The Next Data Revolution is Here – AI Will Understand What Isn’t Being Said

Print Friendly, PDF & Email

In this special guest feature, Ran Margaliot, COO and VP R&D for Affogata, discusses how analyzing and sorting through unstructured data saves countless hours and recognizes patterns in seconds that even skilled data professionals may never uncover. Affogata is a leading Customer Intelligence Platform that enables brands to be truly customer-obsessed by bringing their customers’ voice to the right teams in the organization. Affogata’s powerful portfolio of clients includes names like, Fiverr, Lemonade, eToro, MyHeritage, Plarium, and more. Previously, Ran was Co-Founder of Communit360, a social media management and intelligence platform, which helped thousands of marketers and small businesses increase their ROI from their social media activities.

Imagine for a moment an evolutionary chart from hunched over ape to modern-day humans, the way data is leveraged is similar. First, we had the most unevolved, manually combing through data for insights wasting countless hours and speeding up the process simply by adding more people. The second step in the evolution was being able to apply “logic” to the process [i.e. if X is mentioned then take Y action], which was followed by all sorts of AI and machine learning processes which gathered data thanks to APIs and simply gave literal interpretations. Now, we’re beginning the next phase, analysis of unstructured data, which enables companies to use AI to predict and understand crucial context that is neither spoken nor written.

What exactly is unstructured and semi-structured data anyway?

That cliche “data is the new oil” – whether true or not – highlights the indisputable fact that data provides countless businesses with critical insights, but we’re just scratching the surface. In big data, unstructured and semi-structured data [i.e. media, images, audio, etc.] is what appears the most, comprising 80-90% of all data, yet its potential remains untapped. Every second we’re sending voice notes, emojis, and GIFs that convey crucial context and meaning, yet all of it is invisible to AI models that gather and analyze structured data alone.

Tapping into unstructured data is more important than ever

With agile development driving rapid product updates, constantly shifting privacy regulations, and occasional outages affecting third-party marketing channels, and PR crises one mistake away, knowing the customer is a requirement.  That means truly understanding what the customer thinks in real-time by going beyond simple direct mentions. Leveraging unstructured data provides the ability to identify indirect brand mentions and segment those mentions by topic, keywords, sentiment, intent. Having the full picture leads to better product development, more robust marketing insights and initiatives, particularly across owned channels, and avoiding a crisis by understanding users’ frustrations and intent in real-time.

Applying AI to data analysis involves comparing datasets, looking for patterns, key differences, etc. that reveal meaning, and using various data preparation techniques to turn unstructured data into a format that machines can understand. Particularly aggressive language from a customer might mean that they will not engage with a brand again even though they didn’t literally say it. Repeating cases with language that is aggressive but wholly inaccurate may reveal manipulation or that a problem isn’t worth addressing because there is a low probability they’re actually a customer. Data scientists build a data model using training data first, teaching it how to recognize patterns that indicate certain types of reaction, sentiment, opinion, etc. not just literal meaning. Subtext often says more than the words themselves.

Online insights have real-world impact

In a rapidly growing gaming industry, whose value stands to almost double to $314 billion by 2026, unstructured data insights can lead to understanding player difficulties before they abandon the game and sizing up updates in real-time. In fintech, it translates to truly understanding increasingly powerful online forums that speak in slang, subtext, and inferences to better prepare for a rush towards particular stocks and avoid a subsequent Robinhood-esque PR crisis. And when it comes to marketing, we’ve seen first hand not only when it comes to better targeting customers by knowing their true preferences, but also with an ancestry platform client that leverages unstructured data to connect long-lost relatives and publicizes the stories.

The potential of unstructured data is vast

First and foremost, unstructured data fuels consumer intelligence platforms that allow companies to better understand their customer, bolstering product development, customer success, and marketing functions, but unstructured data’s potential doesn’t stop there. Additional applications are wildly creative and have a significant potential impact. Substrata for example, which takes the analysis of unstructured data in another direction entirely. Instead of using it to understand customers online, they apply the technology to sales conversations to understand the subtext behind how an email is written, the time it takes to send, and the hidden meanings that dictate power dynamics on calls. Further applications include voice or image data sent between medical professional and patient, collected by IoT devices, or present in various business documents – the possibilities are endless.

The AI-powered data revolution is already here, and it’s up to data analysts and multiple teams across organizations to embrace the monumental shift to reap the benefits and get ahead. Analyzing and sorting through unstructured data saves countless hours and recognizes patterns in seconds that even skilled data professionals may never uncover. That data “superpower” can lead to better products that adapt in real-time, responsive customer service,  and a sharing of insights across an organization magnifies its impact even further. The impact is well-demonstrated, and there is plenty of data and case studies to prove it; the final step is for human decision-makers to make the decision sooner rather than later.

Sign up for the free insideBIGDATA newsletter.

Join us on Twitter: @InsideBigData1 –

Speak Your Mind