Loading today’s stories
Build1 distinct publisher3 min readUpdated
A twenty-site test found one free listing that was both followed and indexable. The other nineteen were killed in the head of the document or by a pricing tier, not by the anchor.
The Engineer · Build desk
Compiled by The EngineerSomething wrong?How this is made
A `rel` value is an instruction about one link on a page a crawler has already decided to keep. The robots meta and the canonical are instructions about the page itself, and they resolve first [4]. That ordering is the whole result: a followed anchor on a `noindex` page has nothing to carry [6], and a tool page declaring the directory homepage as its canonical is asking Google to fold it into a URL where the vendor link does not appear [10].
The list-makers are not lying about attributes; by the author's own account they are often right about them [4]. They are reading the one layer you can read without leaving the anchor. Two of the most-recommended platforms noindex free accounts specifically, and on one hosted-changelog product indexing is itself the paid upgrade [7]. A large GIF platform renders brand-channel website fields as `rel="noopener noreferrer"`, followed, while user channels stay noindex until the brand is verified [8]. Both appeared on every list the author read, and neither list mentioned the robots meta [9].
Nineteen of the twenty measured listings produced nothing usable for a free account [1], a 5 percent hit rate [2] in a category whose entire pitch is the attribute. Three of the four gates the author describes sit above or outside the anchor [3].
The fourth is not in the HTML at all. On one directory, product links carry `rel="nofollow noopener noreferrer"` while sponsor slots in the same template carry `rel="noopener"` [13]. That is not a crawl policy, it is a price list rendered as markup. Another platform's free tier requires a reciprocal badge, and none of the five followed listings sampled carried one anywhere on their site, which points to those five not being free listings [14]. Measuring a competitor's live page tells you what that competitor's tier bought [15].
Worth noticing what the author declines to claim. On the canonicalised directory he could not prove the tool pages get indexed independently and says so; a canonical is a strong hint Google can reject, and a link that depends on that rejection is a wager either way [11]. The same restraint produces the only honest label for a queued free submission: unproven [15].
Then there are the pages where the attribute question never arises. A large software-comparison site serves `index, follow`, self-canonical, dedicated category and comparison pages, and zero outbound anchors to the vendor [16]. A startup directory's homepage is full of followed outbound links, all of them sponsor slots, while the permanent detail page for each listing links nowhere [17]. An integrations marketplace gives each app its own page and no link out [18]. Any tool that scores a domain rather than a rendered page will file all three as opportunities.
Follow any of these and your For You feed starts watching them — no settings page required.
Ranked by verification strength, evidence, and original report placement.
Exactly one of the 20 platforms gives a free user a followed, indexable link.
Three launch platforms measured as followed vendor links on indexable pages and were written up as recommended; after the author submitted to one, the confirmation email listed "do-follow backlinks" as a benefit of the $49 tier.
The author checked 20 platforms one at a time on real published pages using a script.
The author's fourth gate is which pricing tier the measured page was published under: a live page's rel is evidence about that listing, not about yours, and the honest answer for a queued free submission is "unproven", not "dofollow".
The test script fetches a published listing page with curl and a desktop user agent, then extracts the robots meta, the canonical link, and the rel value of every outbound anchor.
First failure mode: the link is followed but the page it lives on is not in the index, so nothing flows. Two of the most-recommended "free dofollow" platforms do this to free accounts specifically.
Evidence-backed comparisons of source perspectives and observed adoption signals. Read the methodology
Which Builder, Operator, and Investor concerns the observed source mix emphasized—not a truth score.
Evidence, demonstrated adoption, hype gap, incentives, and confidence are assessed independently, each on its own current evidence. How these are measured.
Direct HTML measurement, single unreplicated observer
The strongest element is method transparency: the full audit script is published, the signals checked (robots meta, canonical, rel) are directly observable in served HTML, and specific artefacts are quoted verbatim. Against that, everything rests on one self-published author's single run: no platform is named except WordPress.org, so no verdict can be independently checked; the sample is twenty measurable platforms with seven more bot-walled; and the tier-gating conclusions rely on one confirmation email and a five-listing badge sample. The author's own refusals to over-claim (canonicalised pages, unmeasured rows) raise credibility without adding external corroboration.
One founder's audit; platform behaviour observed, practice uptake unknown
There is real observed behaviour to point at — twenty platforms' live link and indexing policies, one platform's $49 do-follow tier, a reciprocal-badge free tier, and one confirmed referring domain in Bing Webmaster Tools — which is why this is not zero. But the observations come from a single operator's own site and sample; nothing in the supplied source shows other practitioners running this four-gate check, any tooling adopting head-level checks, or any measured ranking or traffic outcome from the one passing listing.
Slightly understated relative to its own evidence
The framing is stark ('One of them was real') but the body is unusually disciplined about its limits: the author refuses to claim canonicalised tool pages are indexed independently, downgrades bot-walled platforms to 'unmeasured' after nearly recording a block page's noindex as a verdict, and insists the honest status of a queued free submission is 'unproven'. That restraint pulls the piece just below its evidence. The offsetting risk is generalisation — a twenty-platform, unnamed, single-observer sample is used to state a rule about 'modern directories' — which keeps the gap close to aligned rather than strongly negative.
Founder-authored link-building content on a self-publish platform
The author states he runs a link-in-bio SaaS and needed backlinks, and publishes on dev.to, where there is no editorial gate; SEO-adjacent 'I measured X' posts are themselves a distribution tactic, and the piece closes by disclosing that the one passing directory is his site's only referring domain. Platforms are left unnamed, which limits both defamation risk and reader verification. Countervailing signals exist — a published script, a self-correcting retraction of three earlier recommendations, and explicit non-claims — so this reads as ordinary founder-marketing incentive rather than an undisclosed sponsorship; the supplied source shows no payment, affiliate arrangement, or vendor relationship.
Method credible, verdicts unverifiable
Confidence is limited mainly by structure, not sloppiness. The signals measured are objective and the script is reproducible, so the mechanism claim — that robots meta, canonical, and pricing tier decide outcomes above the anchor's rel — is fairly reliable. The specific findings are not independently checkable: one publisher, one observer, twenty unnamed platforms, seven unmeasured, one tier disclosure, and a five-listing inference the author himself hedges. Platform templates and tier policies can also change after a single point-in-time fetch.
build
A 5x publishing increase cost one site 1,000 indexed pages and every impression1 distinct publisher
build
Before you spend quota on an agent skill, make it pass an eval harness1 distinct publisher
build
Geofencing beats GPS polling on power, then loses to the OEM battery optimiser1 distinct publisher
build
Google Trends returns 200 OK with an empty body when it blocks you1 distinct publisher
Distinct publishers with included, body-backed reporting in this cluster.
dev.to
1 article · August 23, 2026