Published: October 30, 2018 | Comments
Think alternative truths don't affect you? If you're like many contact centers, you may be operating your centers on misleading metrics.
A recent insurance client of mine was confounded with overly negative morale that manifested in a high attrition rate. The resulting lack of agent expertise was hurting customer satisfaction as customer satisfaction is consistently affected by the tenure of the agent. The turnover was also costing them a lot of money to replace the staff. I've seen this elsewhere, so I analyzed how they measure and optimize C/Sat. Their process was to survey customers quarterly and respond to any negative comments or Emails with an all-hands on deck approach, immediately bringing each issue to the team's attention to remedy the tactical issue and to prevent reoccurrence. It was a great idea to repair occasional problems, but a consistent barrage of bad news was taking a toll. Too many of their talented agents had enough and left.
I performed statistical sentiment analysis on all the survey responses. In the over 10,000 replies analyzed, their customer sentiment was distinctly appreciative and positive for their customer contact services, but since the communications were based the misleading sample from a small subset of urgencies, the contact center staff felt like failures and were highly demotivated. After sharing my analysis results, the leadership developed a more balanced communication procedure, had a floor to ceiling mural painted of a word cloud I developed of the survey results, and attrition reduced by 60% in the next year.
Success for your call center relies on effective operational analysis, but most centers can likely get feeds from 10+ data sources. Since those sources often calculate similar metrics very differently, the output can be very misleading and confusing.
How can you tell fact from fiction? Here are some examples where misleading statistics were misguiding operational decisions:
- Metrics not reflective of your core mission - most competitive companies list providing excellent customer service as a primary contact goal, but a majority operate by closely monitoring productivity-based metrics. It's helpful to your bottom line to watch the average talk time, but you need to use CSAT and FCR as your barometers for success.
- Insufficient data - Many companies use survey responses as a basis for changing their service level targets. As most get 3-6% response rates and typically from customers who really love you, or really don't, the sample size basing the analysis is not reflective of the typical customer sentiment. You'll change important elements of your operation to please the extreme few, and not the majority of your customers.
- Incomplete data - Text analytics analysis is often based on CRM entries, but typically over 50% of entries are truncated, use inconsistent abbreviations or conventions, or are merely missed details in transcription. That results in a lot of effort to normalize the inputs or lost data to define the frequency of trends.
What news do you need to hear to keep your customers happy and customer interactions successful? I suggest you can get that from applying these five best practices -
1. Define KPIs matched to desired outcomes. If your business is positioning itself in the market as a low-cost operator, use performance-based metrics as your Key Performance Indicators. And keep your KPI as key. Other metrics may be valuable but limit your KPI to a maximum of 5. This keeps the operational focus on those metrics that matter to success in your business.
2. Share good news and bad news. See the above cautionary story.
3. Looking at a single data source has most analysts missing the real headline of Customer Experience insights. A customer journey, even if in a single channel, is a combination of several events, from caller inter determination to queueing, to resolution, and finally after-action perception. Looking at any of these events myopically limits the scope and thereby the value of any insights. This increases as more channels are adopted. The idea is to merge several data sources and visualize normalized data.
4. Invest specifically with a thorough, conservative business case. Define the desired outcomes in detail and keep the expectations tempered. As an example - don't think that investing in Artificial Intelligence (AI) will solve all your text analytics needs. AI needs considerable time to become fruitful.
5. Analyze with purpose. Don't wait for insights to jump out to you. Define an area of focus that will move your analysis from data-driven to insights-driven. This will provide affirmation if positive trends are found.
How many of these suggestions are you using? What have your results been? Please share your comments below to help us all learn. Be sure to join me for my presentation at ICMI Contact Center Demo in Las Vegas, where I'll unpack these concepts and best practices in greater detail. And be sure to follow me on Twitter as I'll be live Tweeting the event at @RobertLambcxpro.