For a while, I published technical articles in two places: on my own website and on developer platforms such as Medium, DEV Community, and Hashnode. My site was the place where I kept my writing together, including versions for readers in other languages. The platforms were supposed to help people discover it.
Eventually, I stopped routinely cross-posting. I was spending time adapting, formatting, and maintaining articles in several places without seeing enough benefit to justify that work. Meanwhile, my own blog was slowly attracting readers and occasional emails about collaboration, including requests to exchange links.
That experience led me toward a simple preference: publish on a domain you control first, and treat every other platform as a distribution channel that has to earn its place in your workflow. I would not go as far as saying Medium or DEV is useless for everyone. But for a developer who is also building software, publishing everywhere can become an expensive habit disguised as promotion.
Publishing and distribution are different jobs
Publishing produces an article that readers and search engines can find at a stable URL. Distribution puts that article in front of people who might otherwise never encounter it. Developer platforms can be good at the second job. They do not automatically improve the first job for your own website.
When I publish an explanation on my own domain, I can connect it to related articles, documentation, and projects. I can update it when the underlying technology changes, preserve its URL, and decide how it fits into the rest of the site. Every useful page becomes another potential entry point into a coherent body of work. Google's link guidance explicitly describes links as a way to discover pages and understand their relevance.
That does not mean each new article deposits a predictable amount of “SEO authority” into a domain. Search performance is not a reward for accumulating posts. Pages still need to answer real questions, be accessible, compete with other results, and earn attention or references. A decade of weak articles can remain a decade of weak articles. But a useful archive on an address I control gives me opportunities to improve, connect, and preserve the work over time.
My own blog has had a modest audience. I have also seen accounts on large publishing platforms with long posting histories but little apparent engagement. Those observations are not a representative study of either publishing model. They were enough, however, to make me question why I was maintaining multiple copies of my writing when the main asset I wanted to build was my own site.
The backlink argument is weaker than it sounds
The most tempting justification for cross-posting is often backlinks. Publish a story on a well-known domain, link to your project, and perhaps some of that site's search reputation will flow to you.
It is not that simple. A link I place in my own article on a publishing platform is not the same thing as an independent editorial recommendation from the platform. Google's outbound-link documentation recommends qualifying user-generated links with rel="ugc" and describes rel="nofollow" for links a publisher does not want to endorse in the usual way. Platforms set their own link policies, and Google does not publish a transferable SEO value for any given self-placed link.
That does not justify claiming that every such link is worth exactly zero. A link may send real visitors, help someone discover a tool, or eventually lead to an independent mention. Its treatment as a ranking signal depends on more than the host site's name. The practical mistake is to assume that creating a profile and inserting a project URL is equivalent to earning a recommendation.
The same caution applies to links from my personal blog to my projects. If an article genuinely explains a problem and points readers to a relevant tool, that link is useful. If the destination is on the same site, it can also help users and crawlers navigate between pages. If the project is on a different domain, the link is still useful to readers, but I should not pretend that a self-controlled cross-domain link is automatically a powerful independent vote.
I also receive occasional outreach proposing reciprocal links. I take that as evidence that somebody found a way to contact me, not as proof of search-engine success or a reason to accept the proposal. Google's spam policies identify excessive link exchanges undertaken for ranking manipulation as link spam. A meaningful collaboration and a manufactured link scheme are different things.
What Medium, DEV, and Hashnode can actually provide
There is a serious counterargument to my preference: an article on my own domain is not necessarily an article anyone will read. A technical community may already have people interested in precisely the problem I am explaining. A good discussion there can produce feedback, professional connections, subscribers, direct visits, or later links from independent writers. For some developers, the audience and conversation are worth more than owning every page view.
That benefit is distribution, not a guaranteed search-ranking shortcut. A post can collect views or reactions on a third-party platform while sending little traffic to the author's site. Equally, a small number of the right readers can matter more than a large view counter. If the goal is to find collaborators, attract users, or share knowledge with a specialist community, a platform can succeed without becoming a measurable SEO win.
My problem was that routine cross-posting had become the default, regardless of whether an individual article had an audience there. I was treating the existence of a publishing button as a reason to press it.
Canonical URLs help, but they do not make duplication free
One thing I would correct in the strongest version of my original argument is the claim that publishing the same article elsewhere always requires rewriting it. Some platforms support importing or syndicating content. Medium's import tool adds a canonical URL pointing to the original. DEV's publishing documentation describes RSS import and an option to designate the source as canonical. Hashnode has also documented republishing with canonical URLs, although older instructions should be checked against the current interface before relying on them.
Those features can reduce copying and formatting work. They also address an important technical issue: when several URLs contain substantially the same article, which one should a search engine treat as the primary version?
Google's canonicalization documentation says that a declared canonical URL is a hint, not a rule. Google can select a different canonical page. Duplicate content by itself is not automatically a spam violation or a ranking penalty, and the presence of a canonical tag does not guarantee a ranking boost or a stream of visitors from the republished copy.
There is also an important distinction between true duplicates and broader syndication. Google's guidance on syndicated content notes that canonical tags may be unsuitable when partner pages differ significantly; if the goal is to keep those copies out of Google Search, partner-side noindex is a more effective approach. A developer cannot assume that every publishing platform offers that control. And if I substantially rewrite an article for a different audience, I should not treat the canonical tag as a magic declaration that two different pages are identical.
So cross-posting can be technically cleaner and less laborious than I once made it. It still creates another published artifact to review when a code sample becomes obsolete, a link breaks, or the original article changes. The question becomes whether the incremental readers justify that maintenance—not whether copying is technically possible.
My multilingual blog is useful only if the writing is useful
I publish versions of my articles for readers in different languages. Keeping those versions on my own website lets me organize them under a consistent navigation and update the underlying explanation without scattering its language variants across unrelated profiles.
For search engines, multilingual publishing requires more than translating navigation labels. Google's localized-page guidance explains how hreflang helps identify language alternatives and notes that pages with untranslated main content can still be treated as duplicates. Good translations, appropriate language-specific URLs, and correct alternate-page relationships help make the archive usable. Publishing more language versions does not multiply rankings automatically.
This is another reason I value control of the original site: it is easier to maintain a deliberate information architecture than a collection of platform-specific copies. But control is useful only if I keep the material accurate and worth reading.
Owning a domain is not the same as running a server
Another part of the decision has changed over time. An individual developer no longer needs to operate a complicated server just to publish a technical blog. A static site can be generated from text files and served by a hosted service. For example, GitHub Pages supports personal sites and blogs through its free offering, subject to its terms and limits.
But “my own website costs nothing” is too absolute. A custom domain normally has registration and renewal costs, hosting tiers have limits, and building or maintaining the publishing workflow still takes time. A managed platform can be the right answer for somebody who would rather write than maintain a site.
The more useful distinction is control over the address and the content, not who owns the physical server. It is possible to use hosted software while publishing on a domain you own. The commercial details matter: Hashnode's June 2026 Pro announcement lists custom domains among its paid features, with an exception for previously configured sites. A hosted blog on your own domain may satisfy the goal of building durable URLs, but it is not necessarily free or free from provider lock-in.
How I would allocate the next hour
For a developer with a product to build, time is the real constraint. Cross-posting has to compete not just with writing another article but with fixing a bug, improving documentation, answering users, or making an existing post genuinely more useful. I use a simple order of priorities:
- Publish the useful, complete article on my own domain first. Give it a stable URL, sensible internal links, and the technical details a reader actually needs. Keep the original up to date.
- Connect the article to relevant work. Link to related posts, documentation, or projects when the connection helps the reader. Do not manufacture links simply to influence rankings.
- Distribute selectively. Share an excerpt or a useful standalone discussion where the community is likely to care. A full republication is an option, not a requirement.
- If republishing, preserve source attribution. Use the platform's canonical controls where appropriate, verify the published page, and account for future corrections. Do not assume a canonical tag determines Google's final choice.
- Judge the incremental return. Look for referral visitors who actually read, meaningful conversations, project interest, and useful relationships—not merely impressions or a count of posts published. Compare that return with the extra editing and maintenance time.
I do not have a controlled experiment proving that personal blogs outperform Medium, DEV, or Hashnode for every developer. My observations would not support that claim. The decision is instead an allocation rule: if a distribution channel consistently contributes little to the outcome I care about, I would rather put that attention into the work and website I can improve directly.
Where I landed
I no longer feel obliged to post every article in several places. I want my technical writing to become a useful, connected archive on my own domain, with a clear path from a problem I have explored to the projects or explanations that might help somebody else. The occasional inbound message is encouraging, but the long-term value is the body of work itself—not the promise of a particular ranking.
Medium, DEV, Hashnode, and communities such as Indie Hackers can still make sense when they deliver a specific audience, discussion, or opportunity. I would not dismiss those outcomes. I just would not maintain parallel publishing workflows solely because the platforms are popular or because I hope a self-placed backlink will improve SEO.
Build the home for your work first. Distribute when the distribution is worth the cost. For me, that distinction made it easier to stop writing for every platform and get back to building.