Line of Sites

I kept submitting products to launch directories and never knowing whether it did anything. The lists that rank directories are written from the directories' own marketing pages. Nobody fetches the listing page afterwards to see what it serves.

So I started fetching them. lineofsites crawls launch directories and records, per listing: whether the page still responds, the outbound link to the product, the rel attribute on that link, and whether Google has the listing page in its index. Every number carries the date it was measured. History is append-only, so when a link disappears that is a fact with a date on it rather than a gap.

Where it stands today (17 Aug 2026): 202 directories tracked, 34 of them read by the crawler so far, 6,376 live listings, 5,791 products. Across those listings the rel attribute reads 3,895 dofollow, 2,342 nofollow, 139 ugc. Of 148 listing pages checked against Google's index, 17 were not in it, which is a small base and worth saying so.

The gap between 202 and 34 is mostly queue rather than refusal. The crawler has no server yet and runs on demand from my laptop, so most of the rest are tracked and waiting for a first read. Their rows show "not verified yet" and no numbers at all, because a directory I have not measured does not get a score.

Refusal is a separate list. I have assessed 139 further domains and left them without a parser, each recorded with its reason and the date I last looked: 42 render listings client-side, so the product URL is nowhere in the served HTML; 29 turned out not to be listings sites; 23 are dead; 18 publish no outbound product link at all, which means a submission there produces no backlink of any kind; 17 block my crawler, which I record and publish rather than work around. Those verdicts decay. I re-probed five of them two days ago and four had changed behaviour since the previous read.

Two things I cannot show you yet. Per-directory indexation rate is the number I care most about, and it is published for 2 directories, because I hold back any rate with fewer than 30 real checks behind it rather than print one computed from four. Domain rating is wired up and populated for none of them.

One correction, since it is the kind of thing this exists to catch. My first reconciler treated a directory's front page as evidence and recorded 642 links as lost. All 642 were my own bug: the listings were live and the feed had rotated. I reverted them and rewrote the rule so a rotating feed is discovery and never evidence of removal. Any tool that reports lost backlinks off a feed or a cached index has that failure mode.

The scan is free and needs no account: enter a domain, get the tracked directories linking to it, the rel read on each page, and index status where I have it. The catalog is free to browse. There is a paid report and a paid monitor subscription, and neither is needed for the free parts.

My own product is a row in the catalog, measured by the same crawler under the same rules. I publish measurements with dates, not rankings or verdicts on other directories.

Method questions welcome. Criticism of the measurement is the most useful thing I could get out of this thread.