Shared services leaders are increasingly expected to deliver real business value in addition to keeping the business running. However, attempts at shared services transformation and process improvement are handicapped by the opaqueness and inefficiency of email and comms-based workflows. Fortunately, natural language processing is now giving them unprecedented insight and business intelligence.
Running shared services and global business services (GBS) has become much more complex over the last year. COVID-19 and the rise of hybrid working have seen service request volumes explode, new cost and efficiency targets imposed, and the emergence of a pressing need for seamless, effective customer experiences.
It isn’t enough simply to hit your service targets. Increasingly, a service function has to create real business value. What’s more, this has to be achieved with limited resources and often lean teams. Business as usual won’t cut it - service leaders need new approaches for growing productivity and responsiveness, improving CX and driving competitive advantage through greater operational efficiency.
The first step in making worthwhile improvements is understanding the causes of workflow - that is, the specific requests that are triggering work for service agents. Even if change and process improvement aren’t priorities, this insight is crucial for planning and making the right resourcing decisions. The problem is - do you really know what your people are working on right this moment?
Service leaders increasingly depend on the likes of BPM, Process Mining and Task Mining to understand what’s going on in their line of business. But this provides an incomplete picture at best. Current tools almost all share the same fundamental flaw - they don’t provide visibility into service requests, the most important as well as the least efficient source of work.
Where work begins
The work of a shared services agent is inextricably based in comms. Almost every request and workflow starts with an email or service request. As a result, teams are spending enormous amounts of time in email, communicating with colleagues and internal customers. Indeed, the average agent sends and receives over 120 emails a day, and spends 40% of their time in Outlook alone.
There’s a very good reason why this is the case - communications are an invaluable source of insight into the running of a business. The messages an agent receives are a window into the problems plaguing employees, the causes of process breakdown, and opportunities for new services that haven’t been discovered yet. The challenge has always been to distill this valuable information from masses of emails and messages at pace.
However, email is also a cost centre and the leading source of service latency. Email costs businesses up to £10,000 per agent each year, but the cost rapidly scales with the size of the service workforce. Processes halt when an email is sent and only resume once it’s had a response. Service requests and support tickets are often sent to the wrong agent and must be manually triaged.
These delays are only magnified when work is primarily performed across shared mailboxes with no clear owner. When workloads are high, agents may ignore these types of requests in the hope that someone else will pick them up. The pressure to maintain response SLAs and performance metrics often results in more straightforward, transactional requests being prioritised instead.
Manual email processing is a necessary task for agents to extract the information they need from email. However, it is also a highly inefficient one. Even when performed correctly, there are few mechanisms for agents to collate and analyse all of this insight for the good of the business. The email is read, the request processed, and then the agent moves onto the next task. The shared services leader still lacks a complete overview of their service function and the work of their people.
NLP: The answer to unstructured data
Understanding service requests is the crucial first step in boosting productivity, improving insight and responsiveness in shared services and GBS. A lack of visibility here is a major limitation, so service leaders need to move decisively to close the gap.
However, this will be difficult without the right tools. Traditional digital transformation solutions - such as Task Mining and Process Mining - have struggled to extract information from email because it’s expressed in freeform natural language. This represents unstructured, complex data that isn’t machine-readable. Humans understand this kind of language innately and can interpret context to fill in the gaps and make the best decision. But this has been very difficult to train machines to replicate - hence why the bulk of email processing is still undertaken by human employees.
However, an important shift is building in shared services. The leading organisations are beginning to solve the challenge of email and unstructured data with solutions based in natural language processing (NLP). More than two-thirds (67%) of SSO leaders now view NLP as an important investment priority for extracting data from emails. This follows game-changing advancements in the technology over the last few years. The newest NLP models now routinely beat humans in language and reading comprehension, enabling NLP to accurately and reliably understand masses of unstructured email communications at speed and scale.
Conversational Data Intelligence is a new class of enterprise software that uses machine learning and NLP to give service leaders total insight and visibility over their communications channels, including email. Through a process called Communications Mining, shared services leaders can rapidly extract and analyse the most important information from email channels, gaining unprecedented insight into the drivers of workflows and the causes of process breakdown. This process also supplies the structured data that fuels automation, uncovering new automation opportunities that build on agent productivity.
You can’t manage what you can’t measure. A services leader won’t be able to understand what their people are working on, what requests are driving workflows, and where manual processes are breaking down, without having real-time visibility over email. Conversational Data Intelligence does just that, supplying the raw data needed to make the right decisions and level up shared services as a whole.
To see the benefits of Conversational Data Intelligence in action, watch our product tour of the Re:infer platform.