Directory backlinks: what we can prove, and what we cannot

Every directory list on the internet, including our own, is sold on one implied promise: a listing gives you a backlink, and a backlink helps you rank. Most of those lists then decorate each entry with a “DR 72” badge and a green “dofollow” tag, as if those were properties of the directory. They are not. They are observations somebody made about one page, on one day, with one tool — or, quite often, observations nobody made at all.

This article explains the method the toolkit ships in catalog/PRIORITY.md for deciding whether a directory is worth a submission, in my own words. It is a method for collecting evidence, not for producing badges. At the end it should be clear why the best-of lists on this site carry no “free”, “dofollow” or domain-rating claims, and what you would have to do yourself to earn the right to make one.

The question that actually matters

“Does this directory give a backlink?” is the wrong question, because it has no single answer. A directory is not one page. It is a template that renders hundreds or thousands of listing pages, each of which may or may not be indexed by Google, and each of which contains a product link that may or may not be one a crawler follows and counts.

The question that can be answered is narrower: for a listing like mine — same category, same listing template — is the page findable in Google, and is the product link on it a normal crawlable anchor? That is a question about a sample, not a directory, and it is answered with three checks that take about five minutes per platform.

Check 1: is the listing page in Google?

Pick three public listings on the directory that look like yours: same category, same listing template, ideally one established product and two recent ones. Record the exact URLs and the date.

For each URL run a Google search with the site: operator on the exact page, and a second search for the product name plus the directory name. Record the query, the date, the URL Google actually returned, and whether it was the exact listing page.

Then apply the one rule that most backlink lists get wrong: a positive result is evidence; a missing result is unknown. Google does not return every indexed URL for a site: query, the operator is a filter rather than an index dump, and a page can be indexed today and absent from the result tomorrow. If you cannot find the page, you have learned nothing about whether Google has it. You have certainly not learned that the directory is “not indexed”. Only Search Console’s URL Inspection is authoritative, and only for properties you own — which a directory is not.

Two other traps. A result from a different search engine says nothing about Google. And a result for the directory’s homepage says nothing about the listing pages; homepages are almost always indexed, listing templates often are not.

Check 2: could Google index it at all?

Indexing is a decision Google makes; eligibility is something you can inspect. For the same three sample pages, look at:

• HTTP status and redirect chain. A listing that 302s to a login page or a category hub is not the page you think it is.
• Response headers, specifically X-Robots-Tag. A noindex there never shows up in the HTML.
• robots.txt for the Googlebot user agent. Disallow rules block crawling; they do not, by themselves, remove a page from the index, which is why robots disallow and noindex are two different mechanisms and must be recorded separately.
• Meta robots / googlebot tags in the raw HTML and in the rendered DOM. Directories built on JavaScript frameworks sometimes inject a noindex on the client, or remove one.
• The canonical. If the listing page canonicalizes to a different URL — a category page, a bare domain, the product’s own website — the page you are looking at is telling Google to credit something else.

An accessible page with a 200 status, no robots directive and a self-canonical is eligible. That is not the same as indexed; it means Check 1 could in principle succeed. A CAPTCHA or a browser challenge when you fetch the page is unknown, not proof that Googlebot is blocked — bot walls often treat Google differently from you.

Check 3: is the product link a link?

Find the actual link to the product’s website on the listing page. Not the social icons, not the navigation, the one that points at the product domain.

Record the element, its href, its rel tokens and where it finally resolves. Directories love to route outbound links through a redirect (/go/123, /out?url=), and that route decides everything: a redirect that returns a normal 301 to the product is a crawlable link; one that requires JavaScript, an onclick handler or a javascript: href is not something you should count on.

The toolkit includes a small console snippet that lists candidate anchors on a public listing. It only surfaces candidates; it does not follow redirects, collect headers or decide anything. You do that by hand. A rendered <a href> can be crawlable even if JavaScript inserted it, and a button with a click handler is not a crawlable link no matter how it looks.

Then the rel tokens. nofollow, ugc and sponsored are hints to Google about how to treat the link. Since 2019 Google has described them as hints rather than directives, which means a nofollow link is not proven worthless and a bare link is not proven valuable. The honest record is “crawlable anchor without rel hints” or “anchor with nofollow”, and nothing more.

Why one page is never the directory

The single most common mistake in backlink lists is inferring a directory-wide policy from one listing. Directories change templates, run paid and free tiers on different templates, and treat featured listings differently from the long tail. The method therefore requires at least three distinct sample pages and treats a mixed sample as unknown: if two listings carry a plain link and one carries nofollow, you do not know what yours will get.

This is also why the same directory can be “dofollow” on one list and “nofollow” on another, and both authors can be reporting exactly what they saw.

Turning evidence into a queue

The scoring in the toolkit is deliberately blunt. Observed Google result plus a crawlable link gives the SEO component its full credit. Observed result plus a link with rel hints gives a third of it. Every other combination gives zero — unproven, not “no value”. Product fit and evidence of a relevant audience count separately, so a directory with real visitors and an unproven link can still be worth a submission as a referral experiment.

All evidence carries a checked_at date and goes stale after 30 days. The submission-status guide describes the tracker side; the priority file is the evidence side. Nothing in it fetches pages or submits anything. It reads what you recorded and prints an order.

What we therefore do not claim

Given that method, here is what the best-of lists and directory profiles on this site can honestly say about each platform: its name, its official URL, its catalog category, and a sentence about what it lists. Here is what they do not say:

• “Free.” The catalog stores the current free route of every one of the 346 platforms as unknown until someone checks it on the day. Free tiers appear and disappear; a listing that was free in the spring may be paid-only by autumn. Your agent checks the form when it gets there, and the tracker records deferred_paid if the only route costs money.
• “Dofollow.” For the reasons above: it is a property of a sample on a date, not of a directory, and we have not run three-sample checks for all 346 platforms. Where we have run them for our own product, the result belongs to that product’s private workspace.
• A domain rating or authority score. Those numbers come from third-party crawlers, change monthly, and are not a Google signal. Publishing them next to a directory name would be borrowing authority we cannot verify.
• Traffic. We have no measurement of any directory’s traffic and will not print a guess.

What we do publish is the pilot: a real product, listings we can point to, and the outcomes recorded in the tracker — including the ones that ended as pending for weeks or as deferred_paid.

How to earn the claim yourself

If you want a “this directory gives a crawlable, indexed link” line for a platform, do the work for your product and your category: three samples, site: searches on the day, header and canonical inspection, the actual product anchor and its final destination. Write the dates down. Recheck before you submit if the evidence is older than a month. The toolkit’s audit template is the field reference for exactly that record, and the priority script will refuse to award SEO credit for anything less.

That is more effort than reading a badge, and it produces something a badge cannot: a claim you could defend. LaunchRepo gives your agent the playbooks and the tracker to run this for every product you own, at a launch price that does not rise with the number of directories. Which platforms deserve a submission is still your evidence to collect.

Sources: catalog/PRIORITY.md and templates/priority-audit.json in the LaunchRepo repository; Google’s documentation on the site: operator, crawlable links and the 2019 rel-attribute change. Catalog counts as of 17 September 2026.