Disciplined Process Design Lets Customer Teams Absorb Volume Without Hiring

Customer service teams answer rising contact volume by hiring, and that answer costs more every time volume rises again. Only 14% of customer service issues are fully resolved in self-service, which sends most contacts to a person during a marketing push or a system outage. Convenience retail runs both of those events on a regular cycle across hundreds of sites and several brands. The teams that handle them without proportional hiring have documented their process and calculated their staffing before either event begins.

Adam Wood is Head of Customer Engagement at OTR, the Australian convenience retail network. He leads customer operations, customer intelligence and transformation across the group, and he held similar roles in financial services, software, education and home services before joining OTR. Wood puts the network he supports at roughly 1,000 locations and more than 24 brands, carrying over 600 million customer transactions and 150,000 customer contacts a year.

"Scale is always an ongoing discussion of how we can handle getting bigger without what I call 'the human body problem.' The traditional approach is to just hire more staff, which creates its own layers of complexity and its own layers of expense," says Wood. Wood's alternative is what he calls "full utilization", and it depends on knowing what is arriving and sorting it with the tools a team already owns. Filtering routes contacts to the right place, self-service removes the ones a customer can settle alone, and customer education prevents a share of them from being sent at all.

Starting with the SOP library

Everything Wood does in a new organization starts with standard operating procedures, ahead of any question about tooling. Six people can each reach the correct outcome by a different route, which leaves a manager with six processes to improve instead of one. Writing down a single method separates the mechanical steps from the judgment calls so a manager knows which parts can be automated and which need a trained person. "If you've got a team and they're all doing good work and coming to the same outcome, but they're all doing it a different way, that's not scalable, that's not repeatable, it's not trainable," says Wood.

At OTR the finished library moved into the CRM as a like-for-like replication, with no attempt to improve anything during the migration. Wood's team pared the inbound, outbound and unique-event journeys down to a minimum viable process, then rebuilt from there in increments. Redesigned workflows and service levels cut case resolution time by 37%, a figure he reports from his own team. "I always start with the fundamentals first, the basic lines of communication," Wood notes. "Get that clean, repeatable, scalable through repetition."

A team that has written down its process can plan capacity with arithmetic. The steps, the time each takes, and the expected volume are the only inputs it needs. A promotion planned six months out and a payment system failing overnight pose the same question. Wood reports a 66% cut in his cost to action, the amount the business spends resolving a single customer request. "You compare that structure to your FTE calculations and volume," Wood adds. "The calculation will instantly tell you I need Y number of staff to handle this volume over Z amount of time."

Training beyond the script

A documented process handles the contacts a team already understands. The rest arrive as questions outside what procedures cover, and most operations respond by writing more procedures. Wood spends the training time on product knowledge. His staff learns how the product works and what happens when it reaches a customer. He then teaches a language technique they use to build the answer themselves. "The mistake people make is trying to plan for every eventuality and trying to plan for every individual question," Wood says. "I don't use a lot of scripts, I use examples."

Training pays off the first time a question lands outside the documentation. A representative who understands the product can adapt the standard procedure and answer on the spot, without escalating to a team leader. "Rather than saying, here's 60 templates to use, I teach them about the product and then teach them how I would like them to convey that information," Wood explains. "If something suddenly breaks and you get 40 times your usual weekly volume in a day, you're not just providing cookie cutter responses to frustrated customers. You're providing detailed, unique, customizable answers based on unique circumstances."

Building the data triangle

Wood applies the same minimum viable approach to his technology stack, and the design principle is one record per person. A representative opening that record knows what the customer bought, what they complained about last year, and how the last issue ended, without asking them to explain any of it again. Wood puts the feedback side of that record at more than 500,000 customer interaction points and over two million ratings and feedback records. "Everything from Google reviews to emails to sales queries all come into a single CRM and are attached to a unified customer profile," Wood notes. "We know that this person who made a Google comment last year is also this person who's asking about this receipt."

Wood combines three data types on that record and calls the result a "data triangle". Contact information forms the first point, demographic data by region the second, and sales data the third. Wood describes a customer in her forties whose suburb carries an average household income his team can identify. Those two facts alone determine which products get promoted to her and whether the message goes by text or email. His team applies the same three points to ten customers or a hundred thousand without adding anyone to do it. "You need to know how to replicate those answers more readily than throwing more people at it," Wood concludes.