Infinite scroll and lazy loading
The engine does not interact with the page: it scrolls nothing and clicks nowhere. Content that only appears after a gesture stays invisible. The documentation asks for loading triggered by visibility in the viewport, and, for infinite scroll, a persistent unique address per chunk.
The fact that governs everything else
Google Search does not interact with the page: the recommended lazy-loading methods are precisely those that do not rely on user actions such as scrolling or clicking.
officialSource : Google Search Central, “Fix lazy-loaded content”
A modern page often assumes a gesture: scroll to load more, open a tab to read a card, click “show more”. Each of those gestures is a condition, and the condition is never met on the engine’s side.
This is not a limit of power but a design decision, and it has a simple consequence: anything that appears only after an action stays out of reach, however carefully the rest of the page is built.
Load on visibility, not on gesture
For all content to be seen, the implementation must load every relevant element whenever it becomes visible in the viewport; three means are named, among them the browser's built-in lazy loading and the IntersectionObserver API.
officialSource : Google Search Central, “Fix lazy-loaded content”
The distinction is fine and decisive. Loading tied to scrolling depends on a gesture; loading tied to visibility in the viewport fires because the element enters the frame, which also happens during an automated render. The first formulation excludes the engine, the second includes it.
The documentation advises against applying lazy loading to content likely to be immediately visible when a page opens: the content would take longer to appear, which the reader would notice very clearly.
officialSource : Google Search Central, “Fix lazy-loaded content”
That recommendation serves the reader before the crawler, and it describes a very common fault since lazy loading became a single attribute: the header image, deferred like the rest, arrives after the text and makes the layout jump.
Our infinite-scroll catalogue is fully optimised: Googlebot scrolls the page like a user and sees every product.
Infinite scroll, and the price of an address
An indexable infinite scroll requires each chunk to have its own persistent, unique address whose content stays the same every time it loads; the documentation recommends absolute page numbers rather than relative parameters such as a relative date.
officialSource : Google Search Central, “Fix lazy-loaded content”
The detail about the parameter is worth underlining, because it is easy to miss: an address built on a relative value points to different content depending on when it is loaded. It looks like an address, it does not behave like one, and nothing stable can be indexed at it.
Those addresses must be linked sequentially so they can be discovered, and the displayed address must be updated through the History API when a new chunk becomes the primary visible element, so that the reader can refresh, share and return.
officialSource : Google Search Central, “Fix lazy-loaded content”
A list without an address per chunk cannot be shared: a reader sending the link to what they have just read actually sends the top of the list, and that usability failure comes long before the indexing question.
hypothesis
The status is that of a reading: the documentation lists sharing among the benefits of updating the address, without making an argument of it. We go further by noting that the failure is paid for on the reader’s side long before indexing — a link sent that does not lead to what you meant to show is an immediate disappointment, and there is nothing technical about it.
The check, in one capture
The check consists of examining the rendered HTML with the URL Inspection tool: the implementation is correct if image or video addresses appear in the src attribute of the corresponding elements.
officialSource : Google Search Central, “Fix lazy-loaded content”
It is the one check that settles the question, and it applies to the rendered code, not the source. A placeholder attribute left in place and an image address missing from the src attribute say the same thing: what the engine receives is not what the browser eventually shows.
Giving every chunk of an on-demand list a stable address
hypothesis
- Effort
- high
- Time to effect
- weeks
- Risk
- none
- Durability
- structural
The effort is the highest among the technical cards on this site, because it touches the architecture of a list rather than a setting. The delay follows the crawling of the new addresses. The risk is low: addresses are added, none removed. Durability is excellent, a stable address staying useful long after the display mechanism that prompted it.
What does not work, or no longer works
Counting on the crawler to scroll. The documentation writes that it does not interact with the page.
Then deferring the load of what is visible on opening. The documentation advises against it explicitly, and the reader notices before anyone else.
Finally, building chunk addresses on relative values. The same link will not lead to the same content tomorrow, which strips those addresses of the one property that made them useful.
Sources
- Fix lazy-loaded content — Google Search Centralaccessed on
- Pagination, incremental page loading, and Search — Google Search Centralaccessed on
Block revalidated on
Frequently asked questions
- Does the crawler scroll the page?
- No, and that fact governs the whole subject. The documentation explicitly recommends methods that do not rely on user actions, because Search does not interact with the page. Content arriving only after a scroll or a click does not exist for it.
- Is lazy loading discouraged?
- Not at all: it is a performance and experience best practice, provided the trigger is tied to visibility in the viewport. What the documentation discourages is applying it to content immediately visible on opening, which would then take longer to appear.
- How do you make infinite scroll indexable?
- By giving it addresses: one per chunk, persistent, unique, always returning the same content. The documentation recommends absolute page numbers rather than a relative parameter. Those addresses must be linked to each other sequentially so they can be discovered.
- What is the History API for here?
- For updating the displayed address when a new chunk becomes the primary visible element. The benefit is the reader's first: they can refresh, share and return to where they were. Without it, the address shown stays that of the first chunk.
- How do you check the content is really seen?
- By looking at the rendered HTML in the URL Inspection tool, not at the source the server sent. For an image or a video, the sign that the implementation works is that its address appears in the src attribute of the rendered element.