Cannes Lions

Image based targeting using AI

CLUEP, Toronto / Cluep / 2018

Presentation Image
Supporting Images
Supporting Images

Overview

Entries

Credits

Overview

Description

Cluep Pics is a mobile ad platform that delivers advertising to people based on the images they post on Twitter, Instagram and Facebook. With access to the full fire-hose of public posts on these platforms, Cluep analyzes these images within milliseconds of being posted to identify products, brands, people, activities, animals and scenarios. Cluep then uses a proprietary device ID matching model to follow these individuals into their other apps and browsers, where it serves insight-driven advertising. For example, share a photo of your puppy, get an ad (standard, rich media or video) for puppy food; share one of your adult dog, get served an adult dog food message. If desired, exclude overweight dog images. This personalization at scale ensures consumers have more relevant experiences resulting in greater engagement, better brand perceptions and more efficient programs for Marketers.

Execution

Cluep built an image recognition engine that uses convolutional neural networks (CNNs). It is inspired from what we know about the architecture of the human visual system and uses a mathematical process known as convolution to be able to analyze images in non-literal ways, such as identifying a partially obscured object or one that is viewable only from certain angles. We feed our convolutional neural network vast amounts of training data, labeled by humans so that it can essentially fact-check itself as it’s learning. Let’s say this labeled data consists of pictures of pizzas and burgers. The pictures are the data; “pizza” and “burger” are the labels, depending on the picture. As pictures are fed in, the convolutional neural network breaks them down into their most basic components, i.e. edges, textures and shapes. As the picture propagates through the convolutional neural network, these basic components are combined to form more abstract concepts, i.e. curves and different colours which, when combined further, start to look like an entire pizza, or a burger. At the end of this process, the convolutional neural network makes a prediction with 75%+ accuracy as to what’s in the picture.

Outcome

Although Cluep has run 321 Pics campaigns for 196 Brands in the US, Canada, India, and Asia, there are many factors that point to exceptional growth.

• People’s shared images provide timely insight that supports various brand and business building strategies, including driving sales among category users, competitive conquesting, establishing brand relationships during transitional life events etc.

• Applied to virtually any type of product use, lifestyle or context, meaning that the client categories for which Pics can be applied are practically unlimited

• Accessible for all clients – Pics can run at varying scale, from $5K pilots to $500K+ programs

• Only targeting people who are publicly sharing their images allow marketers to be hyper-personal while respecting privacy

This is supported by proven performance: 2.1% CTR average (4.8X industry norms), completed video views above norms, effective on-site traffic and behavior, and back-to-store traffic are driving on-going client business.