Most "best SEO tools" articles are a feature table someone copied from a pricing page. Nobody logged in. Nobody exported a report. Nobody checked whether the rank tracker actually agrees with what Search Console says.
We do it differently, and this page is the whole protocol — so you can decide whether our scores deserve your trust.
The three test sites
Every tool is pointed at the same three properties, which we own and have full analytics access to:
- A 4-year-old content site in a competitive niche (~90k monthly sessions). This is where crawl depth and keyword coverage get stressed.
- A six-month-old site with under 40 pages. New domains expose how a tool handles thin data — some simply refuse to report.
- A local services site with a Google Business Profile, for the local-pack and citation features.
Using our own sites matters. It means we can compare a tool's traffic estimate against the real number in Google Analytics rather than against another tool's estimate.
The seven tasks
Each tool runs the same tasks, timed, by the same person.
1. Keyword research from a cold start
Given one seed term, how quickly can we build a 50-keyword cluster with volume, difficulty, and intent? We record the time and the number of clicks.
2. Rank tracking accuracy
We track 25 keywords we already know the position of from Search Console, then compare. A tool reporting position 4 for something sitting at 11 loses points here, and no amount of interface polish earns them back.
3. Site audit depth
We introduce five known technical faults — a broken canonical, an orphaned page, a redirect chain, a missing hreflang return tag, and a bloated LCP image — then see how many the crawler reports without being told where to look.
4. Backlink index freshness
We check for links we placed ourselves in the previous 30 days. Index size claims are marketing; recency is what you actually feel.
5. Competitor analysis
Given one competitor domain, how long to a defensible content-gap list? This is where the difference between a database and a product shows up.
6. Reporting and export
Can a non-specialist read the output? Is there a scheduled PDF or a live dashboard a client would accept? Does CSV export preserve the data or mangle it?
7. Support response
We open one genuine support ticket per tool and record the time to a useful answer — not the time to an autoresponder.
How the score is built
The published score is a weighted average, not a vibe:
| Dimension | Weight | What moves it |
|---|---|---|
| Data accuracy | 30% | Rank and traffic figures against our own analytics |
| Feature depth | 20% | Coverage of the seven tasks without a second tool |
| Ease of use | 20% | Clicks and elapsed time on the timed tasks |
| Value | 20% | Capability per dollar at the plan most readers buy |
| Support | 10% | Time to a useful answer |
A tool cannot buy a higher score, and no vendor sees a review before it publishes.
What we deliberately do not do
- We do not score on feature count. A tool that does five things well beats one that does twenty badly.
- We do not test on demo accounts. Vendor sandboxes are curated. We pay for our own subscriptions at the tier we recommend.
- We do not refresh scores silently. When a re-test changes a score, the date and the reason both change with it.
When we re-test
Every tool is re-tested annually, and immediately after any pricing change or major release. The "last tested" date on each listing is the real date somebody ran the seven tasks — not the date the page was edited.
If you think we have a tool wrong, tell us. We publish corrections.
