For many years engines analysed sites with a desktop crawler, treating the mobile version as a variant. Today it's the other way round: Google indexes sites mainly using its smartphone crawler. What counts, therefore, is the version a phone sees. If the mobile version has less text, fewer links or less structured data, it's that reduced version that ends up in the index.
What it means in practice
Mobile-first indexing is not a bonus for "mobile-friendly" sites. It's a question of which version gets read. The consequences:
- content present only on desktop risks not being considered;
- internal links present only in the desktop menu count for less or not at all;
- metadata must be identical across the two versions;
- the performance that matters is the one measured on mobile.
The three possible configurations
| Configuration | How it works | Risk of disparity |
|---|---|---|
| Responsive | same HTML, layout adapted with CSS | low |
| Dynamic serving | same URL, different HTML depending on user agent | medium |
| Separate URLs | mobile site on another host, such as m.example.com |
high |
Responsive
It's the simplest choice to maintain: one HTML, one URL, no logic to detect the device. Differences only arise if CSS hides parts of the content or JavaScript loads different things depending on screen width.
Dynamic serving
The server returns different HTML depending on the user agent. It must send the Vary: User-Agent header, so that caches and CDNs don't serve the desktop version to phones or vice versa. Every change has to be made twice, and disparity is common.
Separate URLs
An m. site with its own URLs requires bidirectional annotations between versions, consistent device-based redirects and the duplication of everything: content, metadata, structured data, hreflang. It is the most fragile configuration. If you still use it, consider moving to a responsive site, treating it as a migration.
Where the differences hide
In my work, parity checks almost always uncover something in these areas:
- Reduced text: paragraphs cut on mobile "to keep things light", shortened product descriptions, FAQ sections removed.
- Content hidden behind interactions: text in tabs or accordions is fine if it's present in the HTML; it isn't fine if it's loaded from an API only on tap.
- Different menus: the desktop menu has fifty items, the mobile one ten. The missing links disappear from the internal link graph.
- Images: different or missing images on mobile, missing alternative text in the phone version, images lazy-loaded with techniques the crawler doesn't trigger.
- Metadata: title, meta description, canonical and meta robots differ between versions, typically with separate URLs.
- Structured data: present only in the desktop template.
- Blocked resources: mobile-specific CSS or JavaScript excluded by robots.txt.
How to check for parity
Comparison with a crawler
Run two crawls of the same set of URLs, one with a desktop user agent and one with a smartphone user agent, both with JavaScript rendering enabled. For each URL compare:
- word count;
- number of outgoing internal links;
- title, description, canonical and meta robots;
- presence and type of structured data;
- number of images and their alternative text.
Every significant difference must be explained.
Spot checks
The URL inspection tool in webmaster tools shows which crawler was used and the rendered HTML it saw. It's the most direct way to see the page "through the eyes" of the engine.
From the browser
Developer tools let you simulate a mobile device and change the user agent. It's useful for a first look, but it doesn't replace a systematic comparison.
Performance and usability
On mobile these matter too:
- load times on slower networks and less powerful processors;
- interactive elements large and spaced enough to tap with a finger;
- no pop-ups covering the content as soon as the page opens;
- text readable without zooming, thanks to a correct meta viewport.
These aren't indexing issues in the strict sense, but they shape the experience of users arriving from search, now largely on phones.
In short
- Engines mainly index the version a smartphone sees.
- Responsive design minimises the risk of differences.
- With dynamic serving use
Vary: User-Agent; separate URLs are the most fragile solution. - Text, links, metadata and structured data must be the same on mobile and desktop.
- Check with two comparative crawls and the URL inspection tool.
