Subdomain vs subdirectory: the actual difference
A subdomain is a separate hostname that shares your registered domain — blog.example.com is a subdomain of example.com. A subdirectory is a path on the same hostname — example.com/blog/. From a DNS and serving perspective they are completely different: subdomains have their own DNS records, can be hosted on entirely separate infrastructure, and can sit behind different TLS certificates, CDNs, and origin servers.
From a search-engine perspective the distinction is more nuanced and has shifted over the years. The current (2026) reality is that subdomains are related but not identical properties. Some signals aggregate across them; many do not. The argument is not whether one is universally better — it is which signal-aggregation behaviour you want for the specific content you are publishing.
How Google actually treats subdomains in 2026
Google has been publicly consistent for over a decade: both architectures can rank, and the choice should be driven by user and engineering needs, not perceived ranking advantage. That public position is true. It is also incomplete. In practice, here is what we observe in client work:
| Signal | Same hostname (subdirectory) | Different hostname (subdomain) |
|---|---|---|
| Internal link equity | Flows freely across the hostname | Treated as external links between the two hosts |
| Topical authority assignment | Aggregated to the parent host | Assessed per host, with partial transfer |
| E-E-A-T signal carryover | Strong — same publisher entity by default | Inferred — requires explicit signals (sameAs, publisher schema) |
| Search Console | Single property covers everything | Each host needs its own property (or use Domain property) |
| Structured data graph | Single Organization graph applies | Each host needs its own publisher graph |
| Crawl budget | Shared with the rest of the host | Allocated independently per host |
The single biggest practical difference is the link-graph one. PageRank-style equity does not flow between subdomains the way it flows within a single hostname. For a marketing site whose primary goal is to compound authority around a small set of commercial pages, that matters a great deal.
In every site migration I have run from a subdomain blog to a /blog/ subdirectory, the parent site has gained ranking strength on its commercial pages within the recovery window. Not because subdomains are penalised — because the link equity finally aggregates.
When a subdomain is the right call
Subdomains exist for good reasons. They are the right architecture when one or more of the following is true:
- Genuinely separate audienceDevelopers consume docs.example.com; consumers consume www.example.com. Treating them as separate properties matches user intent and lets each be optimised independently.
- Genuinely separate infrastructureIf your help centre runs on Zendesk and your marketing site runs on Next.js, putting help on a subdomain is operationally simpler than reverse-proxying through the marketing host. The SEO cost can be paid down with internal linking.
- Language or country targetingfr.example.com / en.example.com or ca.example.com / us.example.com are officially supported hreflang patterns. Subdomains are appropriate here.
- Commerce on a different platformshop.example.com on Shopify, www.example.com on a custom site. Common, sensible, no realistic alternative.
- Brand separationA separately operated sub-brand with its own editorial voice and link profile that should not be conflated with the parent.
- Compliance or security boundariesCustomer portals, banking flows, or anything where you want a hard isolation boundary between marketing and product.
When a subdirectory is the right call
For most marketing-site decisions, the subdirectory is the safer default. It is right when you want every page you publish to compound authority on a single host. Specifically:
If the blog exists to rank for top-of-funnel queries that funnel users to your service pages, a subdirectory is almost always correct.
These are link-magnets whose authority you want flowing back to the host. Subdirectories make that automatic.
Hundreds or thousands of templated pages benefit from sharing the host's authority budget rather than starting from zero on a fresh subdomain.
Same publisher entity, same audience, same goal. No reason to fragment.
If your marketing pages and your blog have the same audience, same brand, and same goal, a subdirectory is the right call almost every time.
Migrating between the two without losing rankings
The most common engagement we run in this category is moving a blog from a subdomain to a subdirectory using a reverse proxy, where the blog continues to be served from its original CMS but is exposed at example.com/blog/. The migration is straightforward in principle and unforgiving in execution.
- Map every URL, in advance, with no gapsEvery blog.example.com URL needs an exact /blog/ counterpart. Programmatic mapping plus manual review of the top 100 by traffic.
- 301 redirect at the edgeImplement permanent redirects from the old subdomain to the new path before the new path is publicly linked. The redirect must be a 301, not a 302 or a JavaScript redirect.
- Update internal links sitewideReplace every internal link that points to the old subdomain. Do not rely on the redirect chain to do this work — it slows crawling and dilutes equity.
- Resubmit a fresh sitemapSubmit a sitemap of the new URLs in Search Console the day the redirects go live. Keep the old subdomain's sitemap submitted for 30 days so Google can discover the redirect chain.
- Validate canonicals on the new pathThe new /blog/ pages must self-canonical. The old subdomain pages must redirect, not be left to canonical to the new ones — canonicals are weaker than 301s.
- Monitor for 90 daysIndexation, impressions, average position. Expect a 4 to 8 week trough; investigate any losses still present at week 12.
The single most common cause of ranking loss in these migrations is incomplete redirect mapping — pages that existed on the subdomain but were not given a redirect target. Run a full crawl of the subdomain before you start, not just an export from the CMS, because content can exist outside the CMS that the CMS has no record of.
Anti-patterns we still see in 2026
Manual penalties attach to the publisher and the link patterns. A new subdomain inherits both unless you change everything else too. Almost always a wasted move.
Worst of both worlds — you neither concentrate authority on the parent host, nor build enough authority on the subdomain to compete.
shop.example.com, store.example.com, and buy.example.com all selling the same thing. Pure dilution. Pick one.
Page A canonicalises to page B which redirects to page C which canonicalises to page D. Google honours the simplest signal it finds; chains are routinely broken.
If you operate a subdomain as a separate property, give it its own Organization graph. Otherwise you confuse Google about who the publisher actually is.
best-toronto-seo-tips.example.com helps no one. The keyword in the hostname is not a meaningful ranking signal.
The honest summary
Subdomains are not bad for SEO. They are an architectural choice with consequences. For most marketing sites, subdirectories concentrate authority more efficiently and are the safer default. For genuinely separate properties, subdomains are correct. The mistake to avoid is the unconsidered default — picking one because your CMS made the choice easy, and discovering twelve months later that you have been splitting equity for no reason. Make the decision deliberately, document why, and revisit it the next time the architecture is touched.
Frequently Asked Questions
The questions we get asked at the start of nearly every site-architecture engagement.