Modern day customer support possibly has its roots in AT&T’s ‘automated collect calling’ launched around 1967. Since then internet based digital channels and now social have changed the customer support landscape dramatically when it comes to channel options. Demographic likes and dislikes are now well taken care of across voice to social with chats and mails bringing up the middle.
Some basics however haven’t changed much. Agents still do the heavy lifting despite the keyword or expression based ‘intelligence’ and pleasant discussion starters with sentiment sensitive choice of words coming from Conversational AI.
Consulting firms are still moving the needle with innovative team set-ups and lean, Coaching firms are doing their best with latest in coaching methods, system integrators doing their bit with scripts and BOTs and L&D folks are burning midnight wattage scheduling smart agent onboarding and refresher programs.
These efforts collectively have had a seminal effect and completely reset your and my support expectation baseline. That said no one still finds it easy to get a 5 on 5 in C-Sat surveys. And the VP-CX never finds the right day to call on her Boss.
We believe 5 fundamental gaps still remain to be solved.
1. Challenge in understanding what the caller wants ~ That’s why experienced agents are so precious. Keywords and programmed expressions do help a bit, but rules looking for “password-reset” invariably miss the meaning of “system’s not allowing me in” unless of course someone just keeps adding these expressions endlessly
2. Difficulty with remembering or locating the answer instantly ~ FAQs run short very quickly. Training materials are great but who remembers it all? Manuals have it somewhere but how does one get to page 74 (that shows the port where caller should push the plug in) ? And how many CRMs allow searches across internal and external knowledge bases anyway?
3. The holy grail of ‘satisfaction’ despite meeting all the SLAs that ever were ~ Customers would want their problems fixed and questions answered instantly, So Is it a surprise that 15 minutes for first response and 6 hours for resolution miss creating satisfaction? It is a race that only John Henry won in the ballad and he was the last.
4. Cost of providing 24x7 support with agents waiting on the other side of mail, chat or even a tweet
5. Lastly, inability to spot the bubbling issues of tomorrow early enough so as to plan and mitigate. Analyzing individual customer journeys have helped a lot in moving towards ‘care for one’ but there’s still nothing better than the Top 20 billboard and pivot tables to help prepare for tomorrow’s chats and calls
All these are playing out as new product and service categories enter the play and customers start calling on topics that were never thought to be needing any support whatsoever.
Customer Support system landscapes have therefore come to need few base capabilities:
1. Ability to cut through the wordplay and even language combinations as they are spoken in multi-lingual parts
2. Search and recommend right answers from across knowledge sources - internal or external, pdf or engineering drawing and render it in a way that obviates the need for last mile f
3. Elimination of non-value adding hand-offsunless it truly deserves a deep dive and hence needs an escalation. This is only possible when a system is constantly learning - once solved by an expert agent it is learnt by the system for good.
4. Ability to spot emerging issues from the mails to social chatter much before they get frequency heavy
Options that demand change of underlying CRM or engagement channels will not quite fit the post pandemic budgets as they would come with significant process, training and adoption challenges. An intelligence layer that could ideally slip into an existing landscape would be the real smart approach.
With patent filed purpose-built Bayesian and Deep learning methods Sainapse from Bayestree Intelligence targets filling each of these gaps that have challenged even best run customer support teams.