How this directory is built

What qualifies as an SEO tool

The test is not "can this be used for SEO". Almost anything can, and applying that test would eventually admit every spreadsheet, writing assistant and automation platform ever shipped. The test is whether search work is what the product is for.

So a crawler, a rank tracker, a link index, a content optimiser and an AI-visibility monitor all belong here. General analytics, general AI assistants, project management, outreach CRMs sold for sales rather than links, PPC tooling and design software do not — even though several of them appear in a working SEO stack, including Alston's own. Where they matter, they appear in the workflow guides on this site instead of being reclassified as SEO products.

Two categories are deliberately marked as adjacent rather than folded in: App Store Optimization and YouTube SEO. Both are genuine search optimization, for an index that is not the web, and listing them beside a technical web crawler as if they were the same product class would mislead the person comparing them.

The four evidence levels

Every listing states which of these it is, near the top, before any assessment. The distinction is the point of the directory.

Alston tested — 12 listings

Alston used the product himself and the test is written down: the date, the plan, the workflows actually run, and an artefact you can open — an original screenshot from his own account, or a video. A listing cannot carry this label without those, because the schema behind the directory refuses to build the page.

Used in real work — 3 listings

Alston has used the tool in his own SEO work, but there is no formal write-up of a test. The label links to the page where that use is visible. Facts on these listings still come from the vendor's documentation rather than from memory.

Primary-source researched — 87 listings

Desk research from the vendor's own site, pricing page and documentation. Nothing on these pages is written in the first person, there are no test results, and no conclusion is offered that research cannot support. This is a legitimate thing to publish; describing it as a review is not.

Vendor submitted — 0 listings

A vendor told us. Submission is not verification, so these are never indexed and never presented as checked facts until someone has actually checked them.

How facts are verified

Sources are used in this order, and the first one that answers the question wins:

1. The vendor's official product site
2. The vendor's own pricing page
3. Official documentation and help centre
4. Official API and developer documentation
5. Official company, changelog and status pages
6. Official store and extension listings
7. High-quality third-party sources, only where a primary source does not exist or where independent corroboration is the point

Each source stored against a listing records what it verifies — pricing, plans, features, limits, platforms, company or a deal — and the date it was read. That is why the sources section on a listing is a table rather than a row of links: a link with no claim attached to it proves only that the vendor has a website.

Research is done with automated primary-source retrieval rather than from model memory. The scripts that do it live in the repository, take their credentials from the environment, and write a dated dossier per tool that a human reviews before anything is published. Nothing goes from a scrape straight onto the site.

Pricing

A figure is only published with a verification date beside it, and the two cannot be separated: the schema rejects a price without one. Plans you can buy on their own are separated from add-ons, because a $10 reporting module listed beside a $139 tier reads as an entry price, and the "from" figure in the directory's price filter only ever reads standalone tiers.

Where a vendor publishes nothing, the listing says pricing is not published rather than quoting a number found elsewhere. Currently 97 of 102 listings carry a dated pricing check.

Prices change without announcement. Every listing shows how many days old its check is, and says so in public when it is past its review window — 45 days for AI-search products, 90 for commercial SEO software, 180 for genuinely free tools.

Limitations

A published listing must carry at least two real limitations. A profile with none is either unresearched or advertising, and the directory treats both the same way: the page does not build. 102 of 102 listings meet this today by construction, because the ones that did not are still in the queue.

Why there are no ratings

There is no score out of ten, no star rating and no aggregate rating markup anywhere in this directory. The data model has nowhere to put one. A number implies a measurement, and a measurement implies a published method — and the honest version of that method for most software comparisons is "someone's impression, expressed as a decimal". Where judgement is offered, it is written as a sentence with a reason attached, and only on listings where there is first-hand experience to base it on.

How videos are chosen

Alston's own video on a tool takes priority and is badged as his. A video existing is not, on its own, a testing claim — a walkthrough recorded from a trial account is not the same as a documented test, so the evidence label and the video badge are separate things. Where a third-party video is used it is because it demonstrates the workflow rather than because it has the most views, and transcripts are never copied into listing copy.

Alternatives and comparisons

An alternative is only listed with the condition under which it is the better choice. "Related products" lists are filler; "choose this instead when your link index matters more than your reporting" is a decision.

Comparison pages are not generated for every possible pair. A dedicated comparison exists only where there is real demand for it, enough verified data on both sides to fill it honestly, and something to say that the two listings do not already say.

Deals and coupons

An offer is only shown when it is real, has a source, and has been confirmed on a recorded date. Expired offers are not hidden — they are unreachable, because the only function the templates can use to fetch a deal returns nothing once the end date has passed. There are no countdown timers, no invented scarcity, and no empty coupon box on an offer that has no code. "Exclusive" appears only where the offer genuinely is.

Affiliate links

Some outbound links pay a commission. Where one does, the listing says so on the link itself rather than in a footer, and the link carries rel="sponsored".

What a commission cannot do: change a listing's position, add a badge, remove a limitation, or keep a non-paying competitor out of the alternatives. Every listing in a category is ordered by the same rules whether it pays or not, and the alternatives section names the best option for the reader's situation regardless of who pays. Trust is the asset; commission is a by-product of it.

Vendor submissions

Anyone can suggest a tool. A submission buys inclusion in the research queue and nothing else — not a listing, not a recommendation, not a placement, not a tested badge. Vendor claims are verified against the vendor's own published documentation before they become facts on a page. If paid placement is ever introduced, it will be labelled as such and kept out of editorial ordering.

Corrections, rebrands and dead products

If something here is wrong, say so and it gets fixed with a dated note in the listing's change log. Rebrands are normalised to the current product name and keep the former name searchable, so looking for DeepCrawl finds Lumar rather than nothing. Discontinued products are marked as discontinued rather than deleted — a reader searching for a tool that no longer exists deserves to be told that, not to hit a 404.

What gets indexed

A listing is only allowed into search results when it has a verified official URL, a useful description, at least three features with their SEO use case, at least two limitations, at least two use cases, an alternative, a primary source and a pricing position. That check runs on the record itself, so nobody can mark a thin page ready: 102 of 102 published listings currently pass it.

Category pages follow the same logic. They are built as soon as they hold one listing, because the links are useful, but they are kept out of search results until they hold at least 3 and carry a written editorial layer. Filter combinations and search results are never indexed at all.