• July 27, 2020
  • By R "Ray" Wang, founder, chairman, and principal analyst, Constellation Research

Automation Is the Future of CX

Article Featured Image

The pressure to reduce margins, technical debt, and investments in core systems creates a tremendous incentive for increased automation. The benefits are numerous and obvious: less staffing, reduced errors, smarter decisions, and security at scale. The quest for an autonomous enterprise starts with the need to consider what decisions require intelligent automation versus human judgment.

Vendors from multiple fronts intend to deliver on this promise. Legacy CRM and customer experience providers, cloud vendors, business process management suppliers, robotic process automation providers, process-mining vendors, and IT services firms with software solutions are attempting to compete with pure-play vendors for both mindshare and market dominance in the intelligent automation market, which Constellation Research expects to hit $10.4 billion by 2030.

Almost every marketing leader has sought to intelligently automate processes as part of critical operational efficiency initiatives. From campaign to lead, order to cash, incident to resolution, and concept to market, no department is immune and no business process is exempt. While these efforts to automate often start with the desire to cut costs, they can evolve into something more. The advent of artificial intelligence (AI) components such as natural language processing, machine learning, and neural networks present opportunities to deploy fully autonomous capabilities that have strategic and long-ranging impacts. Seven forces drive the quest for autonomous capabilities in the enterprise:


Widespread business disruptions and the growth of disruptive business models have shifted boardroom and organizational priorities. Organizations expect to spend more on agility and business continuity, and they no longer seek to invest more in legacy technologies and systems that do not support those two areas. Key investment themes include self-driving, self-learning, and self-healing systems. While the long-term goal is sentience, the short-term capabilities enable redundancy at scale as well as rapid development, testing, deployment, upgrades, and refreshes.


The ongoing battle to address short-term, quarter-to-quarter profitability and the scarcity of top talent gives companies an incentive to invest in automation to augment the labor force. The good news: Enterprises have the technology to automate business processes at an unimaginable scale. Thus, every organizational leader must determine when to trust the judgment of a machine, when to augment a machine with a human, when to augment a human with a machine, and when to trust human ingenuity. In this autonomous future, machines will deliver services that are continuous, auto-compliant, self-driving, self-healing, self-learning, and self-aware. Access to larger datasets and more engagements to refine algorithms will be needed to ensure precision decisions and ever-higher confidence levels.


Many industrialized countries face declining populations. Japan, for instance, faces a projected population decline of 16 percent, dropping from 127 million in 2014 to 107 million by 2040. Europe is projected to have 0.3 percent to 0.5 percent negative growth by 2040. Furthermore, aging populations, declining birth rates, and minimal immigration create systemic declines that hamper productivity gains, reduce the labor force, and erode any economies of scale. Meanwhile, rising labor costs and regulations drive up labor inflation for both services and manufacturing. Leaders seek ways to drive down labor costs from recruiting, re-skilling and retraining—by replacing with automation.


Leaders seek to mitigate compliance risk and reduce errors through the automation of manual tasks. With more than 70 percent of employee time focused on manual and repetitive tasks, many seek relief from the mundane. Manual entry and labor for transactional systems lead to higher risk of errors. Today’s volume of transactions and the downstream implications of improperly entered data, bad data, and late data create exponential issues in human-led errors that must be addressed. Consequently, every enterprise must automate at an unprecedented scale. One compliance fine or privacy breach caused by human error could lead to hundreds of millions to billions of dollars in losses.


Successful AI projects seek a spectrum of outcomes. Automation and training models will improve with more data and more interactions. The disruptive nature of AI comes from the speed, precision, and capacity for augmenting human workers and delivering on the goal of a more automated enterprise. Seven AI outcomes show the progression from perception to sentience on the spectrum:

• Perception describes what’s happening now. The first set of outcomes describes surroundings as manually programmed. Perception provides a first-level report of activity.

• Notifications tell you what you asked to know.
Notifications through alerts, workflows, reminders, and other signals help deliver additional information through manual input and learning.

• Suggestions recommend action. Suggestions build on past behaviors and modify over time based on weighted attributes, decision management, and machine learning.

• Automation repeats what you always want. Automation enables leverage as machine learning matures over time and tuning.

• Prediction informs you about what to expect. Prediction starts to build on deep learning and neural networks to anticipate and test for behaviors.

• Prevention helps you avoid bad outcomes. Prevention applies cognitive reckoning to identify potential threats and to augment human judgment.

• Situational awareness tells you what you need to know right now. Situational awareness comes close to mimicking human capabilities in decision making.


In this world of relativism and enhanced technologies, humans have more trouble discerning authenticity. The blurred line between reality and fiction creates conditions that can sway public opinion, incite violence or riots, and bilk others of value. The need for authenticity still remains, and those individuals and enterprises that can deliver authenticity will win trust and significant business. AI and automation must quickly identify, notify, respond to, and eradicate deepfakes and prevent them from intruding on existing systems. With an increasing number of systems networked to outside systems, customers can expect the greater attack surface to spawn high volumes of denial-of-service attacks, phishing scams, fake invoices, and usage of stolen identities. Autonomous systems will effectively combat these at scale.


Despite massive efforts to grow and train talent, foster innovation, and create institutional knowledge, regressive factors such as high turnover, agile project methodologies, mergers and acquisitions, and short-term thinking challenge the ability to retain and share institutional knowledge. Without easy approaches, organizations quickly forget, facing a degradation of knowledge with each departure and each organizational restructuring. Autonomous enterprises capture the informal and people-centric institutional knowledge from processes, leading to best practices and nuance in decision making. This enables consistent planning, shared institutional knowledge, and a permanent and living memory.

The future of CX points to a more automated enterprise. The more we automate, the more we can build models to improve next best action. The ultimate goal is to deliver precision decisions. Keep in mind that AI enablement requires a strong data strategy, deep data governance, and mature business process optimization.


Seven factors play a significant role in identifying which AI-driven smart services deliver the greatest opportunities:

1. Repetitiveness. The greater the frequency a process is repeated, the more likely the process should be AI-powered. One-offs and custom processes with minimal repetition are lower-priority candidates for AI.

2. Volume. When the volume of transactions and interactions exceed human capacity, the service should be AI-powered. Volumes within human capacity can remain human-powered.

3. Time to complete. High time-to-market requirements favor AI-powered approaches. Lower time-to-completion requirements can remain human-powered.

4. Nodes of interaction. Simple interaction nodes will lean toward the human-powered option. AI serves best in complex and high-volume nodes of interaction.

5. Complexity. Good candidates for AI-powered uses include complexity beyond human comprehension or, at the other end of the spectrum, simple tasks that can be optimized by AI.

6. Creativity. Today, the cognitive processes required for creativity mostly reside with humans; higher creative powers are less likely to be AI-powered. But with advancements in cognitive learning, one can expect creativity to improve with AI-powered approaches over the next decade.

7. Physical presence. Processes that require a heavy physical presence will most likely require human-powered capabilities. However, processes that put lives in jeopardy serve as great candidates for automated, AI-powered options. In general, low physical presence requirements play well to AI-powered approaches. 

R “Ray” Wang is founder, chairman, and principal analyst of Constellation Research. He is the author of the business strategy and technology blog A Software Insider’s Point of View. His latest best-selling book is Disrupting Digital Business, published by Harvard Business Review Press.

CRM Covers
for qualified subscribers
Subscribe Now Current Issue Past Issues

Related Articles

The CRM and E-Commerce Convergence Begins

Companies benefit when customer data can inform e-commerce personalization.