Predict How Consumers Will React to Content Using Machine Learning
What if you could predict which types and versions of content would resonate best with your audience? Or walk into a client or executive pitch meeting with data that backs up each of your creative choices before those choices are made?
Those kinds of oracular powers could mean millions more in the bank for companies that use content to fuel leads, conversions and sales.
That’s the goal of Cortex (@meetcortex), a suite of artificial intelligence, machine learning, and natural language generation and processing technologies designed to turn creative content marketing into a fully data-driven process. We learned how by talking to Brennan White (@Brenomics), Cortex’s CEO.
1. In a single sentence or statement, describe Cortex.
Our software helps social and content marketers make the right decisions about content creation, deployment and promotion.
2. How does Cortex use artificial intelligence?
Cortex uses machine learning to predict what reactions consumers will have to various aspects of content. What colors work best for photos, what timing works best on Tuesdays, what the cost of promotion will be at that time of day.
We use natural language processing (NLP) to break written content down into software-understandable database entries to find patterns about what works best for various audiences.
We use machine vision (deep learning) to automatically understand what is in photo and video content, what colors are used, the size, the brightness, and resolution of the photo and create optimal plans for what content should be created for that brand.
We are using natural language generation (NLG) in our staging environment to be able to create basic captions for photos.
And we use machine learning to evolve many of the machine-generated content deployment calendars and arrive at the optimal choice without needing to A/B test.
3. What are the primary use cases of Cortex for marketers?
Cortex has two major uses cases:
First is to help marketers keep up with the sheer volume of social media content that companies need to produce to stay competitive in this saturated environment.
Every customer gets custom machine-generated content calendars and recommendations about what to post. We also integrate with Getty Images and the client’s digital asset manager to automate photo selection.
The result is a single marketer being able to finish a week of social media posts in an hour or less. Once marketers are able to easily handle the volume of content they need to create, our focus shifts to helping them produce stellar results.
We look at their historical data, competitors, and industry to identify what content is resonating with their audience, then deconstruct what subjects, colors, aspect ratios, messaging, and other creative elements are performing for each channel. This way the creative team can be sure every piece of content they create will inspire their audience to action.
4. Who are your ideal customers in terms of company sizes and industries?
We work with over 60 major brands. While Cortex works for companies of any size and we have a pricing model that supports that, we focus on large brands with multiple products, brands, or locations (fleets of assets) that are marketed online.
With Cortex, those organizations receive all of the benefits of other companies, plus our automation and alerts have made it simple to manage the entire fleet.
5. What makes your company different than competing or traditional solutions?
We're the first software to combine all of the new tools (machine learning, deep learning, NLP, NLG, etc.) into one workflow for marketers.
We've built the self-driving car for marketing. It empowers marketers to be more effective and support their work with data. They'll never again have to redo a creative effort because someone thought it “didn't feel right.” Numbers will support and defend marketers from the ephemeral winds of internal politics and opinions.
6. What do you see as the limitations of artificial intelligence as it exists today?
For a long time, the limitation was technological. Today the limitation is cultural. Businesses are adopting it quickly and the speed of adoption is up to each organization.
Technologically, having relevant data sets large enough (one of Cortex’s secrets) will always be a hurdle.
7. What do you see as the future potential of artificial intelligence in marketing?
Further deployment through creativity. Cortex has taken creativity out of the realm of magic, something unable to be understood or measured, and placed it firmly in the realm of data-driven decision-making. As we advance down this path, marketers will become more powerful in their organizations by being able to support their creative choices with data and prove the ROI of their ideas.
8. Any other thoughts on AI in marketing, or advice for marketers who are just starting to explore the possibilities of AI?
The biggest thought I'd share is a word of warning. Don't buy into huge deployments and expensive AI and machine learning projects. The correct use of these technologies should feel like a self-driving car—sign up, set goals, sit back and start benefiting from a better strategy and better results.
If someone is selling you something complicated or with a long lead time, tell them to build their product first. As a benchmark, Cortex landed a $24B beverage company recently. We had our system optimizing five of their brands and had on-boarded 43 users in less than 48 hours. The experience with machine learning and AI should be easier and more effective.
About Paul Roetzer
Paul Roetzer (@paulroetzer) is founder and CEO of PR 20/20, author of The Marketing Performance Blueprint and The Marketing Agency Blueprint, and creator of The Marketing Artificial Intelligence Institute and Marketing Score. Full bio.