With Google and Pinterest both recently investing in improving their image recognition, The Drum's Ishbel Macleod takes a look at how image search is becoming mainstream, exploring the implications for search marketers and brands.
The need for a more visual approach to marketing has been apparent for the last few years, as a consumer trend towards sharing images online means social media marketers worth their salt are now au fait with incorporating images into their campaigns.
This movement towards the visual also has implications for other parts of the marketing mix; the exponential growth of image-sharing sites such as Pinterest, coupled with the consumer propensity towards the visual, means image search is becoming an important part of marketers’ search strategies.
“The visualisation of the web is continuing to grow thanks to the likes of Pinterest, Snapchat and many others; as more and more users upload their own content and share it, this is bringing fresh challenges as internet companies can no longer manually control and classify their content – particularly images and video,” says Adrian Moxley, CMO and co-founder at visual classification company, WeSee.
With so much visual content out there, how can search engines and social platforms ensure that people are presented with accurate image results following a search?
Pinterest is looking into this area, as evidenced by its acquisition of image recognition company VisualGraph in January. As explained by VisualGraph CEO Kevin Jing at the time, the company’s approach is to “combine the state-of-the-art machine vision tools, such as object recognition (eg shoes, faces), with large-scale distributed search and machine learning infrastructures.”
And the social media platform isn’t the only one to have cottoned on to the importance of images as a key element of search. At the start of the year, Google bought artificial intelligence company DeepMind for an estimated £400m, while last year, Yahoo bought image recognition start-ups IQ Engines and LookFlow.
IQ Engines was acquired to join the Flickr team, to work on improving photo organisation and search for the community. The SmartAlbum API from IQ Engines allows for photo analysis and facial recognition for online photo albums and mobile apps, something that will undoubtedly prove useful for Flickr.
While Google would not disclose its reasons for buying the AI company, or if its focus on image search would grow following the DeepMind move, the new talent is set to report to Google’s senior fellow for search, Jeff Dean, which suggests that the acquisition is more to do with helping improve Google’s image search functions than to with its driverless car plans.
Google’s acquisition came after it discussed how deep learning techniques meant that computer programs didn’t have to rely on labelled data. In fact, in June last year, the search giant ran an experiment which asked, if we think of our neural network as simulating a very small-scale “newborn brain,” and show it YouTube video for a week, what will it learn?
“Our hypothesis was that it would learn to recognise common objects in those videos. Indeed, to our amusement, one of our artificial neurons learned to respond strongly to pictures of... cats,” Dean explained. “Remember that this network had never been told what a cat was, nor was it given even a single image labelled as a cat. Instead, it “discovered” what a cat looked like by itself from only unlabelled YouTube stills.”
Google isn’t the only search engine understanding the importance of images, as Bing integrated Pinterest images within its image search results at the end of 2013. And image search is on the increase; the engine says that ten per cent of its searches were for images, with the number of image searches growing by 520 per cent since the site’s launch in 2009.
According to Meenaz Merchant, senior program manager at Bing R&D, the search engine has “used a number of approaches including natural language processing, entity understanding, computer vision and machine learning to provide high quality, relevant images,” in order to achieve this growth.
When users search for images, be it in a search engine such as Bing or Google, or through a social media site such as Pinterest or Instagram, they naturally expect to be understood and for the results to reflect their query. However, 74 per cent of UK consumers said that traditional text-based keyword queries are inefficient to help them find the right items online, according to WeSee research. This has clear implications for brands, says Moxley.
“When keyword search alone often fails to locate the relevant products, online retailers need to take action and implement visually powered discovery shopping, or else miss out entirely on valuable customers.”
Play.com is an online retailer who is taking advantage of this, reporting that revenue for commercial posts with strong social product imagery can be up to 300 per cent higher than those without, while its image-focused Facebook product posts reach an audience over 100 per cent larger than standard posts.
Carol Dray, marketing director at Rakuten’s Play.com comments: “Our figures show the huge power of images to drive engagement across social for a brand. Products posted should be captivating and conversational in order to boost engagement - and we’ve found being humorous can work to push this even further. The visual search trend continues to widen the opportunities for retailers in social commerce; as shoppers move from channel to channel in their purchase journey, customers will be able to find the products they want and discover news ones simultaneously across the retail web. It will be particularly useful for consumers wanting to surface quirky and unusual products. Shopping is about entertainment and images can really help to drive this engagement.”
Caragh McKenna, group account director at The Search Agency, has seen a lot more integration of images into content recently.
“Combined image promotion is something that we have found to be an important factor for the majority of our clients,” she says. “Ranging from the banking sector to a variety of online retailers, optimisation of image content has ensured our stories are not just telling the target audience about the product or brand messaging but are rather showing the user exactly what they are looking for.”
Topshop is one such brand using visual content to its advantage, with Pinterest playing a large role in its search strategy. Its Christmas ‘Dear Topshop’ campaign saw customers in its flagship stores able to pin, share and shop for items on giant touchscreens, as well as users at home being able to get involved. The retail brand could use the most searched for and pinned of these to discover data about its customers’ intent.
Sheena Sauvaire, Topshop’s global director of marketing and communications, explains: “The sharing and pinning of product items from our digital gift guide meant that we were able to view which were the most coveted items. We then used this data to highlight those key items in-store and online; the results were really positive and I think it shows how peer to peer influence continues to be a driving factor, particularly for the younger shopper.”
Research by Piqora in 2013 discovered that one pin generates an average of $0.78 on e-commerce sites – up 25 per cent from the end of 2012. Add that to the most overused Pinterest stat from Convertro – that it generates four times more revenue per click than Twitter and 27 per cent more than Facebook – and it becomes clear why brands like Topshop and House of Fraser are embracing the platform.
And although the platform is “generally not going to be a major direct response channel”, according to Matt Brown, head of earned media, Unique Digital, it’s important that the opportunity of visual search is not overlooked by marketers.
“If you use image search well, a brand can harvest the interest that a great design has stimulated. It makes it easier for people to find and explore what is new,” says Brown.
That is the area which image search is heading. But what does this mean for marketers, and how can brands embrace this?
“To date we have found that Instagram and Pinterest lend themselves largely to the more “attractive” brand propositions – food, travel and clothing – when it comes to promotion of images,” says McKenna, “but image optimisation is certainly not limited to pretty pictures as proven by Google’s recent increase in the introductions of images in search results that align to the business sector, news and medical research. For more data led clients we have found the seeding of graphs and infographics on relevant platforms hugely successful in returning latest pieces ahead of more static site pages.”
Just as people judge books by their covers, brands today are judged by their visual content online. And with recent partnerships making image search a stronger proposition for search engines, it’s important brands take steps to not only optimise images accordingly, but ensure a visual approach is central to their marketing efforts.