Not such a blank space – how marketers will be using computer vision in 2019
In a technology scene flooded with artificial intelligence, machine learning, voice, blockchain and mixed realities, Ben Plomion warns marketers not to get left behind as computer vision changes the way we find context online.
‘What do marketers and Taylor Swift have in common?’ is not the setup for a cheap gag, but it is perhaps surprising. Both began to deploy computer vision technology in 2018 and are expected to increase deployment over the next few years.
Swift made headlines last month when it emerged that her security team used computer vision software to detect stalkers at a Los Angeles concert in May 2018. Although it raised privacy concerns related to the ownership and storage of the images, Swift’s use of the technology is likely to be replicated at similar events. Earlier in 2018 Ticketmaster invested in Blink Identity, a startup designing sensors that can identify people walking past at full speed, without the need to stop and stand in front of a lens.
While Swift’s use of the technology thrust computer vision out of the tech press and into the entertainment blogs, the technology has been around for some time and is increasingly being used by marketers to help them take control of their visual strategy as they shift towards using visual content as their primary canvas for storytelling and influencing consumers online.
Interviews conducted in July last year by GumGum and Digiday Media – with 379 digital media executives from agencies, brands, publishers, social media platforms and technology providers – identified the strategies and struggles of media in the digital world and how they are using computer vision to help.
This is what we found out.
Living in a visual world
Unsurprisingly, 80% of our executives said that visual content is either ‘very’ or ‘somewhat’ important to their marketing strategy (with none saying it wasn’t important at all). However, there were a few surprises when we asked how they were using visual content.
While 79% said they post images and videos on their owned and operated platforms, 29% said they use videos and images on social media platforms, 36% use augmented social media graphics such as lenses and geofilters and only 19% use images and videos in display ads. This speaks to a need to maintain control of both the context of the images and the technology that underpins it, but also reflects concerns over viewability and brand safety when a marketer can’t control the context in which the visual content appears.
As the study dug deeper into the issue, we discovered that there were three clear areas of concern:
- Viewability – 42% of marketers said insufficient viewability challenged their ability to raise brand awareness)
- engagement – 41% of marketers said that contextual irrelevance was their biggest concern in terms of getting users to engage with context, and
- generating revenue – 46% said contextual irrelevance was their biggest overall concern when it came to generating revenue.
Getting in context
To help assuage these concerns, marketers are beginning to rediscover the art of ‘contextual targeting’ – essentially placing marketing materials adjacent or in environments that are related or complimentary. In a post GDPR-world, this also requires little to no demographic information which is another reason for its renaissance.
While the concept isn’t new (think putting a travel ad in the travel section of a newspaper) contextual targeting in the online world is almost impossible to achieve manually, the old-fashioned way at least, due to the sheer scale of images out there. This is where computer vision technology comes in.
Image recognition technology can help comb through millions of existing images and videos, then place relevant marketing in the right context in a fraction of the time it would take for a human to. For instance, an advertiser marketing a luxury car might use the technology to run an ad on a website geared toward car enthusiasts. Similarly, it can be used in the reverse way to ensure brand safety and prevent images from appearing next to certain unfavourable content. The technology has been around for a while and is now sophisticated enough to be used effectively at scale.
At the beginning of the curve
Although the tech exists, only 12% of marketers are currently using computer vision technology. Of these, 59% are using it to detect unsafe brand content, 44% use it to determine the reach or value of visual branding and 10% use it for ad targeting.
Despite low levels of adoption for now, interviews confirmed high levels of awareness of how it could be used, with many executives expecting to introduce image recognition technology in the coming year(s).
As we move through 2019 and into the next decade, there’s little doubt we are going to see a significant increase in the use of computer vision technology in digital marketing to help brands cut through the visual clutter and drive better business results. The technology already exists and is there to be harnessed.
The question is – are you ready for it?
Ben Plomion is CMO of GumGum
- Down the garden path? Six ways for brands to blend martech with customer journeys »
- What marketers should expect from consumer tech in 2019 »
- AdRoll VP of international on adtech martech convergence and a year of change »
Image credit:Paul Skorupskas