Why this page exists
Most comparison sites will not tell you how their rankings work, because the honest answer is often 'the order was negotiated.' ATIV publishes its method because our scores are the main thing we sell you — not literally, you pay nothing, but they are the reason to trust a list here over a list assembled from vendor landing pages. This page explains what a score means, what it ignores, and where the whole exercise can mislead you. If you finish it and decide our weights are wrong for your work, the per-dimension notes in each article should still let you re-rank the tools yourself.
One more thing before the rubric: a score is a summary, not a measurement. There is no laboratory in which 'creator fit' registers 8.6. Scores exist so that a reader skimming five tools can see our relative judgment at a glance — and so that we are forced to commit. Two tools a full point apart are meaningfully different in our estimation. Two tools 0.2 apart are, for most people, a coin toss decided by taste.
The four things we score
Every tool we rank gets assessed on four dimensions, each from 1 to 10. The headline score in our tables is a weighted blend, but the dimensions matter more than the blend, so here is what each one actually looks at when we evaluate a tool for a specific roundup.
Ease (30%). How long does a competent new user need to reach a first good result? Not a first result — a good one. We count the friction honestly: mandatory onboarding tours, concepts you must learn before the tool does anything (operations, nodes, blocks), and how understandable the failure states are. A tool can be extremely powerful and still score 6 here; that is not an insult, it is information for the reader deciding whether they have a free weekend.
Pricing clarity (25%). This dimension grades the pricing, not the price. A $40/month tool can score 9 if the tiers are legible, the limits are stated in plain units, and a busy month cannot ambush you. A $5/month tool scores badly if the real constraints hide in credit systems, expiring balances, or per-seat fine print. We check the pricing page against the billing reality described in each vendor's documentation, and we note where 'unlimited' has an asterisk doing heavy lifting.
Creator fit (30%). Does this tool solve problems creators actually have, in the shape they actually have them? For a video tool that means vertical formats, captions, and export presets for the platforms people publish on — not a generic 'AI features' checklist. For a writing tool it means it can hold a voice, a backlog, and a publishing cadence. Generic excellence loses to specific usefulness here, which is why a famous tool sometimes ranks below a narrower rival in a roundup aimed at a particular job.
Export options (15%). How easily does your work leave? Files in standard formats, data export that includes the useful parts, API access that is not gated behind the most expensive tier. This is the smallest weight and the one readers thank us for later. Tools that hold your archive hostage — no bulk export, proprietary-only formats — take a real deduction, because switching costs are a feature of the product whether vendors list them or not.
How the numbers map to words
Scores from 9.0 up mean a tool leads its category on most dimensions and has no disqualifying weakness for the reader the article addresses; expect only a handful per year. The 8s are where most good tools honestly live — strong, with a specific trade-off we name in the text. A score in the 7s means 'right for a narrower reader,' and the article should make that reader easy to identify. Below 7, a tool usually does not appear in a roundup at all unless its one exceptional use case matters enough to document the rest.
You will notice scores are never all bunched at the top of a single table. If we cannot separate five tools by at least a point and a half across the list, the roundup is not finished — either the picks are too similar or we have been too polite in the per-dimension notes. Disagreement with our ordering is expected and fine; disagreement you cannot reconstruct from our stated weights is a bug, and the corrections route below exists for it.
Money, independence, and what we will not do
ATIV currently earns no affiliate commissions — no link on this site pays us anything today. These rules are published in advance so they are on record before any money exists: affiliate income would never change a score, an ordering, or an inclusion decision, and every article states the relationship near the top in plain language. Where it gets subtle is selection pressure — it is easier to write about tools with affiliate programs — so we keep a standing rule: every roundup must seriously consider the best non-paying option we know of, and as things stand, every tool in every roundup pays us nothing. If a vendor offered payment for placement, the answer would be no, and the offer itself would be worth a disclosure note in the affected article.
We also do not publish categories where a wrong pick can seriously harm a reader. At launch that means no finance, medical, or legal recommendation roundups. The stakes change the standard of evidence, and we would rather be a small site that is right about video editors than a broad one that is casual about chemotherapy or credit cards.
How articles are made and maintained
Each roundup starts from the reader's job, not the vendor list: we define who the article is for and the handful of tasks it must cover, then longlist tools, then cut. Evaluation combines structured checks against the four dimensions, the vendor's own documentation and pricing pages, published changelogs, and the accumulated user reports in support forums and communities — the places where a tool's real failure modes are described by the people paying for it. Where a free tier or trial exists, we weight what that free version actually includes — its limits, watermarks, and export options — because that is the version most readers will meet first.
Software changes under its own articles. We re-check pricing and plan details on a rolling basis, update the date shown on each page when a change is material, and retire picks that have decayed rather than quietly leaving a stale recommendation in place. When we spot an error — a price that moved, a feature that shipped or was removed — we fix the page and change its date; we would rather fix a page in public than defend it in private.