Is Google’s AMP carousel working?

Publishers are questioning the effectiveness of Google’s AMP carousel. Columnist Barb Palser analysed a total of 235 million AMP search impressions in order to find answers.

Google’s AMP rich cards look like they ought to work. The format is undoubtedly visually appealing, and AMP rich card carousels generally appear at the top of mobile SERPs, thus dominating the first viewport.

They also offer a brand showcase publishers should love.

However, lately, publishers have been confused over analytics indicating that AMP content has a lower click-through rate from rich cards than they do from general SERPS. They are discovering this data in Google Search Console, which now enables publishers to filter AMP performance in the SERPs based on appearance as rich cards in non-rich text links vs. a carousel.

Reason #1: Relevance and supply

In a nutshell, the rich card carousel favours wide-ranging searches. The carousel is raised when a query returns a minimum of three AMP results Google deems applicable enough to present together. The more popular and broader the search, the more likely it is that the threshold will be met and the results are shown as a carousel of up to 10 rich cards. However, wide-ranging searches often result in lower click-through rates than exact searches.

Reason #2: Carousels are not great

As a user interface, scrolling carousels do not have a particularly good reputation; mostly, they appear to be good at keeping content from users. It seems that Google’s AMP carousel is no exception to this.

On mobile devices, the first two cards in an AMP carousel actually appear — or three cards in landscape view. It seems users are not scrolling or clicking far beyond that.

When publishers rank at the top of the SERPs, this is not a huge issue — but for publishers who appear deeper in the carousel, it could factor to lower rich result CTRs. While this sounds pretty bad, remember that Google frequently duplicates a publisher’s results in the AMP carousel as well as in text links.  If the publisher’s text link placement is not affected by carousel appearance, then deep carousel slots are like extra exposure. If deep carousel links replace or displace text links, then fears would be justified.

In the meantime, Google appears to be doubling down on AMP carousels — now showing publisher-specific carousels when certain publishers have a number of results for a query. It would be quite interesting to know whether engagement for these single-publisher carousels is different to the multi-source Top Stories carousel. Scrolling notwithstanding, single-publisher carousels is a great content and brand showcase for publishers who are lucky enough to get them.

Conclusion: Do not blame the cards, blame the carousel.

All this analysis suggests AMP carousel CTR is being controlled by the minimum result threshold which stops rich cards from being displayed for granular queries; and the deficits of carousels in general, which lower CTR for “hidden” content which has to be scrolled to view.

Google could probably boost AMP rich card CTR instantly by eliminating or reducing the threshold for showing results as rich cards, and by presenting them as tiles rather than a scrollable carousel. However, this could lead to trimming the maximum number of cards in a collection from 10 to approximately three or four, in turn decreasing rich card impression and click volume.