The billions of selfies, glamour shots, and candids that we take each day and publicize on social media are a vast treasure trove of data when you think about it. They display what we’re doing, what we like, and who we are (at least, demographically). Netra, a startup with nearly $2.5 million in funding, wants to use all that information to improve your marketing.
Leveraging convolutional neural networks, a type of advanced AI, Netra analyzes visual content on social media to help brands and agencies better understand and reach their ideal customers. We spoke with CEO Richard Lee (LinkedIn) to learn how Netra achieves results for marketing and sales professionals using AI in social media.
In a single sentence or statement, describe Netra.
Netra helps brands and agencies better understand and reach their ideal consumers by analyzing billions of consumers’ photos—and identifying their activities, interests / passions, key life events, demographics of their tribe, and brand preferences.
How does Netra use artificial intelligence (i.e. machine learning, natural language generation, natural language processing, deep learning, etc.)?
Netra uses computer vision, AI, and deep learning to essentially teach computers to see. We’ve created models that detect objects and scenes (over 4,000, including categories and hierarchies), humans (classifying age, gender, and ethnicity), and brands / logos (over 2,000, including categories).
During the training process, we feed annotated imagery (a combination of synthetic and real-world imagery) through convolutional neural networks. This process enables the software to learn how to identify, localize, and classify objects. These are all things that humans could identify—it’s just that our software now enables performing these tasks at the scale of hundreds of millions of images per day (which wouldn’t be manually possible).
What do you see as the limitations of artificial intelligence as it exists today?
We’re still in the early innings for AI. As it exists today, image recognition is good but not perfect, so the industry as a whole needs to continue to invest in improving accuracy which will enable many new use cases.
One other limitation today is that training still requires some supervision to identify new objects. We’re increasing the proportion of synthetic data (and using adversarial networks to create more realistic synthetic imagery), which will significantly decrease our training times over the long run. But we still need to “teach” our software new brands and products, which is time consuming.
What do you see as the future potential of artificial intelligence in marketing and sales?
We believe that AI will eventually touch virtually every aspect of marketing and sales—with the greatest impact in areas with large amounts of unstructured data. We’ve focused initially on social media imagery, as consumers are openly sharing close to five billion pictures per day.
This represents one of the largest sources of consumer behavioral data, yet it’s a data source that’s been essentially untapped thus far. Longer term, we believe consumers will use visual technologies as one of the primary ways that they search for information, as well as shop and transact.
So, even though our near-term focus is on social media, what we’re building is foundational to this future state.
What makes Netra different than competing or traditional solutions?
Netra provides a visual intelligence layer on top of existing text- or hashtag-focused analytics.
Our solution goes beyond logo recognition, as brands show up organically in less than 10% of photos. Netra also identifies objects and scenes (present in over 80% of social photos), and the demographics of people in photos (present in about 50% of images).
We are one of the only providers who can identify brands, context, and humans and believe we are best-in-class across all three models.
Who are your prototype customers in terms of company size and industries?
We’re working with a few types of companies. The first type is large marketing platforms (social listening and marketing clouds) that have layered our visual intelligence technology on top of their own analytics.
Our other customers are agencies and brands (essentially the end users for our technology). For larger agencies and brands, our solutions let them track a portfolio of brands and their followers (not to mention their main competitors). Smaller agencies and brands that often don’t have dedicated insight teams leverage Netra to provide them with low-cost and near real-time insights and intelligence about their brands, their categories, and their consumers.
What are the primary use cases of Netra for marketers and sales professionals?
Our thesis is that consumers take pictures of what’s important to them—their friends and families, their favorite activities, or key life events, as well as key brand preferences. Consumers are sharing more and more photos and videos on social networks, yet to-date, brands and agencies haven’t had the ability to mine this data at any scale. So marketers are using Netra’s visual intelligence technology for understanding consumer insights, monitoring brand health, identifying influencers, as well as targeting audiences based on what’s present in photos.
Any other thoughts on AI in marketing, or advice for marketers who are just starting to explore the possibilities of AI?
AI is going to automate a lot of the data collecting and number crunching that marketing creativity builds off of, so marketers can look forward to investing more of their time and energy in taking action on consumer insights than building them themselves. Our advice for marketers is to dive in with an open mind and get their hands dirty. Explore tools first hand to see the tech in action and really understand the capabilities. A small pilot is a great way to get started.
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).