Is chat an effective and efficient customer interaction channel for customer service? Absolutely yes. However while chat adoption for customer service is healthy the leverage of chat as a strong service channel is not so. There are many myths on chat. Consequently its effectiveness as an alternative channel to voice gets constantly questioned by both adopters of chat as well as by those who have not yet adopted chat.
Some of the common myths are that cost per contact in chat is more expensive than phone, customer experience and first contact resolution in chat is lower than voice and that it is not possible to reduce phone volumes through chat. All these myths have some common underlying realities. One, the common approach to implementing service chat has either been a technology or as a website experiment, through small pilots to evaluate the ROI and other benefits compared to voice. Second is the low awareness levels on how to leverage and grow chat as a service channel.
Consider this. 90% of customer service chat experiments/ pilots fail to produce the desired results within 4 to 6 months despite choosing a best in class chat technology. Chat cannot be tested as a technology or a website experiment. Running a customer service chat is not about putting a button chat on the website and seeing if customers click to chat or creating simple rules for targeting website visitors. Creating effective chat interactions requires deeper understanding of consumer contact behavior across channels, segmenting, targeting and driving smart interactions that can avoid potential calls from the website.
Secondly, most chat pilots start with less than 10-15 agents, with the expectation that it can be grown once the ROI is proved. This is a recipe for disaster. Such a small pilot cannot decisively prove anything vs voice and it is guaranteed not to get any executive attention especially in large contact centers. Chat programs need to start with an optimum number of agents that is modeled based on contact volumes using analytics. Growing the chat volumes requires an integrated execution of the analytical models with sophisticated operations. In addition chat also needs to be run as a unique channel without trying to lift and shift practices from voice operations. Many chat pilots start with tenured voice agents under the assumption that tenured voice agents have the product/ process knowledge and would be the best fit for a pilot. Incorrect. Chat operations require different skill sets, personality types that impact performance and productivity.
In effect running a service chat program effectively to produce significant results needs to be done with the vision to make the service transformation happen, backed by executive commitment and with a clear project plan that is not restricted to pilots. It is a science and not an experiment in art. And only few companies that have truly succeeded in transforming their service operations have cracked the code.
Are opinions on social media a lead indicator for CSAT? Absolutely yes. With millions of opinions expressed on social media sites such as Twitter/ Facebook everyday it has become a reality today. For example as a company we analyse tweets about several companies through 247tweetview (http://www.247tweetview.com) and when we compare the sentiments expressed by consumers with the end customer CSAT for our clients there is a clear pattern. Companies do not need to wait for a survey to be done/ analyse it and then figure out their CSAT scores are good or bad. The moment of truth is happening every minute/ every day.
Read the full post »
Customer Satisfaction Surveys have become the default to measure how content customers are. By nature, CSATs tend to be reactive; they measure the satisfaction of customers long after the customers and the business have completed their interactions.
Read the full post »
According to industry sources, an aggregate view of customer contacts and channel mix in the United States of America across major industry verticals (The US Contact Center Operational Review, 2nd Edition, 2008) shows that 89% of all contacts are served by the phone channel either through a live agent (65%) or through automated solutions (14%) such as Interactive Voice Response (IVR) or speech recognition. This data only considers “customer-company” interactions and excludes web based self-service and peer-to-peer interactions through social media/community forums/blogs whose growing impact in providing resolutions to customer service issues has not been satisfactorily estimated till date.
Read the full post »
The power of N=1, R=G has become very relevant in the Web 2.0 world, where a lot of end customer data is accessible and can be used to customize the experience for every individual visiting an online site, as opposed to the traditional retail model. The R=G part is an ecosystem.
In brick-and-mortar stores, traditionally, a lot of analysis was done to segment and drive the right traffic to the store, followed by advanced merchandising and pricing methods to drive up sales. However, in the new 1-to-1 online world, we are more advanced in how every movement of the visitor to a store can tracked.
Read the full post »
Look at the numbers from ATT’s point of view. They have some 60 million customers, and the average customer calls with a question or complaint six to eight times per year. The average call costs about $1 per minute to handle, and lasts four to five minutes. That means that ATT, according to industry averages, must spend around $2 billion per year on call centers.
Read the full post »
It seems like every subscription business (wireless carriers, cable/internet providers….) is worried about one thing in this down economy – customer retention. Analytics teams in these companies are focused on building “customer attrition” models. Typical attrition models take structured attributes and historical behavior of customers to segment them based on their “propensity” to attrite. The end result of these attrition models is simply a statistical prediction of a customers’ likelihood to remain loyal and/or leave. By applying the model to current customers, a prediction of future loyalty is obtained and customers are separated into high-risk and low-risk groups. Retention campaigns can be targeted to the high-risk groups.
Read the full post »
Most of the analysis and outputs from CSAT surveys are focused on what needs to change at the contact center. Whether it is an improvement in agent performance or the type of training or “bringing up a center” to the network average, the call center agent directly come under the microscope. However a quick analysis of the extent of problems/ poor customer experience will show that nearly 50% of the problems originate outside the contact center.
Read the full post »
I was recently talking to a few customers about their opinion on CSAT surveys. One of the interesting view points that came out was the actual number of CSAT surveys that were being done and the impact of that on CSAT. The typical average collection of survey data was around 1.5% to 2% of the interactions; however important action items or decisions were being taken on the basis of this small sample. (Also many of the action items from all of these surveys center around agents performance, which is still only part of the story)
Does a 1.5% to 2% response warrant sweeping changes across the call center? Or should companies try to increase this to 20% and start taking actions? In a quick sample of some surveys and the survey questions one thing was evident. One of the reasons for the low response rate was the number of survey questions in the questionnaire. The average survey questions ranged from 22- 27. So, one approach is to reduce the number of questions per survey from 22 to say 7 and that will improve survey response rates. One company in the media/ entertainment space did that recently and it dramatically increased their surveys. However, the question that remained was - did it provide more accurate insights that could have driven different actions to improve CSAT? The variation was not very high. So what is better- fewer questions, higher number of responses or more questions fewer responses? More importantly, is there an optimum statistical mix in terms of number of questions per survey and the percentage of surveys that will throw up better insights for actionable recommendations? It differs by company, by industry vertical, their CSAT measurement infrastructure and the ability to translate insights into actions.
While surveys present one part of the solution they do not adequately cover the whole picture to take the right decisions. The answer lies in looking at all aspects of interactions - post interaction, pre interaction, and in the moment of interaction, and other attributes to arrive at the right set of actions. Else this will continue to remain only a half solved puzzle.
While I was thinking about how we at 24/7 Customer are helping our customers to serve their end consumers using the N=1, R=G model, I started looking at how we are using it for our own business. Being a global company gives us a lot of opportunity to leverage both those formulae internally, and there are several areas where they have been applied. A couple that stood out for me are in the areas of product innovation and process improvement.
Read the full post »
|
