Conventional wisdom is that Caller ID is captured for each incoming call, and a query to enterprise databases attempts to match the caller to an existing customer record. If there is a “hit,” profile information and historical transactions can be processed in real-time to create differentiated Customer Experience (CX) through routing, self-service, and handling by a live agent.
Yet, too often, Caller ID is unavailable or a match is unattainable, and so default rules are applied, resulting in a one-size-fits-all experience. This happens when an existing customer places a call from a corporate office, where the same Caller ID is associated with all outbound interactions. It happens when a customer uses a mobile phone instead of the landline that is associated with her record in the Customer Relationship Management (CRM) database. But perhaps worst of all, it happens when a prospect is calling to conduct business for the very first time. Matching is impossible because there is no record to match to. So just when it matters most — winning new business from prospects — the customer experience tends to be generic and mundane.
Fortunately, enterprises can overcome these obstacles by using AI and decisioning technologies. Insights gleaned from “context” can be used to orchestrate differentiated experiences instantaneously, even for first-time callers. Some examples elucidate the broader context of “context.”
Leveraging Consumer Marketing Information
Marketing service bureaus (e.g., Experian, Merkle, Acxiom) curate volumes of consumer information indexed by phone numbers and accessible via the RESTful API. Matching on Caller ID could yield pertinent demographic information (e.g., college-grad), purchases (e.g., vitamins), lifestyle attributes (e.g., “jetsetter”), and attitudinal behaviors (e.g., “extrovert”). In the absence of internal information, enterprises can license and use third-party data to gather the
context needed to orchestrate differentiated CX.
Mining Call Metadata
Suppose we know that a caller has already been transferred twice and has now waited on hold for more than 180 seconds to speak with a supervisor. Or that a call originates on a mobile handset that is roaming in the Midwest, where there happens to be a severe thunderstorm. Or that a call came in less than 30 seconds after a previous call from the same handset ended (suggesting poor network coverage).
Call metadata (e.g., location, date, time, duration, queue, port, DNIS) is readily available and provides context about what the caller may be experiencing. Having discerned what already occurred, the enterprise can take proactive steps to determine what happens next.
Processing What’s Been Said
Voice processing solutions powered by elements of AI — natural language understanding, machine learning and analytics — can transcribe speech in near real time and then tag pre-defined entities and intents. It is possible to measure a caller’s fluency, dominance, and engagement. And to assess sentiment from both explicit utterances and changes in volume. What is said and how it is said provide context that enterprises can leverage to orchestrate subsequent experiences — even within a single call.
Context is King
In the past, enterprises relied on matching Caller ID to their own customer records to create differentiated CX. Yet so much more can be done by leveraging the broader context of the call, including third-party consumer information, call metadata, and what’s actually been said in the talk-path. By drawing upon these data sources, enterprises can instantaneously create differentiated CX for both returning- and first-time callers!
About the Author
Michael Sisselman, MBA, M.Sc. is an expert in Customer Experience platforms that delight customers, empower the workforce, and drive business outcomes. He most recently worked at Avaya, where he led a team of data scientists in Conversational AI, Intelligent Automation, and CX Analytics. During the autumn of 2020, Michael developed methodologies pertaining to Value Consulting and Forensic Customer Success.
Michael lives in New York and plays classical guitar every day. For more information, click his LinkedIn page.
Categories: Conversational Intelligence, Intelligent Assistants, Articles
Related Articles
-
Exploring Microsoft’s New Magentic-One Framework: A Flexible Agentic System for Complex Enterprise Tasks
-
InstructLab and RAG: A New Oppportunity for Enterprise AI with IBM’s Granite Models
-
Opus Research Welcomes Ian Jacobs as VP and Lead Analyst
-
Navigating a Future of Automation: Strategic Insights from NICE’s Analyst Summit 2024
Mike —
Thanks for sharing. Being context aware and acting on it is an advantage. Can you say more about “It is possible to measure a caller’s fluency, dominance, and engagement.”
Appreciate the guidance on first-time calls — crucial make-or-break opportunities.
Michael,
I enjoyed your article and perspective on how Caller ID is used relative to the CX. It’s interesting how the area code in Caller ID is becoming less meaningful over time, as individuals keep their numbers after moving. I have so many clients in the San Francisco bay area with New York phone numbers, only because they would never want to change it. Thus, location based campaigns by number are becoming less accurate. Your insight how to leverage the number and focus the CX campaign on the person is spot on. I’ve seen you highlight how much information can be accessed for even a new customer and still provide a personalized customer experience based on their demographics and the business rules.
Interesting article and the power of AI is top of mind for many business executives these days. Agree with what you have here, mostly leveraging customer marketing data and context is king. We all know, a great CX drives loyalty and increase sales and these are great suggestions. Thanks for sharing your insight.
To improve the conversation, I’m sharing comments and replies (slightly edited for clarity) that appeared on Mike’s LinkedIn POST announcing publication of the article by OpusReseach:
Greg Pelton [https://www.linkedin.com/in/greg-pelton-631789/]
It is interesting to contrast the experience of a customer reaching a contact center with that same customer using FaceBook or Google or Amazon. When you interact with a social network, almost everything is known about you because you have been monitored and tracked for years. Conversely, when you reach a contact center there is very little information or insight. It is like predicting the weather using an almanac vs the National Weather Service.
There is movement now for consumers to own their data and control access to it. This is driven by privacy concerns and corporate overreach. If the right technology and regulatory constructs are put in place, why couldn’t contact centers leverage the same infrastructure? Imagine if you called your cable company and allowed them access to your search history exploring alternatives to cable? Would they change their pricing or work harder to solve problems? Maybe not because, well, they are cable companies. But many other vendors would use that data to improve service and tailor their offers to meet customer needs. This would be a win-win for consumers and vendors, and the contact center would be the lynchpin in securing and managing their interactions.
Michael Sisselman
It’s not difficult to envision a centralized repository wherein data subjects manage consent for third-parties to access their personal information. (This already exists in healthcare vis-à-vis access to medical records). But centralized, holistic management of the personal information itself (e.g., preferences, spending patterns, attitudes) is more futuristic. No doubt contact centers would be both contributors to and consumers of such information.
*****
Kieran Gilmurray [https://www.linkedin.com/in/kierangilmurray/]
Customer Experience (CX) through routing, self-service, and handling by a live agent alone is not a differentiated enough experience today. Prescriptive analytics, sentiment analytics; chat bots; smart & integrated IVRs etc will all help, but it’s technology-augmented agents who possess exceptional EI, empathy and skills that truly create differentiated people experiences. It is people, not just technology, that transform customer experiences and organizations.
Michael Sisselman
How information is communicated to callers is as important as the content itself. Human agents are especially adept at adapting to a caller’s communication style, emotional state, and general mood, and no doubt this can have a positive impact on overall CX. Luckily there are great tools (e.g., knowledge-bots, task-bots) to assist agents both during and after an interaction. There are also training-bots that can capture behaviors exemplified by the most successful agents and then train others.
*****
Peter Finney [https://www.linkedin.com/in/peterfinney1/]
Orchestration of a differentiated experience driven by Context here are all great valid points. The contextual journey should also be saved once an interaction is completed, and this historical data mined and analytical insights around these could also be considered and pulled into future inflight orchestration decisions. AI/ML for continuous customer experience improvement, active learning from contextual interactions?
Michael Sisselman
For CUSTOMERS we embrace the “journey” paradigm in an attempt to optimize individual interactions over time and to derive insights that might apply to others. No doubt this works for PROSPECTS as well if they are identifiable. The more interesting case is that of anonymous prospects. Even then behaviors can be discerned, modeled, and leveraged to yield optimal treatment for others.
*****
Julie Gershman [https://www.linkedin.com/in/juliekgershman/]
Excellent points on differentiated CX. It’s what consumers are coming to expect in their experiences.
Michael Sisselman
When consumers on browsing online and messaging on Social Media, they’ve come to expect a highly-calibrated personalized experience. Contact Centers have been moving in that direction with existing customers, but there is a long way to go in providing differentiated CX for prospects that are calling in for the first time. This can only get better.
*****
Robin Foster [https://www.linkedin.com/in/robinfosterroi/]
Thanks for sharing. Being context aware and acting on it is an advantage. Can you say more about: “fluency, dominance, and engagement.”
Michael Sisselman
Fluency, for example, is whether English is a mother tongue or a second language. If the latter, then it’s appropriate to slow the dialog and clearly annunciate each syllable.
Dominance is whether the caller is strong-willed and purposeful or open to guidance and suggestions. If the latter, then it’s appropriate to proactively steer the conversation.
And finally, engagement speaks to the implicit level-of-interest or lack thereof. If the latter, it’s appropriate to ask questions and try to incentivize the caller toward completion of a task.
In general, these psycholinguistic attributes can be measured during the course of a call and used to affect subsequent content and communications style.
*****
Dan Miller [https://www.linkedin.com/in/danmiller/]
Glad to see the topics surrounding #ConversationalIntelligence brought up in this framing. Thanks for your insights. In the short term, it is definitely possible to lean on #ThirdPartyData aggregators for information to inform those first encounters. In the long-run, concerns about accuracy, currency and privacy will shift focus to #FirstPartyData (permissioned info from individuals or their agents) to inform conversations between companies and their customers or prospects. Recognizing the challenge of rapidly recognizing and meeting the needs of first-time callers is a great use case for #conversationalAI
Michael Sisselman
Did you say ACCURACY? This important qualification is often overlooked and potentially wreaks havoc when a project moves from design to implementation. When a query to a Data Provider returns information, there is typically a confidence score that comes back as well. Ranging from 1 (= “highly confident”) to 0 (= “no confidence”), the Provider is explicitly scoring the validity of the data attribute itself. Needless to say, the higher the confidence score, the more precise we can be in orchestrating fit-for-purpose interactions. And the converse is also true: the lower the score, the more general we need to be in orchestration an interaction. By the way, I learned the hard way that whenever demonstrating a use case that is based on real-time access to third-party data, it’s best to test the inputs/outputs beforehand, just to make sure that there are no surprises due to accuracy or lack thereof.
*****
Bernard Gutnick [https://www.linkedin.com/in/bernardgutnick/]
I so miss the days working with you since I left Avaya. Your insight on CX topics was so spot on. The demonstrations of how my mobile number could open up my complete demographic information and purchasing patterns was incredible. It’s so interestingly creepy that when I say something to Alexa, it appears in the Facebook ads. How long will it be before retailers associated my caller ID with my Alexa questions so they can personalize the CX experience. It seems like Big Brother is listening and will lead to a Sony 65″ TV on Black Friday. Damn…… Facebook now will be showing me ads for a month. And the agent will answer with….”would you like a Sony TV that matches your entertainment center and buy a subscription of an Apple TV device to match you watch. You need a new band” It will never end.
Michael Sisselman
The breadth and depth of what the marketers know about me is astounding. There are literally thousands of attributes that speak to everything I’ve said and done, especially in the digital world. As a result, I expect that every contact center knows that I am a “Jet-Set-Urbanite”, spend a lot on money on fine Bordeaux reds, and if they are going to play music-on-hold, it should be a piece from the Classical Guitar repertory. As to whether this leaves me with a feeling that they are hyper-sensitive to my needs or disrespectful of my personal privacy, I guess that depends on the context.
*****
Jon Davis [https://www.linkedin.com/in/jon-davis-5589311/]
Interesting article and the power of AI is top of mind for many business executives these days. Agree with what you have here, mostly leveraging customer marketing data and context is king. We all know, a great CX drives loyalty and increase sales and these are great suggestions. Thanks for sharing your insight.
Michael Sisselman
Through great CX, value accrues to both the products and to the brand. On top of massive budget that generates inbound traffic in the first place, enterprises need spend only a bit more to create differentiated, optimized experiences.
*****
Will Nihan [https://www.linkedin.com/in/willnihan/]
Contact centers are predicated on human interaction but yet rarely have the context available to them to understand and empathize with why the customer is calling. Using AI and 3rd party data to enable them to get the context of why the customer might be calling will drive better experiences, shorter calls and happier customers and reps! Great insight- thanks.
Michael Sisselman
Let’s address agent reps, who may be happy, sad, or lackadaisical about the infusion of AI into talk-paths and workflows. Regardless of their disposition, agents play an extremely important role in providing the human-insights that are needed to train the underlying AI models. For that reason, and for the fact that they typically care for the most difficult interactions that are beyond the scope of bots, agents remain an inextricable part of best-in-class contact centers.
*****
Wayne Butterfield
Interesting perspective, a unique customer identifier is of course the holy grail for most organizations, but using 3rd party data as an accompaniment to your own is a novel way of gaining similar insight to a customer. What’s your view on how this works in Online Chat, Email, Social channels also? The Phone is not generally the channel of choice for people who are the ones expecting personalized service right?
Michael Sisselman
Analogous to voice calls, digital communications are typically comprised of “self-service” (interactions with a bot) and “assisted” (interactions with a live-agent) modalities. But there are salient differences, not the least of which is the asynchronous cadence of digital messages. Thanks for the prompting; I’ll be writing about differentiated CX in digital messaging in a forthcoming article!
*****
Pinku Reen [https://www.linkedin.com/in/pinku-reen-614aa11/]
Interesting Use Cases for leveraging “allowed” data that individuals may provide directly as first party or through third party….
Michael Sisselman
Throughout a conversation, algorithms are cranking continuously behind the scene to assess context and ultimately determine any /all the following: “Next-Best-Response”, “Next-Best-Action”, “Next-Best-Content”, and “Next-Best-Offer”. The point is that at a low-level, it’s the localized, first-party data that fuels most of the decisions that get made in near real-time.
*****
Peter Davis [https://www.linkedin.com/in/pdaconsulting/]
I have been looking at sentiment analysis for my law course as it is helpful during eDiscovery. I can assure you that Bell Canada does not use it when I call for technical support, or they would talk to me much differently.
Michael Sisselman
If they only knew that you wrote the book—literally—on technical support and other interesting topics in IT auditing.
*****
David Wisotsky [https://www.linkedin.com/in/davewisotsky/]
If this caller ID technology could be successfully developed, might it have uses in other contexts outside of an inbound contact center?
Michael Sisselman
Beyond inbound there are outbound calls, as would be the case with notifications of product recalls, civil safety alerts, and payment collections. Keep in mind that some of the underlying technologies and models could be adapted for digital experiences, spanning online browsing, messaging, and social media. More to say about differentiation of CX in digital channels in an upcoming article.
*****
Edward Williamson [https://www.linkedin.com/in/edward-williamson-72885527/]
I know about this one all too well, but still a great read. Interesting to see how some of these use cases will be adopted in the future.
Michael Sisselman
Edward, how many times did we demo at tradeshows together a use case wherein third-party data was accessed for the purpose of creating an appropriate travel offer for a first-time caller? What an excellent experience that seem to have catapulted your career as a CX analyst.
*****
Andrew Maher [https://www.linkedin.com/in/andrewfmaher/]
You know my love of wearing my own t-shirts and just a minute before reading this article I just put one of my “context” t-shirts in the wash.
Whether it is a call to a business, a walk-in to a shop or a chat at the curb with a neighbor, context drives our decisions, our handling; why don’t businesses make more use of this?
Thanks for sharing your article.
Michael Sisselman
Close your eyes and imagine one day that you’re in a videoconference with a live agent. You’re wearing that specific T-shirt (already washed and pressed), and they’re trying to make context of “context”. My guess is that you’ll get anything that you ask for…within reason. Please record that video (with permission from the agent), and upload to this blog for posterity.
Andrew Maher
What you describe is magic. Is that what companies need to aim to deliver?
Michael Sisselman
Desktop sharing moved the industry one step closer toward a productive, immersive conversation between consumer and agent. But if 2020 marks the era of Zoom, my bet is that 2021 will be the year that consumers get a chance to “see” the agents if they choose to. No doubt this will have significant impact on both customer service and sales interactions. The answer to your likely next question is…YES, all of the AI models will need to be recalibrated for what could be a paradigm shift in providing differentiated CX.
*****
Zachary Bamberger [https://www.linkedin.com/in/zeb3/]
Another interesting point is that perhaps people with similar language patterns (demonstrating a certain level of education, fluency, and care for particular entities) have similar end goals from calls. Is this a fair assumption? If so, even callers whose ID cannot be queried can instead be matched to a “nearest neighbor.” This could facilitate a hybrid between aggregated data and contextual data on customers.
I couldn’t agree more that the use of evolving NLP technologies in this application is critical.
Thank you for an awesome read!!!
Michael Sisselman
You’re correct that short of explicitly identifying the caller, the psycholinguist profile can be determined and then typecast to pre-existing personas. Based on troves of historical information, data scientists can model treatments that seem to work for various personas, and that becomes the basis for handling the anonymous caller—without the use of Personal Identifiable Information (PII).
*****
Constantin Garbe [https://www.linkedin.com/in/konstantingarbe/]
Interesting article, thanks for sharing.
Using CRM/profile information and call meta data to enrich/navigate automated conversations is nothing new to me. Using third-party data related to a caller ID is something I hear the first time indeed. Maybe it is owed to my operating location.
Is that a use-case that you see commonly? I noticed that you are saying „it could yield pertinent information“, so I guess we are talking about theoretical use-cases, although technologically feasible, right?
Michael Sisselman
With respect to third-party, consumer marketing information that is based on PII, coverage is much wider in certain countries (e.g., US, Canada, UK) than others (e.g., Germany, China). Largely this is a function of differences in data privacy regulations and societal norms. That said, there is widespread use by contact centers of external, real-time information that is unidentifiable. For example, severe weather, civil emergency alerts, and dramatic changes in the financial markets are all leading indicators, a priori, of the reason for a call and can be used by the enterprise to affect treatments.
*****
Michael Pawlak [https://www.linkedin.com/in/michaelpawlak44/]
Nice to see a focus placed on unknown or first-time callers. A company “should” know how to respond when they know who the caller is. The challenge is when you don’t know, and your article provides a roadmap to improve this experience. Great job.
Michael Sisselman
Needless to say, it’s a continuum ranging from 100% (“We know this caller very, very well”) to 0% (“We don’t recognize this caller at all”). But in all cases, it makes sense to mine context and endeavor to differentiate CX appropriately.
*****