The right advice at just the right moment can make all the difference to employee satisfaction and performance. New AI-driven approaches are making truly personalized, real-time coaching a reality.

Great employees are not born. They develop their skills over time, through a combination of on-the-job experience, formal training, coaching, and mentoring—inputs that employers have a high degree of control over. Most companies, however, struggle to give their people the breadth and depth of support they need. Coaching and training interventions are usually infrequent, limited in scope, and delivered using a one-size-fits-all approach.

In some organizations, entrenched cultures and habits have proved difficult to shake off. In others, the use of generic and repetitive feedback has left staff disillusioned about learning and development. In the worst cases, poorly considered coaching and feedback can actually harm employee experience, morale, and engagement, driving up attrition. Organizations with a high proportion of transient or distributed workers are especially exposed, as they have little time to develop and instill high-performance behaviors.

Organizations with a strong performance management culture know that there is a wide gap between their highest-performing employees and their weakest. Other companies suspect the same, even if they lack the data to prove it. Leading companies also understand that on-the-job training, coaching, and support must be tailored to the strengths and aptitudes of the individual (Exhibit 1).

In coaching and performance management, one size does not fit all, and support should be tailored to individual performance.

Harnessing the power of behavioral insights

Today, behavioral psychology is starting to transform the way companies engage with both employees and customers. In recent years, a growing number of organizations have discovered for themselves that seemingly simple nudges can be a remarkably effective way to promote positive behaviors. An insurance company encouraged customers to exercise and lead healthy lifestyles, thereby reducing claims. A steel manufacturer saw a 35 percent increase in employees following safety procedures in one of its plants after placing posters of watching eyes to remind employees where safety procedures are critical. A major financial institution started to offer small prizes to loan officers to encourage the early achievement of monthly targets. That led to an 18 percent increase in the sourcing of new loans in the first two weeks of the month. And these are just a subset of nudging techniques, out of a wide range of proven options.

As customer interactions are conducted increasingly through digital channels, leading organizations have been able to greatly refine their use of nudge techniques during the customer journey. Advances in AI and machine learning (ML) allow companies to identify critical decision points for customers and offer timely and targeted advice that can dramatically improve sales, customer experience, and retention. These sophisticated and personalized approaches are now inspiring innovation in AI-driven employee engagement, enabled by the ubiquity of digital hand-held devices.

Enter the AI coach

Companies are combining behavioral-science insights with the latest AI technologies to create a new kind of tailored coaching experience for their employees. Like the best customer-facing AI-driven tools, these new “smart AI coaching” systems combine multiple sources of data to identify specific strengths and weaknesses in an employee’s performance. They use these data to select timely, context-specific nudges, delivered via the employee’s computer or hand-held device, with intelligent, deliberate repetition. These nudges can be delivered in various ways: as a regular, structured program; as an on-demand service; or in real time while employees execute their tasks. They can cover a broad spectrum of topics too, such as employee health and wellness, customer experience, or operational efficiency. Nudges can also be delivered to supervisors, to encourage them to coach employees on the right skills at the right time (Exhibit 2).

AI-driven nudges can encourage higher job performance in many ways.

AI-driven coaching uses proven principles to change deeply ingrained habits. Adult-learning specialist David Kolb demonstrated that adults cannot learn merely by listening to instructions, for example. They must also absorb the new information, use it experimentally, and integrate it with their existing knowledge. That means new skills are best learned in small chunks, with learners given the opportunity to put their fresh knowledge into practice straight away. Moreover, behavior change requires repetition of actions and consistency over an extended period before it is internalized and demonstrated in a predictable way. A study on habit formation by researchers at University College London found that it took participants anywhere from 18 to 254 days of repetitive action to build a single habit.1

Because people have short attention spans and narrow bandwidth for digesting new information, the AI algorithms help to identify the most critical metrics to focus upon for each individual employee. These systems are also designed to improve their own performance over time. An individual’s response to different nudges can be analyzed to develop a profile of their preferred learning style and the types of intervention that work best for them.

AI-based systems need to be designed to complement traditional in-person performance management and coaching, not to replace it. If they are, AI systems can provide detailed insights for managers, telling them when specific employees deserve praise and public celebration, and where others need additional support to help them address specific areas of their performance.

One utility company used the AI-coaching approach to boost the productivity and quality of the work done by its field service technicians. The company built a recommendation engine to identify personalized coaching opportunities based on an individual’s past performance in key metrics such as time per job or rework rate. It then developed a library of targeted nudges and best practices delivered directly to technicians and their managers.

Those nudges included requests to complete certain steps or actions, as well as summaries of best practices for specific activities, with links to informal training content such as video clips. Initial nudges were followed up with reminders, and technicians who improved their performance in response to these nudges would receive “affirmation” nudges recognizing their progress. The content and overall tone of the messages were adjusted depending on the technician’s response to the combination of personalized nudges sent to them.

The system was piloted with test groups in the field service workforce, with its performance measured by comparison with similar groups that relied on the company’s conventional coaching and performance management techniques. The results were compelling. The productivity of the test groups increased by 8 to 10 percent, while the need for rework dropped by 20 to 30 percent compared with the control group. Together, those productivity and quality improvements helped to drive a 5 to 10 percent reduction in cost. Better still, the system was well-liked by both field technicians and their managers (Exhibit 3).

An AI-coaching pilot delivered compelling results for a utility’s field services.

Another company uses smart AI-coaching techniques to improve performance in its customer service call center. At the start of a conversation with a customer, the AI system determines the most likely reason for the call, based on the customer’s profile and past activity. It then reminds the agent about best practices associated with that kind of interaction. As the call progresses, the system offers automatic real-time nudges, suggesting additional diagnostic steps or cross-selling opportunities, for example. Agents receive weekly reports celebrating good performance in areas addressed by the nudges, such as reducing their average handling time or always rerouting calls to the most appropriate team. The report also suggests actions and resources to address areas that could be improved.

The system supports team leaders, too. At the start of each day, it offers resources for them to share with the teams in morning meetings, based on predictions of the most likely call types expected that day. During the shift, it tracks the progress of ongoing calls and flags those that might require supervisor attention. Weekly summaries offer insights into team- and agent-level performance, with suggestions for effective coaching interventions.

AI-based coaching systems are now delivering significant impact in a range of settings. At the call center of one business service provider, average handling time dropped by 11 percent within three weeks of the system’s introduction. Another company achieved a rapid 10 percent drop in the share of calls that needed to be transferred to a second agent to address the customer’s issue. At one healthcare company, the AI system identified coaching opportunities in more than 60 percent of the calls it monitored (Exhibit 4).

AI-coaching examples in customer service contexts have shown significant improvements in key operational metrics.

Making smart AI coaching work

AI-based coaching systems generate impressive results in a range of different work situations, but making them work effectively requires careful targeting, testing, and development. Beyond the obvious need for employees to have routine access to a digital device, the approach works best for environments that share a few common characteristics. First, the company must have a good understanding of critical KPIs at the individual employee level, and it must be able to define its aspirations for employee performance in terms of those KPIs. Second, it must have clear opportunities for improvement. Typically, that means a group of employees who commonly demonstrate underperformance, or where the variation in performance between individuals is wide.

Once the company has identified a suitable target group, the approach is best deployed using agile principles. Usually, that involves the creation of a minimum viable product (MVP) that can be tested with a select subgroup of employees. The MVP can be used to evaluate people’s responses to different types of nudges, and to refine the manner and timing of insight delivery to ensure a seamless experience for users. Depending on the company’s maturity on the AI/ML/advanced analytics journey, it may be best to start with business intelligence tools first before graduating to more complex algorithms.

As is always the case with new technologies, much of the hard work takes place outside the digital tool. Successful organizations establish a core group of digital and learning specialists to lead the pilot and drive the development of the system, creating and improving AI models, developing new nudges for staff and insights for managers, and extending the system over time to cover additional KPIs and use cases. This team can also work with the wider business to ensure that appropriate training and support is given to all users of the system, and to promote effective change management across the organization.

Source: McKinsey & Company

Previous How Machine Learning Can Support Historians Decipher Ancient Inscriptions
Next Technology University Stops Information-Stealing Cyber-Attack With Darktrace AI