Editorial Policy

At aicustomerservice.cc we publish reviews of AI customer service software that a buyer can act on and check. This policy sets out how those reviews are made.

How We Test

Every tool goes through the same five steps.

1. We run real queries through a test environment. A dummy Shopify account, connected to the tool, with real support questions put through it — order status, a refund request, a policy question, an angry customer, and something deliberately outside what the tool is meant to handle. What we watch for is what it does at the edge: whether it escalates, guesses, or makes something up.

2. We read the fine print. The terms, the billing definitions, the data processing agreement, and the FAQ at the bottom of the pricing page — which is usually where the exclusions live. Vendors are consistently clearer in their documentation than on the page selling the product.

3. We work out the price from every mechanism it is sold through. Per seat, per resolution, per conversation, per minute, per credit, or some combination. We read both billing toggles, because a monthly rate is often 20 to 40% above the annual one and only one of them appears by default. Where a figure only exists behind a product dropdown or a currency selector, we open it.

4. We read everything published about the vendor. Product documentation, release notes, trust and security pages, funding and acquisition announcements. Four to six of the vendor’s own pages is the working minimum, and a single fetched page is not a read.

5. Then we write it, and date it. Every figure carries the day it was read and links the page it came from. A price without a date is a memory, not a fact.

We do not claim this is an exhaustive benchmark. We are not running thousands of tickets through every platform or publishing a performance league table, and where a tool is demo-gated we cannot get in at all — in which case the review says sales call required, no public pricing, because that is itself the finding. What this site is good at is sourcing and verification: reading the primary document, quoting it verbatim, dating it, and correcting it in public when it moves.

Rating System

Tools are rated on a 1–5 scale based on weighted criteria: capability (28%), pricing transparency (22%), cost predictability (20%), independence (15%), and independent evidence (15%).

Sub-scores are shown on every tool page, so a score can be argued with rather than taken on trust. Nothing reaches five — the scale is capped at 4.6 to keep headroom. Twelve platforms carry no score at all, because there is not enough public material to build one, and an unrated tool is marked unrated rather than given a low number.

Content Updates

All reviews are updated at least quarterly, and any pricing figure older than ninety days is treated as stale. When a tool ships a major update or changes its pricing, we re-evaluate the review. The full list is checked for acquisitions monthly, because ownership changes what a product is and vendors rarely mention it on the pages buyers read.

Each review displays the date its pricing was verified, so readers know how current the information is. When a re-read changes a number, we publish the change rather than quietly swapping it in.

Corrections

If you find an error in any of our reviews, please email corrections@aicustomerservice.cc. We investigate all reports and update our content promptly when errors are confirmed. We take no money from vendors, so there is nothing to lose by telling us we are wrong.

We correct in public and date it. Where a published count has been wrong, the page states the old number, the new one, and why the first was wrong. An earlier count of vendor-authored search results said roughly 40 of 48; adding up the published rows gave 35 of 47, so the study says 35 of 47 and explains the error.

The Blog

The blog is a personal space rather than a product surface. It carries writing on customer support generally — how teams are run, what the metrics actually mean, what goes wrong and why — and not everything there is about a tool. Where a post does make a claim about a platform, it is held to the same sourcing and dating standard as a review.

Affiliate Disclosure

Some links on this site can be affiliate links. This means we may earn a commission if you purchase a tool through our link. Affiliate relationships never influence our ratings, rankings, or editorial content. Tools with and without affiliate programs are evaluated identically.

Nobody can pay to appear here, to move up, or to soften a review. We accept no sponsored posts, no vendor-supplied copy, and we do not send drafts to vendors for approval before publishing. Several tool pages advise against purchase, and that is the point of the arrangement.

Who Writes It

AR. I read the fine print. Vendor pricing pages, billing definitions, terms, funding filings and acquisition notices — then I do the arithmetic nobody publishes: what a platform actually costs at your volume, what its headline metric is really counting, and who owns it now. I do not run benchmarks, and no page here pretends otherwise.