How We Score AI Tools
Last updated: September 2026
Every active tool is re-verified at least monthly against the vendor's own site: pricing, free tier status, and whether the product still operates. Each tool page displays its "Last verified" date, discontinued products carry a warning banner instead of being silently deleted, and the monthly changelog records what every sweep found. Between mid-June and mid-August 2026 those sweeps caught 15 tracked products that shut down, rebranded, or were absorbed.
Almost every tool on ToolChase carries a ToolChase Score, a weighted editorial evaluation across 8 parameters. Where we have not finished testing a product's core claims, the page shows Score pending instead of a number we cannot stand behind. This page explains four things: how a score is produced, how a score becomes a position in a ranked guide, where money touches any of this, and what the ranked set is actually drawn from. The last one is the part most directories leave out.
The ToolChase Score
The ToolChase Score is not a crowd-sourced average. It is an editorial score based on hands-on evaluation. We test each tool against its category use cases, score it on a 0-100 internal scale across the 8 parameters below, then convert that total to a 5-point display rating.
The score is produced before any commercial conversation, and independently of one. An editor tests the product, fills in the parameter sheet, and the total falls out of the weights. A score is finalized and dated before any commercial discussion about that tool, and a published score changes only when the product is retested. When a partner's score moves, it is because we retested the product or the product changed, and the "Last verified" date on that tool page moves with it.
External reviews from platforms like G2, Capterra, and TrustRadius are shown separately and clearly labeled with their source and the date we last verified them.
Scoring Parameters
| Parameter | Weight | What We Evaluate |
|---|---|---|
| Core Product Quality | 25% | How well the tool performs its primary job. Tested against category-specific use cases with benchmark prompts and side-by-side comparisons. |
| Ease of Use / UX | 15% | Onboarding speed, setup friction, navigation clarity, learning curve, and time-to-value. |
| Value for Money | 15% | Free plan quality, starter pricing, feature limits, upgrade logic, and pricing relative to competitors. |
| Feature Depth | 15% | Breadth and quality of core + advanced features. Scored against a category-specific feature checklist. |
| Reliability / Performance | 10% | Speed, stability, output consistency, rate limits, and historical uptime. |
| Integrations / Ecosystem | 8% | Native integrations, API quality, export/import options, marketplace, and workflow compatibility. |
| Market Trust / Reputation | 7% | Third-party review sentiment, brand maturity, update cadence, and community adoption signals. |
| Support / Docs / Governance | 5% | Documentation depth, help center quality, support responsiveness, security/privacy policies, and admin controls. |
Category-Specific Weighting
Not all tools are scored the same way. We adjust parameter weights by category. For example, Developer Tools weight Reliability (15%) and Integrations (12%) higher than average, while AI Image Generators weight Core Product Quality (32%) much higher because output quality is the primary differentiator.
Score Scale
Each parameter is scored 1-5 internally, then weighted and summed to a 0-100 total. The display conversion is:
90-100 → 4.8-5.0 | 80-89 → 4.3-4.7 | 70-79 → 3.8-4.2 | 60-69 → 3.3-3.7 | Below 60 → under 3.3
How a position in a ranked guide is decided
A numbered guide is not the score column sorted downward. Position is set by the score and by how closely the tool fits the specific job named in the headline. A tool scoring 4.6 for general use can sit below a 4.4 built precisely for the use case in question, and the entry will say why.
A commercial partner can hold any position in a ranked guide, including first, when its score and its fit put it there. We do not demote partners to look independent, and we do not promote them to be agreeable. Several tools ranked well here before their vendors ever contacted us, and their positions did not move afterward in either direction.
Three kinds of entry, and how to tell them apart
The label at the entry tells you which one you are reading, in visible text, without leaving the page.
A ranked entry with no commercial relationship
Carries a ToolChase Score and holds its number on editorial grounds. There is nothing to disclose because nothing was paid or earned on it. Most of the directory is this.
A ranked entry from a commercial partner
Also carries a ToolChase Score and also holds its number on editorial grounds, and additionally carries a disclosure at the entry naming the relationship and stating that the ranking is editorial. The live wording reads: "Semrush partners with ToolChase. Its ranking here is editorial, scored on our 8-parameter framework, and the partnership does not change it." A partner in this shape can hold any position, first included.
A sponsored placement we did not test
Carries a Sponsored label and sits outside the numbering as its own block, because numbered positions are earned on the framework and are not sold at any price. Where we have not tested the product it carries no ToolChase Score, the entry says so in plain words, and the write-up is marked a vendor-informed summary rather than a tested verdict.
So the test is simple. A number in the score column means we tested it and the number is ours. A score plus a disclosure means both things are true at once: we tested it, and we have a commercial relationship that did not move the result. A Sponsored block sitting outside the numbering, with no score, means we sold that unit and did not test the product.
What the ranked set is drawn from
A ranked guide on ToolChase ranks the tools we have reviewed. It is not a survey of the market.
The directory holds 780+ reviewed tools. Every one of them was reviewed at no charge to its vendor, every one keeps its page, and every one stays eligible for any guide its score and fit qualify it for, at no charge and with no expiry. A tool's ranked position has never depended on whether its vendor has paid us anything.
New tools reach the review queue two ways. We choose them, on merit and on search demand, and no money changes hands. Or the vendor buys review access, which is set out on our Advertise page, which buys a place in the queue and an editor's time, and buys nothing else: not a score, not a verdict, and not a ranked position. A separate paid tier buys a labeled sponsored entry on a named page, which is disclosed as sponsored wherever it appears and is never a ranked position bought with money.
The consequence is worth stating plainly, because you should price it into how you read us. Over time the pool any ranked guide draws from is made up of tools we chose ourselves plus tools whose vendors paid to be looked at, which means a genuinely good tool can be missing from a guide for no better reason than that nobody here has tested it yet.
What payment cannot buy
- A score. No payment has ever produced a score, raised one, or reopened a finished one.
- A ranked position. A ranked entry sits where its score and its fit put it. A sponsored slot is bought and is labeled as bought, which is the opposite of a ranked position being for sale.
- A verdict. We write the "best for" and the "not for" ourselves, including on partner tools.
- A deleted criticism. A limitation that was true when we tested it stays on the page until the product changes. When it changes, we say the product changed, not that we changed our mind.
- Silence. No partner has the option of an unlabeled placement. If a relationship exists, it appears at the entry.
External Reviews
When we display review counts or external ratings, we always cite the source (G2, Capterra, TrustRadius, Product Hunt, etc.) and the date we last verified the data. We never combine external reviews with the ToolChase Score, they are separate signals.
Pricing Verification
All pricing data is verified directly on the vendor's website. Each tool page shows a "Last verified" date. If you spot incorrect pricing, please let us know and we will update it within 48 hours.
What to do with this
Read the entry, not the headline. If there is a score beside a tool, we tested it and the position is our judgment. If there is a Sponsored label where the score should be, we sold that slot and you should weigh the write-up accordingly.
Questions about our methodology? Contact us.