Let’s face it. Simply delivering a digital interaction in real-time conversational channels such as web, chat or mobile with the technology that brands have today isn’t that difficult. The challenge is delivering something that is anticipated, personalized and useful — or in other terms — relevant to the consumer at the current stage in their customer journey.
Brands recognize that this is a challenge. SAS, in a recent report alongside Accenture and Intel, found that more than 80% of executives surveyed by Harvard Business Review Analytic Services said they want to use analytics to generate real-time actions from customer data. However, only 22% of respondents in the same study said their organizations are effective at using analytics and data.
That is a huge gap between desire and reality. Why is this the case? It’s because being able to deliver contextually relevant real-time interactions across all channels can be a daunting task, and it’s a task that must account for many variables. It requires good data, a variety of analytics, integration with a plethora of endpoints and the ability to execute at lightning speed. Easy to write, much harder to do.
Let’s talk about what it takes to power a truly real-time customer experience and get some insights on what consumers think vs. the reality for brands.
What Is 'Good Data?'
I think that the definition of “good data” is very subjective based on the industry, geography and business model in which you operate. For example, digitally native organizations probably don’t think twice about collecting volumes of digital data off apps and owned web properties. For more of a legacy brand, it may be a real challenge due to legacy technologies, organizational structures and process siloes.
- Consumer Perception: Data is easy to manage and activate across organizations. Brand X should have the data to know I just called the contact center yesterday when engaging with me via chat today.
- Brand Reality: “Good data” is hard to come by and requires many things to be mastered.
What constitutes “good data?" In my mind it’s three things — variety, quality and depth.
- First is data variety spanning from zero to third-party data. There are varying levels of importance, but a general rule of thumb is that zero-party data is the most valuable, followed in order by first, second and third. All data types are needed to build a complete view of a customer, from data that is provided by the consumer (zero), to data that is collected (first), to data that is shared by other businesses (second) to data is that collected indirectly (third). Having all types (basic identity or demographic data, engagement data across channels, customer behavioral data, attitudinal or VoC data, etc.) of customer-related data with a focus on zero- and first-party data is the first component of “good data."
- Next, is data quality. Simply put, your data must be complete and correct. If you have holes or gaps in data, techniques like synthetic data generation can be used to help augment data to create high-quality data profiles. Poor quality data results in incorrect and often very misleading results from downstream applications, such as customer engagement technologies.
- Finally, the depth of data is important. When I say depth, I mean more than the last engagement. Contact and response history from multiple interactions over months if not years and across both channels (web, mobile, chat, email, social, etc.), as well as departments (sales, marketing, service, support, etc.), helps to create a more holistic view of a customer.
Related Article: NLP and Text Analytics Enhance VoC Programs, Boost CX Engagement
A Different Type of Analytics
So, once you’ve got the data, what comes next to power a real-time, yet conversational customer experience? Well, it’s going to involve a variety of analytics, some of which you may be familiar with and others that are rather rare.
- Consumer Perception: Analyzing customer data for optimal engagement is straightforward; brands should be able to analyze exactly what I need next.
- Brand Reality: Applying and layering analytics on top of data involves the right people, processes, technologies and experience for real “business changing” insight to be derived.
The report we cited earlier states, “Many organizations already capture or have the means to capture this information, but few have the tools to promptly act on it. They may store the data for batch analytical processing — or simply throw it away.”
The key phrase to call out here is “batch analytical processing — or simply throw it away.” Why is this the case? Well, it’s because data requires the proper processing that can often involve a sequence of steps. Data log files come in aggregate and often sit stagnant until acted upon. The trick lies in being able to automate the incoming data translation process via data management techniques combined with predictive and prescriptive analytics.
And for true conversational customer experience, once the inbound data is translated, it must be queued and available for resultant outbound communication. To do this comprehensively the analytical types can be many — and sometimes daunting. These include but are not limited to:
- Streaming Analytics: Collect data from event streams on IOT-style devices. This can include usage information and behavior, location, device statistics, etc. Users would have to opt-in to this data collection.
- Text Analytics and Sentiment Analysis: This is a must-have for conversational AI. By analyzing chat text strings and the sentiment in those text strings, brands can understand customer attitudes and intent.
- Natural Language Processing and Generation: The ability to process natural language data (NLP) coming from documents such as chats and convert speech-based conversations into natural language text (NLG) are foundational components of conversational AI as well.
- Computer Vision: Images shared by consumers over conversational channels must be quickly analyzed and tagged to provide additional context to the dialogue.
Related Article: Voice of the Customer: What Is It and Why Does It Matter for CX?
Going Beyond Simple 'Conversational AI'
The saying often goes, “Are you listening to respond, or are you listening to understand?" Some conversational AI only works well in “response” mode. Simple, rules-based and often very frustrating for the customer. But the fact is, true and complete customer understanding is what drives long-term brand loyalty and trust.
- Consumer Perception: This brand doesn’t get me. If I type “warranty” in this chatbot, the response will be a laundry list of FAQs that contain the word warranty. I just bought this consumer electronic, and it’s already broken. What they need to understand is that I won’t ever be purchasing from them again.
- Brand Reality: We don’t have the technologies and processes in place to do anything more than a simple query based on a word or text string and return all related content!
So, how do you truly understand your customers — to serve and satisfy their needs? Well, the beginning comes through solid data and analytics practices and processes, some of which we have mentioned above. But going beyond conversational channel-based AI involves rethinking organizational models and processes. It involves thinking comprehensively about your organization. This often involves having individuals and/or teams consider the “enterprise view” of customer experience.
It involves asking questions such as:
- How do interactions on marketing channels (chat, social, email, etc.) impact overall customer loyalty and lifetime value?
- How do interactions in other areas of the business — such as a customer-specific risk or fraud decision (declining a customer for a loan) impact future engagement activities with customers?
- How do service and support inquiries from customers impact the product and services offered to those customers — and at what time intervals should we offer them?
Enterprise decisioning technologies play a key role in going beyond just asking the questions but being able to understand and activate a response. These technologies rely on analytic decisions, arbitrations and optimizations to detect a data change, analyze that change and optimize the ideal time and interaction point in which to deliver a decision or response to that change.
Thinking Beyond Channel-Specific Customer Experiences
That takes us beyond thinking in a channel-specific manner and instead treats the customer experience and the interaction points along the journey within that experience in a more holistic manner.
Seeing things holistically allows brands to guide their customers to end conversion events by using techniques such as customer journey orchestration and guided AI techniques, instead of forcing or pushing them down a channel-specific, organizationally defined path. The result becomes a happier customer and a better experience with the brand.