I checked the rel attribute on every directory listing we own. Nine listings, zero followed links.

We spent three weeks submitting to directories. Around 38 submissions across 11 platforms. This week I finally measured what the ones that went live actually give back.

Nine live listings across four platforms. Zero followed links.

SourceForge, free vendor tier: there is no link at all. Our domain renders as plain text inside a div. Paid advertisers on the same page get six anchors each, through an ad redirector.

AlternativeTo: rel="nofollow noopener".

Indie Hackers, yes, here: rel="nofollow noopener".

Startup Fame: rel="nofollow" before approval. Their free Verified tier does publicly promise do-follow links on approval, so that one is genuinely still open and I am not calling it yet.

The control is the part that makes this a finding rather than a broken parser. I ran the identical read against two paid listings on the same platform in the same minutes. They returned six anchors each. So the platform can emit an outbound anchor. It just does not for free listings.

Two things I would do differently, and the second one is the one that actually stung.

First, read the rel before you count a listing as a win. "We got listed" is not a result. A directory listing is a brand and AI-citation surface, not a backlink. Ours do produce real citations in AI answers, which is worth something. It just was not the thing we thought we were buying.

Second, and this is the one: check your own estate before you go looking for someone else's. On the same day I found the nine zeros, I checked whether our own products link to each other. Four products, twelve directed pairs, twelve zeros. Our corporate hub linked to exactly one of the four, twenty-two times, and to the other three not once. Meanwhile our homepages were emitting nine followed links out to directories that return none. We were exporting the only link equity we fully control and giving our own products nothing.

That was free to fix, it needed no account and no review queue, and nobody had looked.

If you run more than one product, go read your own footers before you submit to a fortieth directory.

The method is one same-origin fetch per platform, reading the anchor's rel straight out of the raw HTML rather than trusting a rendered page. Happy to write it up properly if anyone wants it. The crawl-and-audit side of this is what nexusbro.com does, which is why I had the tooling to check nine listings in an afternoon; the first scan needs no signup and no card.