Transparency · Methodology

How we review AI tools

Every score on this site comes from real, paid usage. Here’s exactly how we test, what we measure, and how we handle affiliate relationships.

Smart Tools Pick exists because most AI tool reviews are written by people who spent 20 minutes on a free trial and then repeated the vendor’s marketing copy back at you. We do it differently. This page explains our process in full — not because we have to, but because you deserve to know how the sausage is made before you trust a recommendation that might cost you $100/month.

Our four principles

🧪

Real usage, real work

We only review tools we’ve used on actual client projects or our own business operations — not demo workspaces or sandboxes provided by vendors.

💳

Paid plans, always

We pay for the plans we review. We do not accept free “review accounts” from vendors because they often unlock features or remove limits that paying customers face.

📏

Specific, not vague

Scores are based on measured outcomes — time saved, tasks completed, error rates — not impressions. If we can’t quantify it, we say so explicitly.

🔄

Updated when things change

AI tools update fast. Every review displays a “Last tested” date. We revisit top-rated tools every 6–12 months and update scores when the product changes materially.

Our testing process

Every review follows the same six-step process. There are no shortcuts — if a step isn’t completed, the review doesn’t get published.

Purchase the plan

We buy the plan most relevant to freelancers — usually the entry-level paid tier, since that’s what most readers will consider. The exact plan and monthly cost are always disclosed at the top of the review.

Define the test scenario

Before opening the tool, we write down the specific job we want it to do. This prevents us from cherry-picking tasks the tool happens to be good at.

Use it on real work for a minimum of 2 weeks

Most tools shine in the first hour and reveal their friction points over weeks. We track time spent, errors encountered, and workarounds required in a private log throughout the testing period.

Stress-test the edges

We deliberately test the scenarios where tools tend to break — large files, messy data, edge-case inputs, high usage volumes. If the tool has a usage limit, we test what happens when you approach it.

Take our own screenshots

Every screenshot in our reviews is taken from our own account, showing our real data. We never use vendor-provided press images.

Score against the rubric

After testing, we score the tool against our standard rubric (see below). Scores are set before writing the review text — not adjusted to match a narrative.

How scores are calculated

Our overall score is a weighted average across five dimensions. The weights reflect what matters most to freelancers making real purchasing decisions.

Dimension	Weight	What we measure
Core functionality	30%	Does the tool do its primary job reliably? Accuracy, speed, and consistency on real tasks.
Value for money	25%	Is the price justified by the outcomes? Compared against free alternatives and direct competitors.
Ease of use	20%	Time to first useful result. How much setup, learning, or workaround is required?
Reliability	15%	Uptime, error rates, and how gracefully the tool handles failures during our testing period.
Support & docs	10%	Quality of documentation, response time on support tickets we actually submitted, and community resources.

Score interpretation: 9–10 = exceptional, use without hesitation. 7–8 = solid, recommended with minor caveats. 5–6 = situational, only buy if it fits your exact use case. Below 5 = skip it, better alternatives exist.

What we don’t do

✗

We don’t accept free review accounts
Vendors sometimes offer upgraded accounts to reviewers. We decline. Free accounts can have artificially removed limits or added features that paying customers never see.

✗

We don’t accept payment for positive coverage
Affiliate commissions are earned only when a reader purchases — not in exchange for a specific score or recommendation. No vendor can pay to influence a review.

✗

We don’t publish reviews without hands-on testing
If we haven’t used it ourselves on real work, we don’t review it. We will not publish a review based on reading other reviews, vendor documentation, or press releases alone.

✗

We don’t hide negative findings to protect affiliate relationships
If a tool we earn commissions from has a significant flaw, we say so — with specific evidence. A review that only shows the upside is a paid ad, not a review.

💰 Our affiliate policy

Some links on this site are affiliate links. When you click one and make a purchase, we earn a commission — at no extra cost to you. This is how Smart Tools Pick is funded.

A few things we want you to know:

Affiliate relationships do not influence scores or recommendations. We review tools we believe are genuinely useful first, then check whether affiliate programs exist — not the other way around.
We link to non-affiliate alternatives in comparison tables when they’re the better choice for a given use case.
Every article that contains affiliate links displays a disclosure at the top of the page, above the fold.
We are compliant with FTC guidelines on affiliate disclosure.

Corrections and updates

AI tools change fast. Pricing updates, features get added or removed, and quality can shift significantly between versions. We handle this in two ways:

Minor updates (pricing changes, small feature additions) are made silently with the “Last tested” date updated.
Major updates (significant score changes, product pivots, tool shutdowns) are noted with an “Updated [date]” notice at the top of the article explaining what changed and why.

If you spot something that’s out of date or factually incorrect, please contact us. We take corrections seriously and respond to every report within 48 hours.

Questions about our methodology? Something doesn’t add up? We’d rather you push back than silently distrust us. Get in touch →

This methodology page was last updated March 2026. It applies to all reviews published on Smart Tools Pick.