage limits and feature claims come from those pages and nowhere else. We do not use review sites, aggregator listings, or a vendor's sales deck as a source for numbers. When the vendor does not publish a number, the page says the price is on request and links to the vendor.
Each page shows the date its prices were read, and links the exact vendor page the numbers came from, so you can check the same page yourself. Where a pricing page was captured as a screenshot, the caption gives the capture date and the URL. We never edit or crop a screenshot.
What we do not do
- We do not test tools. No page here claims that we sent a campaign, ran an A/B test, or measured a reply rate. If a page describes how a feature behaves, it is describing the vendor's documentation, and it says so.
- We do not publish star ratings, scores or reviewer quotes. A score would imply a measurement we have not made.
- We do not estimate prices. If a plan is quote-based, we say that rather than guessing a figure.
- We do not keep a vendor in a ranking because it pays. Position comes from the reasons written next to it.
How a pick is made
A "best pick" on this site is an argument for one reader in one situation, and the reason is written next to it. For writing assistants the questions are practical. Where does the tool live, in the inbox or in a separate app? What does it cost for one person versus a team? Does it write from a prompt, coach a draft you already wrote, or both? What happens to your drafts when you stop paying? For sending tools the questions are about volume: what a plan costs at 10,000, 50,000 and 100,000 emails, what the free tier really allows, and which limits (daily caps, contacts, seats) bite first.
We pick the option that fits the most common reader of that page and name the readers it does not fit. If two tools are close, the page says so and gives the deciding condition, such as "if you live in Outlook, choose the one with the Outlook add-in".
Corrections
Vendors change prices without notice, and an AI product's plan names change faster than most. If you find a figure that no longer matches the vendor's page, send us the page path and the vendor URL that shows the new number. We check it against the vendor page, fix the page in place, and update the date shown. A correction never waits for a quarterly review.
Where a vendor has been acquired or renamed, the page says so. Flowrite, for example, now redirects to MailMaestro, and our Flowrite coverage is written around that fact rather than around the old product.
Who does the work
One editor, working under the pen name Tessa Brightwell, does the reading and writing. See About for what that does and does not mean. For the pages themselves, start at the AI email writing assistants comparison or the sending-tool pricing index.