The Power of Zero-Party, Unstructured Data

Building a Multimodal Customer 360 in an AI-Driven World

I am delighted to introduce Snowplow’s own John Bourous, Senior Partner Marketing Manager, with an extremely timely take on unstructured data and the rise of multimodality:

In the ever-evolving landscape of AI and customer experience, a new paradigm is emerging. While structured data has long been the cornerstone of data-driven decision making, zero-party unstructured data is now playing a significant role. It offers deeper insights to build more personalized experiences than ever before.

Unstructured data joins the Mad Hatter’s tea party, courtesy of Midjourney 6.1

This shift represents more than just a passing trend. It is reshaping how businesses interact with customers on various platforms such as social media, video, chat, AI, and other technologies. 

So, how can businesses make the most of this new data type to enhance customer experiences and stay ahead in the AI-driven world? Let’s explore. 

What Does Multimodal Mean for the Customer Experience

Recently, you may have heard the buzz around multimodal large language models (LLMs), such as OpenAI’s GPT-4o. These advanced models can read and reply with a wide array of data formats, from PDFs and images to diagrams and software code. In a nutshell, multimodal data refers to information that combines different formats like text, images, audio, and numbers. Today, experts are bullish on the prospect of multimodal LLMs reshaping industries such as healthcare, robotics, logistics, and other data-driven sectors.

In marketing, having a multimodal customer data asset is essential in today’s AI-driven world.

But what is multimodal data exactly? In basic terms, it combines structured data, like purchase history, with zero-party unstructured data, like customer service chats. Each of these different types of data is considered a "modality" and when combined creates a sum greater than its individual parts. 

By blending these complementary data types, organizations can unlock unprecedented insights into customer behavior, preferences, and needs. This holistic approach not only provides a 360-degree view of the customer but also enables AI-driven systems to deliver hyper-personalized experiences, predict customer intentions with greater accuracy, and identify emerging trends before they become apparent in structured data alone.

With “multimodal” now defined, let’s revisit first-party structured data and the part it plays in any Customer 360:

Pros and Cons of First-Party Structured Data

Structured data has always been important for businesses. It fits nicely into databases, making it easy to track through cookies and pixels. Information on device type, locations, and purchase history is simple to collect and analyze. This is why it has traditionally been a go to source for business intelligence and customer relationship management, and played a key role in decision-making. 

We use “first-party” to indicate that this activity data is collected directly by a business from its own customers or users through their interactions with the company's products, services, or platforms. This is distinct from second- or third-party data which is mediated by some other provider, with differing levels of consent from the end consumer.

However, first-party structured data has its limitations. It can tell you what a customer bought, when they bought it, and perhaps even predict what they might buy next, based on patterns. But it struggles to capture the nuances of customer intent, preferences, and sentiments, especially in an AI-driven future where customers can interface with brands in new and exciting ways. Any additional context that you can capture about your user could help determine whether or not they ultimately purchase your product, churn from your service, or deliver negative sentiment with their social networks.

Today, every ambitious brand is trying to build a competitive advantage in this new era of customer experience by harnessing the power of AI. It feels like a modern-day gold rush, as pickaxes and shovels are replaced with ML models and data warehousing. However as data scientists tap into the same set of vector databases, open-source models, and other increasingly commoditized tools, the reality is that the underlying data will become your most important, proprietary asset. In order to truly unlock the potential of this data asset, companies need to move beyond the limited and inflexible first-party data tracking imposed by the vendors of yesterday and expand their contextual understanding of the end user. 

Enter Zero-Party Unstructured Data, A Messy Resource

Unstructured data is the wild west of information - it is raw, diverse, and doesn't conform to predefined data models. It includes everything from social media posts and customer reviews to email communications, chatbot conversations, and even audio and video content.

We use “zero-party” to reflect that this data is voluntarily and proactively shared by customers with a brand or business - it is not mediated by any third-party, or even by the brand itself.

While historically challenging to model, store, and analyze using traditional methods, zero-party unstructured data holds a wealth of insights that can dramatically enhance customer experiences when properly distilled and leveraged. 

Zero-party data is incredibly valuable because we can now learn the intention of a user directly. What are they trying to do in our app / website / store right now? This is something that as long as digital analytics has been around, we've been trying to figure out indirectly. Advancements with AI are really exciting as organizations can now ask questions like “What are you doing here?” and act on that information at a scale never seen before.

Examples of Harnessing and Capturing Zero-Party Unstructured Data

The beauty of zero-party unstructured data lies in its ubiquity and richness. Here are some key ways businesses are capturing this valuable resource:

1. Search & Discovery: Every search query on a website or mobile for a product, item or otherwise contains unstructured data revealing high user intent.

2. AI Chatbots: Conversations with AI chatbots provide a goldmine of unstructured data, offering insights into customer queries, pain points, and satisfaction levels

3. Social Media Monitoring: Tracking mentions, comments, and shares across social platforms captures real-time customer sentiments and trends

4. Customer Support Interactions: Emails, chat logs, and call transcripts contain valuable unstructured data about customer issues and experiences

5. User-Generated Content: Reviews, forum posts, and blog comments offer authentic, unsolicited customer feedback

The Blending of Structured and Unstructured Data

When combined, structured and unstructured data create a powerful synergy that can transform customer experiences. Let’s look at five use cases which are significantly enhanced by this multimodal approach to customer data:

1. Multimodal Customer 360

This use case is so powerful we referenced it in our article title: imagine combining a customer's purchase history (structured data) with their recent product reviews and support chat logs (unstructured data). This creates a more expansive and holistic 360-degree view of the customer, allowing businesses to understand not just what the customer bought, but why they bought it, how they feel about it, and what they need from us next.

2. Predictive Analytics

By analyzing patterns in both structured and unstructured data, AI can make more accurate predictions about customer behavior. For example, a sudden increase in negative sentiment detected in social media posts (unstructured) combined with a dip in sales (structured) could predict a potential churn risk, allowing businesses to proactively address issues.

3. Personalized Recommendations

Unstructured data from product searches and chatbot interactions can reveal preferences and intents that aren't captured in structured browse and purchase data. This allows for hyper-personalized product recommendations that go beyond simple "customers who bought X also bought Y" algorithms.

4. Improved Customer Support

By analyzing unstructured data from previous support interactions, AI can identify common issues and suggest solutions more quickly. This can be combined with structured data about the customer's history to provide more context-aware and efficient support.

5. Product Development Insights

Unstructured data from customer feedback and product reviews can reveal pain points and desired features that might not be apparent from sales data alone. This can inform product development decisions, leading to improvements in the overall experience that truly resonate with customers.

Challenges and Considerations

While the potential of zero-party unstructured data is immense, there are challenges to overcome:

1. Data Privacy and Ethics: Zero-party unstructured data is highly sensitive - it is the direct verbal expression of a customer’s opinions, intents and preferences. Businesses must be highly vigilant about privacy concerns and ethical use of all zero-party data

2. Data Quality and Relevance: Not all unstructured data is equally valuable. Businesses need to develop strategies to filter out noise and focus on meaningful insights

3. Tooling & Skills Gap: Processing and analyzing unstructured data often requires different tools and skills than traditional data analysis, potentially necessitating new hires or training

4. Integration Complexity: Combining structured and unstructured data into a multimodal asset requires sophisticated data integration and analysis tools

The Future of AI-Driven Customer Experiences

As AI technologies continue to advance, the integration of structured and unstructured data will become seamless, leading to even more transformative customer experiences:

• Emotion-driven AI: By analyzing unstructured data from voice and video interactions, AI agents will be able to detect and respond to customer emotions in real-time.

• Predictive Customer Service: AI will proactively reach out to customers to solve problems before they even realize they have one, based on patterns detected in unstructured data.

• Augmented Reality Experiences: Combining structured data about product specifications with unstructured data about customer preferences will enable highly personalized AR shopping experiences.

• Natural Language Interfaces: As AI becomes better at understanding context from unstructured data, interactions with businesses will become more conversational and intuitive, mediated by AI shopping assistants.

Conclusion

The future of AI-driven customer experiences lies in the synergistic integration of structured and zero-party unstructured data. While structured data provides a solid foundation, unstructured data adds the color, texture, and depth needed to create truly personalized and engaging experiences. Businesses that can effectively harness both types of data will be well-positioned to lead in the era of AI-powered customer centricity.

When successful, these businesses will be able to leverage a multimodal customer data asset in this AI-driven world. This asset will not just provide the richest-yet 360-degree view of the customer, but also enable hyper-personalized experiences, accurately predict customer intentions, and enable rich agentic experiences which are themselves multimodal.

As we move forward, the line between structured and unstructured data will likely blur, with AI systems becoming increasingly adept at extracting structured insights from unstructured sources. This convergence will open up new possibilities for understanding and serving customers in ways we can only begin to imagine. The businesses that embrace this multimodal data revolution today will be the customer experience leaders of tomorrow.