There’s a galactic amount of data being produced on social media, but the ability to categorise it into meaningful information is “near impossible” without Artificial Intelligence (AI). This is the challenge that Crimson Hexagon is taking on, and a brief demo of its abilities at #DMWF by senior sales executive Yugen Moodley shows the “deep insights” platform is doing a scarily good job.
Crimson uses a combination of Natural Language Processing, image recognition and sentiment analysis, and the result for its customers is practically limitless possibilities to connect, grow and learn from their audiences in ways they may have never even considered possible.
Giving attendees a taste of these capabilities and touching on the company’s most recent developments in image recognition, Moodley explained how the platform can analyse photos shared from an event, such as a festival, to identify the prevalence of certain clothing brands. It doesn’t take a lot to imagine the immense possibilities of this, opening doors for real-time adjustment of on-site sponsorship campaigns and offers and informing future sponsorship decisions based on competitor events.
But with Moodley’s session just scratching the surface; MarketingTech caught Crimson Hexagon’s sales manager, EMEA & APAC, James Humphrey, for a deeper dive into what the platform can do and how it makes it possible.
First of all, could you give us an overview of what Crimson Hexagon does, or perhaps more appropriately, how clients can use the platform?
James Humphrey: We empower our customers to make informed decisions that genuinely impact top and bottom line revenue. How we do it is by providing them with a suite of AI [artificial intelligence]-powered tools that contextualise and quantify over one trillion consumer conversations and images across millions of sources, including social media, blogs and forums.
Our customers use the platform for a plethora of use cases but the most common include audience analysis, brand analysis, competitor analysis, campaign tracking, customer care and of course industry and trend analysis.
We’re talking about masses amounts of data being processed a day – something like one trillion social media posts and over 160m+ photographs. In simple terms (if that’s even possible!), how is any of this possible?
JH: Immensely large data centres and very clever AI and machine learning. In the past, analysts and digital marketers would have had to manually tag and classify posts as being positive or negative in sentiment. That approach became a great deal easier with the advent of natural language processing – which infers the sentiment of a post or comment by relying on a predefined library of keywords. This changed the game a little but as a technology, it’s fundamentally flawed. A predefined library of keywords is by definition, restrictive. Language patterns shift all the time. Words that used to have a negative association now can be used to describe things in a positive light, for example, “those new jeans are sick bruv”. Without continually updating these libraries, the accuracy will diminish over time. Another flaw can be found in the answers NLP [natural language processing] gives you. At the end of all that analysis, of potentially thousands of conversations, the answer is a percentage of positive or negative sentiment. That’s not particularly meaningful and it’s certainly not very actionable.
So, what sorts of insights can be gleaned from this information, and how can marketers and advertisers use them for commercial gain?
JH: Examples of actions our customers have taken from this insight are improved messaging in campaigns, better-aligned product portfolios and roadmaps, improved competitive positioning and crisis management to name just a few. These actions directly impact the business. We’ve seen increased intent to purchase, reduced customer churn, new market entry and new products being developed and released by our customers as a result of them. Not only that, this insight is now available to them in real time. No longer do organisations have to rely on lengthy and costly consumer surveys to find out what their customers are saying! It’s available right now!
The uses don’t stop in advertising though, there are surely broad applications, such as in sociology, politics, even anthropology?
JH: Funnily enough, one of the earliest use cases of our platform didn’t have anything to do with social data. The World Health Organisation [WHO] approached our founder, Professor Gary King with a fascinating challenge. They had collected data from thousands of doctors and nurses across sub-Saharan Africa and were struggling to make sense of it. There were no consistent methods of writing up a doctor’s encounter with a patient so the WHO had no way of accurately predicting where there may be an outbreak of a particular disease or illness.
Professor King hypothesised that he could create an algorithm capable of identifying patterns in the syntax that could be used to measure these vast volumes of unstructured data.
I actually love this story, I still tell it all the time. It’s a great differentiator in our industry and it’s a wonderful analogy as to what digital marketers are trying to do with unstructured, unsolicited consumer opinion. How do we make sense of a trillion voices? How do we use that information to improve our business and the experience our customers have with us?
So yes, the potential use cases are not just reserved for digital marketers and advertising professionals. They are vast and incredibly exciting.
That leads on to a broad but poignant question; what role do you believe online data should play in society and how is that guiding Crimson Hexagon’s mission?
JH: That is indeed a poignant question. It’s also a great divider. On the one hand, I think consumers, like me, like the idea that what we do online is our business. We should have the right to keep what we do online private and to a certain extent, we do. On the other hand, I want the companies I love to develop products I love. We [Crimson Hexagon] help our customers develop campaigns, products and services that are better aligned to what their customers actually want. I’m hopeful we live in an age that the consumer’s voice is at the forefront of every decision a business makes. Should they enter a new market? Should they venture into a new product space? Should they leverage this potentially dividing celebrity for their next campaign? Why would you leave the answers to these questions to chance when you can have them in an instant?
As a society, there is still a big gap between the prevalence and power of online data, and our understanding of how it is used. In just the last few years, we have seen this conversation grow exponentially — among consumers, companies and policy-makers — but it is clearer than ever that this discussion is just beginning.
At Crimson Hexagon, we believe strongly in having, and contributing to, this conversation. There is a fair amount of black and white here — private data should never be analysed, for example — but there is also a lot of grey area still out there. Should government agencies be able to analyse public online data? What about individual politicians? Ultimately, these are questions that we need to answer as a society.
With all that said, we believe strongly in Crimson’s core mission: to help brands and agencies better understand what consumers are saying. This can help them design more useful products, create more compelling marketing materials, and provide more thoughtful and timely customer care.
This month, Crimson Hexagon launched new reverse image search technology. Could you tell us a bit more about this and how brands could use it in practical terms?
JH: Absolutely! There’s no denying that images are being shared more and more and in parallel, text is getting shorter and shorter. As you pointed out earlier, we are now indexing over 160 million images every single day. Last year we developed a cutting edge way to contextualise information in images and this year we’ve just launched our ‘reverse image analysis’ product.
This new product allows you to take any image, be it from social or from your own internal data and use it as your search term! You could identify where a campaign image has been shared and drill down into the psychographics of those sharing it. You could even look at your own customer care data for images being shared of faulty or damaged products. It’s early days but we’re really excited to see what our customers start doing with this new technology.
Finally, are there any developments or announcements on the horizon for Crimson Hexagon that you’re able to share with us today?
JH: We have exceptionally talented and driven in-house product and engineering teams that are working around the clock on a product roadmap that continually blows my mind. I am definitely not going to share any of that today! If your readers would like to learn more, give us a call. We’re always happy to geek out!
Digital Marketing World Forum (#DMWF) returns this November 7-8 for the North American leg of the global conference and expo series in New York. Find out more here.
