You may already be aware of the importance of optimising your images before uploading them to your website as we discussed in our blog, How to Get Better Google Rankings: Tag Your Images. But the world of SEO is constantly changing. And although image optimisation is still crucial, these days it means more than simply looking at the image format, image size, file size, titles, descriptions and Alt Text.

Even now not everyone optimises their images, so Google has expanded its algorithm to interpret them for itself. As a result, Google now has the capability to analyse what an image is about and how relevant it is to the surrounding content, in order to rank it appropriately. Unsurprisingly, given the millions of images uploaded every day, Google automates this process, scanning every image to identify faces and objects, assign labels, recognise text and other properties.

To see what impact this has, we can use Google’s Cloud Vision API  to gain insights as to how Google interprets our images.

Let’s take the image below as an example of a photograph that a small cafe owner is considering using on their website (the image is titled “Coffee Shop.jpg”). With this image, the café owner wants to convey to us (and Google) the customer’s joyful experience as he dives into a stack of tasty pancakes for breakfast.

Original image of a man sitting at a table in a cafe enjoying his breakfast.
Original Image
Man enjoying his breakfast with various elements which Google API has identified highlighted in green.
As seen by Google

Although we may only briefly glance at the overall image ourselves, Google’s Cloud Vision API takes a more forensic examination, looking for much more detail.

Facial Recognition

Let us begin with the faces of the people in the image: –

Man enjoying his breakfast with faces which the Google API has identified highlighted in green.

Although we are interested in the diner, there are actually four people in the shot, three of which have been detected by the API. Any keen-eyed photographer would already have noticed the woman’s head growing out of the diner’s left shoulder and edited that out as well as her partner at the same table.

More interesting though is how the API has tried to determine the emotions associated with the facial expression and in this case, Google is unsure whether our diner is having a joyful or angry reaction to his meal. If anything, it’s erring towards an angry reaction. So if the café owner wants to rank for “café with happy customers near me” we would recommend trying an alternative image with a more obviously happy diner.

Object Recognition

Now let us look at the objects in the image that have been catalogued by the API: –

Man enjoying his breakfast with objects which the Google API has identified highlighted in green.

Although our eye is drawn to the diner’s face, the table and its contents are equally dominant in the image to the extent that the diner and the table (classified as “packaged goods” here) have been given almost equal weighting. Fortunately, though, it has also recognised the “food” and “cake” as well, but with less emphasis.

It’s possible that this café’s location makes it a popular venue for cyclists (why else would you park a bike on the wall here?) but because the bicycle is not prominent enough in the image (or even in focus) it has not been recognised: this could well be an opportunity missed.


Man enjoying his breakfast with labels which the Google API has identified.
Man enjoying his breakfast with labels which the Google API has identified.

So when it comes to labelling the content in this image the Google API has been more successful at detecting the “Smile” and “Happy” it has more heavily favoured the furniture and fittings in the top ten features with “Food” and “Cuisine” labelled only twice.

Ideally, we want these objects and labels to be as relevant as possible to the context and purpose of the webpage, the nature of our business and – even better – closely related to the focus keywords we are targeting.

So how can this image be better optimised?

As always think about your keywords as they apply to images as well as the written text on the webpage. This particular image could perform better in Google searches if it were composed differently, considering the focus keywords related to the content and minimising the impact of the less relevant elements.

Depending upon the nature of the business, one might make the following changes: –

  1. Rename the image from coffee shop.jpg to happy ideal-cafe-for-coffee-and-breakfast.jpg for example
  2. Focus on the diner, by removing the three other people from the shot
  3. Change the facial expression to illustrate pleasure/joy more clearly, not anger
  4. Reduce the emphasis on the bare table and remove/replace some of the clutter on the table i.e. cup and saucer for the coffee in place of the two glasses and the tin can
  5. The API can also read some text in images and so signs with Café or even a text overlay will be detected.
  6. If one of your target audiences is cyclists, for example, have the diner or another in cyclewear and the bicycle more prominently featured and/or the picture frame could contain an image of a bike.

A professional photographer would automatically correct the compositional deficiencies in the image to produce an intriguing and well-constructed photograph, but they are also likely to need some guidance to produce an image that can be readily identified by the Google API and interpreted as intended.

This is another reason using original photographs taken by yourself, using your mobile or camera or professionally, is far superior to using stock images. By taking multiple pictures from different angles, lighting, focus, expressions and general composition, you can select the best image to help Google focus where you want them to.

For more information about our SEO and copywriting services, please get in touch.

Or why not book a 30-minute website review with us and we’ll give you some hints and tips to increase website traffic and conversions.

About the author: Alan Say is a Certified Google Search Marketer and one half of the dynamic duo that makes up CBL Copywriting & SEO. Their combined expertise can help you attract the visitors you want to your website.

Other articles on this topic you may be interested in: