Computer vision company GumGum is using artificial intelligence to unlock the value of every online image. They work with 70% of Fortune 100 companies, such as L’Oreal, Disney, and Sprint, to reach targeted audiences in native and non-intrusive ways.
With 10 years experience in proprietary technology and solutions, it’s no wonder GumGum’s AI can process and identify more than 500 million images and videos each month. We spoke with GumGum CMO Ben Plomion (LinkedIn) to learn more about the AI running GumGum and how it’s driving customer recruitment and retention for marketers.
In a single sentence or statement, describe GumGum.
GumGum is an artificial intelligence company with a focus in computer vision. Its mission is to unlock the value of visual content produced daily across diverse data sets.
How does GumGum use artificial intelligence (i.e. machine learning, natural language generation, natural language processing, deep learning, etc.)?
Since day one, GumGum’s vision has been to extract value from visual content such as images and videos. This is done with computer vision. If you are not familiar, computer vision is the ability for machines to translate an image (or video) into pixels and interpret the big picture that those pixels represent. What we do at GumGum is collect, train and classify what I like to describe as ‘big pixels’.
We process more than 500 million images and videos each month, and over the years, we’ve gotten pretty good at it. Within one image, our technology can identify celebrities, scenes, the make and models of vehicles, even the color of your skin and hair. To accomplish this at scale requires a robust combination of machine and deep learning and NLP (natural language processing).
What do you see as the limitations of artificial intelligence as it exists today?
AI can be very cost prohibitive, but as with any emerging technology application, as it further develops, we’ll start to find more efficient ways to do things.
In terms of marketing adoption, we see a lot of interest, but investing in building out technology stacks, especially proprietary, requires a certain expertise that the marketing/ad landscape could use more of. We expect this to increase as brands look for more sophisticated solutions for evolving marketing plans.
What do you see as the future potential of artificial intelligence in marketing and sales?
In the future, AI practices will become a normal part of campaign development, execution, and measurement. We currently see a few AI practices in steady rotation in the marketing and sales realm—chatbots being one of the most common and consumer-facing examples.
Currently, sales and marketing are already integrating AI practices through deep data set analyzations. We’ve seen increased sophistication with this layer to make sense of big data, help build profiles, and conduct targeting. We expect to see the reliance on emerging AI technologies increase as sales and marketing organizations seek to differentiate their offerings beyond a pure data play, such as with some of the behavioral targeting ad companies.
Obviously, we are biased toward the potential of computer vision. We’ve been building out our technology solution for 10 years and discovering new ways to apply and monetize the tech in new areas. We expect this trend to continue, especially within sales and marketing disciplines as the world continues to create billions of pieces of content and communications that only using computer vision can make sense of.
We also see an increased reliance on computer vision and GumGum technology when it comes to tackling brand safety issues for marketers. Traditional approaches to brand safety rely strictly on analyzing the text that appears on a publisher's site. We leverage our AI to detect unsafe text and imagery, allowing us to deliver ads in brand safe, contextually relevant environments.
What makes GumGum different than competing or traditional solutions?
Ten years ago GumGum pioneered an entirely new ad format, recognizing images as a more native and non-intrusive way to reach a targeted audience on behalf of brands.
This unique "in-image advertising" format allows marketers to deliver contextually relevant ads next to visual content users are actively looking at. We work with 70% of Fortune 100 companies, from L’Oreal to Disney and Sprint. In addition to our advertising business, we use our semantics and computer vision capabilities to determine the value of a brand’s visual content in sports sponsorship or social media.
We’ve built sophisticated proprietary technology and solutions going back to 2008. This gives us an advantage of having a trove of historical data and knowledge that we’ve applied to break new ground, such as with social media and most recently the sports sponsorship category. To that last point, our new VP of sales at GumGum Sports, Ian Partilla, had this to say:
“I knew GumGum Sports would be the one product that finally identifies all the activity (beyond the broadcast) and changes the game. I’ve worked in the sports sponsorship space my entire career and, until now, had never found a tool that was able to provide a true omni-channel view of properties.”
Because image and video content is only on the rise, we see open fields of opportunity to apply our computer vision technology to a range of industries.
Who are your prototype customers in terms of company size and industries?
Currently, GumGum’s customers (rights holders, brands, and publishers) are the world’s largest brands, including the majority of the Fortune 500. Clients include T-Mobile, Disney, and Clorox. GumGum Advertising has integrations with more than 2,000 premium publishers. GumGum Sports works directly with rights holders such as Madison Square Garden, Mets, and New Orleans Pelicans.
The common thread is businesses looking to create more value through visual intelligence to help drive customer recruitment and retention.
What are the primary use cases of GumGum for marketers and sales professionals?
It’s estimated that by 2022, the number of embedded cameras will triple to 45 billion. Most of the pictures captured by your phone, fridge or car will never be seen by a human eye. Generally speaking, the world—or more immediately, our clients—are not equipped to process and make sense of this sheer volume of visual content.
We have a few broad use cases of GumGum solutions.
First, As mentioned above, our proprietary computer vision technology scans images and videos across multiple platforms, allowing you to place contextually relevant ads (In-image) where users are most likely to see them. We use computer vision to scan images, videos, and the content that surrounds them—across millions of pages all over the web. Ads are then integrated within visual content, where our technology knows that customers are paying the most attention. Marketers (and their sales teams) love this capability because it creates a targeted, highly relevant, and non-intrusive brand experience.
Second, in addition to in-image advertising, marketers with a heavy social profile featuring both owned and non-owned content are increasingly turning to computer vision to surface relevant content. We call this GumGum Insights.
For example, Pepsi may want to locate user-generated content across social that includes visuals of the Pepsi logo. Previously, the brand is reliant on contextual searching only - hashtags, Pepsi tagged in the photo or in the description - but brands can’t bet on consumers following those rules. Computer vision enables brands/marketers to surface relevant content just from logo recognition, no words required. This is accomplished by feeding the logo into GumGum’s AI tech stack and using NLP and ML training to recognize certain images based on set parameters. This solution helps brands and marketers stay on top of positive or negative visual sentiment around their brand.
Which, brings us to the third core use case, sports sponsorship and measurement. Once GumGum’s computer vision technology tackled social media, the team saw an opportunity to apply this application to the world of global sports sponsorships, which is estimated at over $60 billion annually. As a result, GumGum Sports was born as a new division of GumGum.
Through patented computer vision technology, we are streamlining the measurement of sports sponsorships. Our technology reduces turnaround times and produces more accurate and consistent valuations across all channels. We provide a holistic view of media valuation, including non owned and operated accounts, unlocking the value of millions of videos and images being shared by thousands of fans.
To see an example, visit this infographic from the Emirates FA Cup Final from last year. Another fun example of GumGum Sports at work is the NBA Scoreboard, a benchmarking tool for analyzing media value generated from NBA team-owned social media accounts.
Any other thoughts on AI in marketing, or advice for marketers who are just starting to explore the possibilities of AI?
AI takes a lot of shapes within marketing - and is most certainly not all created equal. The AI use case that seems to get a lion’s share of coverage is chatbots but Marketing AI practices in computer vision will see an explosion in use cases in the coming years.
As we communicate more and more with visual content, and restrictions on ad formats continue to evolve, monetizing images through data/artificial intelligence, is going to boom. This means have staff that is comfortable navigating the technical worlds of AI - and helping teammates translate that into wins for clients.
Paul Roetzer
Paul Roetzer is founder and CEO of Marketing AI Institute. He is the author of Marketing Artificial Intelligence (Matt Holt Books, 2022) The Marketing Performance Blueprint (Wiley, 2014) and The Marketing Agency Blueprint (Wiley, 2012); and creator of the Marketing AI Conference (MAICON).