A common scenario of interaction event tracking
Let’s paint the picture of a common series of events in the contact center.
1. A technical or performance issue is suspected – either through anomalies in high-level aggregate reports or an agent/customer complaint. From a management stand-point, the immediate question is: “Is this a recurring problem or a singularity?”.
2. Depending on severity, technical resources then get assigned for a deeper analysis to validate and isolate the occurrence, which often comes through a time-consuming data analysis by those technical resources involving log files or interaction level data.
3. If a pattern can be identified, an organization may create another historical report or real-time dashboard to deliver a count or list of repeat occurrences.
4. Companies can then attempt to take corrective measures and insert a process to react to an issue recurrence through that same reporting mechanism.
3 major issues with this process
A typical issue resolution process is a slow process in a fast-paced technological environment where every second counts to the bottom line and Net Promoter Scores drive business success.
1. Historical reports are important, but it is a reactive solution that sometimes delivers limited information only. Validation and isolation of issues must be done through other means of analysis, which is often manual.
2. The log in report creation, delivery, and presentation of information makes the resulting returns unactionable for the initial issue occurrence. If new parameters or data elements need to be tracked for the new issues, companies are forced to let data accumulate for issue recurrence validation. You can only track from the creation date forward without backwards analysis into how long this has been going on.
3. Real-time dashboarding presents aggregate level data, which is not granular down to the interaction level. When looking at aggregated data, it often will take several occurrences to move an average or count metric to exceed a threshold resulted in repeated poor customer experiences that are not auctioned.
We’ve come to accept the above, along with the few dissatisfied customers that experience issue recurrence, while resolution is driven by this slow process.
However, what if we can action poor experience in real-time? The concept is not new and some technology has been around for very specific scenarios. Overflow routing is one example. The longer customers waited, the worse they viewed their experience – so why not have them speak to someone with another focus in the contact center who’s free. Variations on a theme saw skills-based routing with a timeout extension of the target pool to include more and more potential agents as time elapsed in queue. But, these are very specific issues.
Why do we still do it this way?
Standard metrics and contact center best practice are set based on our technical ability to monitor system, agent and interaction behavior. When innovation stalls or is stagnant over an extended period – that standard becomes ingrained in the industry. This happens because companies stop bothering to look for alternative or better tracking methods and accept the limitations as industry norms.
But, every so often there’s a technical leap that pushes limitations to new levels, allowing us to do more and do it better.
Shifting to a new process with innovative technology
A more flexible and agile approach would mean the ability to automatically monitor all contact center activity from an agent, interaction and technology behavior approach. This requires the need for data to be captured at the interaction level. This concept is also one that has existed for a long while.
This is a slow moving process that adds significant time in an ability to take action on a captured event.
Only recently have we seen integrated products with the ability to action system, agent and interaction behavioral data in a real-time manner. We’ve seen companies out there doing real-time voice sentiment analysis and identification, so that immediate action can be applied in the case of a negative or positive tone.
Artificial Intelligence platforms such as Watson, Alexa and Siri are starting to be integrated to filter event data streams and apply machine learning to potential outcomes. AIl can still be somewhat costly and complex to set up the types of learning one hopes to achieve but it is closer to an everyday reality.
Interaction level data capture will permit regressive analysis once an issue is discovered. If you’re capturing all the events – it is a matter of applying your identification algorithm to old data to see how long this has been occurring. Interaction level monitoring through an agile platform can identify single occurrences and action can immediately be taken as mandated.
Now, a new world best practice scenario
1. Report a problem through customer/agent complaint, reporting anomaly or pattern analysis of agile event level monitoring system
2. Analyze with agile event level data capture application to assess impact or existence of hypothesized issue
3. Identify the fix and either attempt to eliminate or reduce occurrences to an acceptable level
4. Create historical report to communicate success or failure of fix over time
5. Use agile event level data capture application to monitor for issue recurrence and automatically alert to take immediate action on resolution. Problem calls can now be flagged for added service as needed until problem is eliminated.
You can imagine scenarios now where your interaction level data monitoring can detect multiple call-backs of a high value customer and immediately enter a callback record into an escalation queue to make sure their issues are dealt with correctly.
Negatively flagged voice tones can initiate immediate barge in of managers to de-escalate. Agents who place calls immediately on hold can be flagged and action taken immediately instead of letting the behavior repeat for multiple customers.
…..And so, a new way of managing customer experience is born. We can see with this possibility more can be achieved in a truly agile manner in every sense of the word. For each now to imagine new scenarios that they would want to monitor and react immediately in their day to day operations.