Conversations about workforce optimization (WFO) have heated up in the past couple of months. Following several passionate discussions about its future – as the transition to the cloud accelerates – I was inspired to write about the category and its evolution.
My first surprise was the broad range of definitions for WFO. I was struck by the number of people equating the three-letter acronym to recording and quality management (QM). Industry analysts concur on a much broader definition including workforce management (WFM), performance management (PM), and more advanced applications like speech, interaction analytics, and gamification. It suggests businesses that buy WFO suites often only use a fraction of its capabilities.
History of WFO
WFO is actually a vendor-created concept. In the early 2000s, two applications, QM for recording and evaluating conversations and WFM for predicting call volumes and creating the optimal agent schedules, were entering the mainstream market. WFO optimization promised an integrated suite of applications to manage the entire agent lifecycle, from onboarding to skills development, and to effectively deploy agents with functions like scheduling.
Witness Systems, then a quality management vendor and now part of Verint, can be credited for coining the term. It made the first consolidation move by buying workforce management vendor Blue Pumpkin and combining the two applications. One year later, NICE purchased IEX, another WFM player, as well as PM provider Performix.
PM was a new application for assembling a global view of agents’ performance metrics from contact center reports, QM scores, and other applications such as CRM for call outcomes or customer satisfaction surveys. PM could uncover the attributes of best-performing customer service representatives and include a workflow to manage feedback and development.
Rise of WFO suites
The following years saw the consolidation of the market by NICE and Verint (each owning a third), the emergence of midsize providers Aspect and Calibrio competing for third, and a long tail of point solution providers.
With IP telephony, call recording became a function you could implement in software, and contact center vendors started to incorporate it into their stack. Gartner predicted that WFO would be subsumed into contact center infrastructure software. Today, almost all contact center providers have integrated recording into their solution. Most of the larger players made QM tuck-in acquisitions, and several also bought their way into WFM. However, Gartner's prediction didn’t fully materialize as witnessed by the two market leaders holding their market share.
The market is dominated by WFO suites with comprehensive capabilities. In Ovum's recent market review, Ken Landoline noted the past decade was marked by vendors continuing to add features to their suite. DMG’s 400-page-long WFO product and market report echoes feature-rich offerings. After years of incremental improvements, I am finding a growing number of practitioners wanting to try new approaches – a second surprise to me.
Getting the most of AI and analytics
The traditional QM model is labor-intensive and built on human listening and grading a random selection of calls. Cloud-based, AI-enabled automatic speech recognition (ASR) can transcribe calls with compelling accuracy levels. The leading technologies enjoy word error rates (WER) between 10 and 15%, hinging on the 4-6% performance of humans. AI can also uncover valuable signals such as silence and emotion. The latest generation of speech and interaction analytics is changing the model, automatically generating key performance indicators (KPIs) and surfacing the specific interaction segments that should be analyzed by a human. Forward-looking companies have started to harness the power of machine learning to uncover patterns of both agent performance and customer feedback.
Not your grandfather’s workforce
The modern workforce has higher expectations, and enterprises are recognizing the importance of engaging their employees better. It has them revisiting performance metrics, increasing their focus on customer satisfaction and outcomes. Customer service organizations are turning to customer and employee experience management applications to address these trends. In response, WFO vendors have started to add these capabilities to their suites. NICE purchased Satmetrix in 2017, and VERINT acquired Foresee last year.
The traditional workforce scheduling model is also put under pressure with the multiplicity of channels, the increased diversity of interactions, and their duration variability. Genesys has introduced AI-based scheduling. Dave Michels argues that the number of variables to optimize has reached a level that makes scheduling too complex. He is urging the industry to consider the benefits of Uber-like schedule-less operations. Indeed, on-demand work arrangements, the combination of self and assisted service, and more asynchronous–messaging-like–interactions are creating new ways to handle staffing.
Easier integrations
The original promise of PM was to assemble a “360-degree” view of agents’ performance across applications. As the industry consolidated, most vendors turned their focus inwards on unifying their suite. The explosion of interactions across channels often managed using different software, and the overall increase of applications used for customer service triggered a renewed interest in federating agents’ KPIs. New entrants such as Playvox or Calabrio’s acquisition of Symmetrics illustrate this trend. Overall, the industry needs to become more open. Businesses want to try these new models through bite-size experiments and simpler addition of point applications into their existing stacks.
It’s time for the industry to look beyond adding incremental features and perfecting existing WFO suites to enable new models for maximizing agents’ productivity and performance. As for any change involving the workforce, organizations want to proceed incrementally and be able to inject targeted innovations in their solution stacks. It requires offers to become more modular and provide more open APIs.
This article was originally published on No Jitter