How We Review AI Tools (and What Our Scores Do and Don't Mean)
Exactly how AIProsNCons builds a review and a Trust Score, what is estimated rather than measured, when a page is labelled hands-on tested, and how we make money.
Most AI-tool sites show you a score and a verdict and never say where either came from. This page explains exactly how our reviews and scores are built, including the parts that are estimates, so you can decide how much weight to give them.
What a review on AIProsNCons is
Every listing is a structured editorial record, not a single blob of opinion. For each tool we publish:
- A plain-language description of what it does and who makes it.
- The main pros and cons.
- Who it is best for and who should avoid it.
- A short verdict in three parts: use it if, maybe if, skip it if.
- The pricing model and the price text shown on the official site.
- An editorial rating out of 5.
Those records are written from the vendor's public product information: its website, pricing page and documentation. We use AI tools to help draft and structure listings, and editors shape the result. We say so because you should know how the content is made.
What we do not claim
We do not hands-on test every tool we list, and we do not want to pretend otherwise. With more than a thousand listings, a "tested" label on all of them would be impossible to back up.
A page only says Hands-on tested when the testing record is published on that page: who ran the test, the date, the specific task, and what actually happened. Every other page says Editorial review, which means it is based on public product information rather than our own testing.
How the Trust Score is calculated
The Trust Score is a number out of 10. It is the average of four factors. Here is how each one is set:
| Factor | How it is set |
|---|---|
| Accuracy | Estimated from our editorial rating: rating × 2 (kept between 5 and 9.8). |
| Ease of use | Estimated from our editorial rating: rating × 2 plus 0.3 (kept between 5 and 9.8). |
| Value for money | Set by pricing model: free 9.5, freemium 8.0, paid 6.5, plus 0.6 for paid tools that advertise a free entry point. |
| Privacy | A flat default of 6.5 unless we have assessed that tool individually. Pages label it "default" when we haven't. |
Two things follow from this. First, accuracy and ease of use are estimates, not benchmark results, because they are derived from the editorial rating. Second, the privacy bar is only meaningful on pages where it is not marked "default". For anything where privacy matters to you, read our privacy and terms checklist and check the vendor's own policy.
The score is a consistent way to compare listings quickly. It is not a lab result, and it should never replace testing a tool on your own work.
How we choose what to list
We list established and emerging AI tools across writing, image generation, video, audio, coding, chatbots, marketing, productivity, design and business automation. We check that the product exists, that its website loads, and that we are not listing the same product twice under two names. When we find a duplicate, we merge the two pages and redirect the old address to the one we keep.
How we make money
Some outbound links on this site are affiliate links. If you sign up through one, we may earn a commission at no extra cost to you. A partnership does not buy a better rating, a higher placement or a softer review. The full wording is on our affiliate disclosure page.
Corrections
AI products change quickly: prices move, features are added, plans are renamed. If something on a page is wrong or out of date, tell us through the contact page and we will fix it. We only show a "pricing last checked" date on a page when someone has actually checked it.
Where to go next
- Use the AI Tool Finder to get matched with tools for a specific task.
- Put two or three options side by side in the comparison tool.
- Learn how to test an AI tool before you pay.
