In the last few years, images have begun to change the way we search online.
An increasing number of people are using visual search – the process of using an image as a search term (instead of keywords) to find out more information, or to discover similar images.
Despite the fact that visual search has been around for a decade already, recent advances in technology have meant we are now able to process vast volumes of images at a time. Naturally, this presents huge opportunities for marketers, and particularly those in the ecommerce industry.
However, the benefits of image recognition technology aren’t just easier or more intuitive search for consumers – there are benefits for the marketer, too.
Econsultancy’s Visual Search Guide for Marketers delves into the topic in-depth, but in the meantime, here’s a look at how the technology can help marketers gain an edge.
Improving merchandising / personalising the shopping experience
The benefits of visual search for customers are clear. When it comes to fashion or home décor and customers have a picture of something they want, it is easier to search for matching products using the image, and the shopper is more likely to find specifically what they are looking for.
However, visual search isn’t confined to a ‘click to buy’ strategy – where the customer searches for something they want and directly purchases it. It can also be used as part of an effective merchandising strategy, in order to create a much more personalised and unique shopping experience during the browsing or discovery phase (not just when customers are in a ready-to-buy mindset).
For example, instead of recommending or categorising products based on past behaviour or purchases – retailers are able to recommend items based on the detailed visual attributes of searched-for products. This means that retailers can analyse and consider specific details of an image, including patterns, shapes and colours, and use this to deliver more relevant and tailored content. In contrast, keywords tend to be fairly broad (e.g. ‘black jeans’), giving no indication into the personal preferences of the customer, and resulting in a less personalised experience.
In some cases, visual search can also be used to recommend products based on the customer’s existing wardrobe. This is the premise of AI-app Intelisyle, which uses visual recognition technology to offer personal styling advice.
— Intelistyle (@intelistyle) March 27, 2018
Finding new trends and insight on social
Social listening is the process of monitoring social media and digital conversations to uncover what consumers are saying about a brand or its products. Most of the time, words are the only thing that’s taken into consideration.
However, when computer vision is applied, social listening platforms can access and understand so-called ‘moments of consumption’ – which refers to the context in which products are being used, and the sentiments of those who are using them. By looking at this, marketers can gain insight into customers, spot trends, and uncover unique marketing opportunities.
One common strategy is to search social feeds for brand logos. When Starbucks did this, it discovered a number of images of dogs enjoying ‘puppucinos’ (take-out cups filled with whipped cream) in cars. This rather unique phenomenon would have otherwise gone unnoticed without image analysis, leaving Starbucks unaware that dogs (or rather, their owners) could turn out to be a valuable target.
Determining future trends (and product lines)
Image analysis on social media doesn’t only discover how people are using products – but it can also uncover consumer desires and be used to predict future trends.
On sites like Instagram, for instance, image analysis and machine vision can determine the most popular and in-demand colours of the moment. Using this insight, retailers can then spot gaps in product lines, and even base future designs on it.
Meanwhile, with fashion retail brands in particular using a fast production model – where consumers are able to access a continuous cycle of trend-based clothing rather than wait for seasonal collections – this real-time context and insight can be invaluable. It’s also a more customer-centric and flexible strategy – one that relies on a brand’s core audiences rather than high-end fashion houses to dictate what is sold in-store and online.
The same applies to other kinds of retailers. For its ‘Built In Pin’ campaign, Home Depot analysed what was currently trending on Pinterest (in terms of interior colour palettes and decorating styles) and used this insight to inform its related campaign video, which showed the before and after of various looks. In doing so, the campaign was guaranteed to pique the interest of Pinterest users pinning similar colours and styles, as well as satisfy their curiosity in terms of how a final room might look when decorated that way.
Forming strategic influencer partnerships
A final way visual search and image recognition technology can aid marketers is in terms of influencer marketing. This is because, often, influencers will be chosen based on their reach or follower count. This is also done manually, involving intensive research by brand marketers.
However, this does not necessarily give a good indication of whether the influencer will connect with the brand’s own audience, or is a general good fit.
In order to take a more nuanced approach to influencer partnerships, image recognition can be used to analyse the influencer’s general sentiment (i.e. the types of things they care about), and real-time engagement. As well as freeing up both time and resources, the technology can also help to uncover influencers that might slip through the net.
If a brand wants to promote a particular product, such as a health food, searching via hashtags might return a certain amount of relevant candidates, but not all. Searching based on imagery, however, is likely to uncover far more.
For more on visual search, subscribers can download Econsultancy’s comprehensive Visual Search – A Guide for Marketers.