No Jitter is part of the Informa Tech Division of Informa PLC

This site is operated by a business or businesses owned by Informa PLC and all copyright resides with them. Informa PLC's registered office is 5 Howick Place, London SW1P 1WG. Registered in England and Wales. Number 8860726.

Applying Automation to Improve Customer Experience

SiberianArt Alamy Stock Vector.jpg

Image: SiberianArt - Alamy Stock Vector
As we advise enterprise customer experience (CX) leaders, one fairly consistent area of deficiency is using automation to improve agent performance and/or customer satisfaction. Sure, most companies use some automation, but the customer experience is one area where the job is truly never done.
 
I’m not advocating a boil-the-ocean approach to artificial intelligence and automation — quite the opposite, actually. Automation strategies for CX should be a continuum, where projects build upon on another. Tackle one problem or opportunity, measure success, refine for further success, and move on to the next one.
 
Fortunately, there is no shortage of innovation among both startup and legacy providers to help automate manual processes in CX — an endeavor ultimately resulting in lower operational costs, better customer satisfaction, and even improving revenue. I often finish briefings with vendors and think, “Wow! That’s a really cool product!” So here and in the future, I’ll periodically provide No Jitter readers with some highlights for you to consider. Today's piece looks at ways automation can improve operations in customer experience.
 
(To be clear, no vendors requested, paid, or provided any incentive whatsoever to be in this post.)
 
What Should You Consider Automating?
From a high level, automation projects should always address a problem or opportunity. These initiatives—particularly when they can change and benefit through the real-time learning of artificial intelligence and/or machine learning—can be truly transformative by changing processes, expectations, and results. Here are a few examples, along with technology providers that I think offer some solid solutions:
 
1. Quality Management – for the Customer and Agent
We’ve all called into banks, airlines, or retailers and heard: “This call will be monitored for quality assurance.” Typically, that process has been manual. The supervisors randomly (or intentionally if they are checking in on under-performing agents) listen to the calls and coach based on the outcome.
 
By using AI to automate this process, supervisors get a more accurate view of what’s happening globally across all agents—and AI can flag conversations that score low in sentiment analysis, for example. Or, they can see which agents are performing flawlessly and listen-in to what they’re doing better than others to replicate their behavior. Reporting can show data, trends, and analytics on the all calls over time. Observe.ai is one company offering such a service. The sweet spot for them is companies with 100 to 5,000 agents.
 
Intradiem also has an interesting approach for what I still partially consider quality management—except it’s on the agent side. Agent AHT Assistance reviews rules and conditions for average handle time (AHT) of calls and takes targeted action (i.e., supervisors proactively intervene to offer assistance). What gave me a “wow” moment is Intradiem’s upcoming release, which will evaluate agents’ current state and dynamically adjust their Key Performance Indicators (KPIs) based on pre-established factors and outliers—continuously challenging agents and improving their performance. For example, AHT for an airline may increase during a strike or storm system, so what’s considered an acceptable AHT during that time may change, so agents aren’t penalized for missing their KPIs.
 
Another possible approach relies on automation. Balto.ai uses AI to track all conversations, and its technology can identify whether agents are reading disclosures or using words (like “guarantee”) that can get the company in trouble. Similarly, Dialpad’s Ai QA Scorecards transcribe calls in real-time and provide a checklist that aligns with required tasks. A future Dialpad release will flag phrases on the QA scorecards to speed the call review process for supervisors.
 
Along those same lines, NICE Actimize recently introduced Compliancecentral, a cloud-based platform that monitors communications to uncover potential risk issues with financial transactions. What’s cool, though, is that it uses AI and analytics to corelate employee behavioral data with their communications patterns and activity. This combined data makes it easier (and more accurate) to monitor and investigate risk.
 
2. Real-time Schedule Optimization
Scheduling the right number of agents each day is a challenge in and of itself. Workforce management applications have been working on this issue for years. Amazon Connect took that to a new level by adding sophisticated algorithms that analyze the accuracy of capacity planning, making adjustments regularly to improve the accuracy (38.8% of the respondents in our Customer Engagement Transformation 2022-23 research study of 724 companies say this is a vital capability for them.)
 
Ensuring agents’ days are as productive as possible, based on interaction volume, is another challenge. For example, agents often get stuck on calls prior to their scheduled break times, resulting in a shortened break or a manual exception task whereby they inform their supervisor they started their breaks late and get permission to extend the scheduled time. Intradiem offers a proactive and automated capability that evaluates call volume and identifies agents with breaks coming up within the next five to 10 minutes. The system can ask the agents if they want to take their breaks early; when they respond affirmatively, the system fires off a series of rules to automate break times.
 
Akixi also helps companies with scheduling through out-of-the box reports that document and help improve productivity and recapture lost callers. The volume-by-time reports help ensure contact centers map the number of agents with volume. They also track the activity of employees regardless of their workplace. Its analytics identify trends to help companies (particularly SMBs) improve capacity planning.
 
3. Automated Coaching
Typical contact centers set aside 5% to 10% (depending on role and industry) of agents’ time for coaching. Often, agents end up skipping pre-scheduled sessions in response to spikes in call volume. Intradiem also offers an interesting solution here by delivering automated coaching during pockets of organic idle time. This micro-training—say, six 10-minute sessions per week—allows companies to reduce their overall staffing because the training happens during times that would have been idle otherwise. The agents become more productive—and they still get their training. The system can find micro-training slots vs. scheduling them to be offline for an hour each week for coaching that often doesn’t happen anyway when the static time scheduled becomes too busy with customer interactions.
 
4. Real-Time Guidance
Also known as Agent Assist, we see this as a big growth area for CX strategies, with 41.9% of companies planning to use it by the end of 2022, according to our Customer Engagement Transformation 2022-23 research study.
 
One common goal of real-time guidance is helping the agent determine the next-best action. Screen-pops typically provide this type of information (in fact, 35.1% of all CX interactions now use screen-pops, according to our study). UJET , which has an impressive ecosystem of partners in QM, business intelligence, virtual agents, UCaaS, and CRM, is focusing on making guidance increasingly more real-time. For example, it provides real-time sentiment analysis and screen pops with next-best actions that include suggestions for sentence completion on webchat, as well as contextual articles. UJET's partnership with Google Cloud Contact Center AI will continue to expand its AI offerings, as well, so it’s a vendor worth watching.
 
Five9 has added several features around its Agent Assist services. One I particularly like is the ability of AI to complete checkbox items for agents. For example, many companies have added sales objectives to customer service representatives KPIs. But some may shy away from asking the upsell question on every call, or they may be running high on their handle time KPI. Some may be tempted to check the upsell question box as “complete,” even though it’s not. By mandating AI (using natural language processing) marks the boxes complete, agents must complete all tasks or risk being flagged to supervisors.
 
Both Balto.ai and Dialpad target the key issue of agent knowledge retention. When agents start their jobs, they go through significant training—and then, let’s be honest, forget half of what they learned. Both companies guide agents through their conversations with access to up-to-date information stored in company knowledge bases, web pages, ticketing systems, etc. Dialpad’s Ai Agent Assist searches through such unstructured data to find answers needed to address customer inquiries, reducing average handle time by 66%. Using AI to predict what the agent needs and delivering in near-real-time gets them up and running faster, because the amount of training they need is drastically reduced.
 
Zoom IQ for Sales focuses on sales people vs. customer service representatives. It uses AI to interpret customer interactions and provide actionable recommendations and insights which are integrated into CRM to provide context for the sales funnel. Sales managers can quickly see deals that need attention and identify pipeline trends. Sales teams can improve their presentation skills by tracking their talk-listen ratios, talking speeds, and sentiment analysis. ZoomIQ is a hybrid sales/contact center tool, particularly, as our research shows 35.4% of contact center leaders hiring people with sales experience to act upon newly created sales objectives.
 
5. CSAT Insights
Wouldn’t CX leaders’ lives be grand if every customer responded to post-interaction surveys? In reality, only about 5% -15% do—and those are typically those who are extremely happy or extremely disappointed. Dialpad uses artificial intelligence to track typical customer satisfaction (CSAT) scores based on pre-established profiles (i.e., people calling from this region at this time of day, with this sentiment typically rate CSAT X). So regardless of whether a customer provides feedback on an interaction, Dialpad can predict what their scores would be with an 85% to 90% accuracy (it does quality assurance testing to validate the figures). Given companies often make decisions based on their customer feedback , the ability to get real and accurate predictive scores can make a huge difference in the accuracy of changes they make to improve customer satisfaction.
 
Cast a Wide Net
Yes, I could go on, and there are certainly many other providers innovating. The key is to map your business problems or opportunities to these existing options. Take the time to evaluate providers, and initially cast a wide net before narrowing your list down to your top three provider options.