Five Ways AI Can Power Branch Workforce Management Success

Forecasting is at the heart of efficient bank branch staffing and cost reduction. But in a post-COVID-19 banking world, branch labor planning and management will be more complex than ever. The largest driver of this complexity is the change in how customers consume banking services.

High volumes of transactions are migrating to self-service channels, such as online and mobile banking, and even in-branch “smart” ATMs. Over the past several years, even during “business as usual,” this shift made forecasting transaction levels more difficult. Now COVID-19 has accelerated these trends, increasing the challenges and burdens for banks’ forecasting and scheduling teams.

Change Management Done Smart

COVID-19 forced banks to quickly make a wide variety of changes to branch processes and operations. Naturally, banks have had to rethink everything, from which branches can open due to available staff, to the need for plexiglass in front of all tellers, to specific cleaning processes, and so on.

Given these issues and the digital adoption ramp, banks increasingly need to re-tool their workforce management tools and processes. These can yield great benefits through smart improvement and optimization. If banks want to keep their remaining branches profitable and efficient, they need to be proactive with workforce scheduling models.

An accurate forecast can optimize network-level staff planning and enable faster, more effective bank responses to these changing trends. For instance, a forecast can enable branch-level staffing guidance to support a reduction in customer wait times, among other outcomes.

Given all this change and uncertainty, leveraging artificial intelligence can support better data-driven decisions in your forecasting and scheduling processes. With that, here are five AI-powered forecasting practices your bank potentially can implement today.

One: Factor in All Banking Variables

Accurate banking forecasting typically requires a wide variety of variables, similar to those in other customer-facing industries, such as retail. They include government check days, holidays (including the lead up and days after), week of the month, and day of the week, in addition to any bank marketing initiatives.

However, when these variables overlap on the same day, traditional forecasting models typically overstate what the actual transaction volume will be. What’s most important are the interactions among these variables, and how the variables will behave when two or more are present.

Deep learning and neural network forecasting models, by their very nature, take into account interactions between variables. It’s best to use a forecasting module that utilizes the latest and best-in-class forecasting algorithms. This will not only result in a more accurate result, but should simplify the forecasting process, as well.

Two: Forecast with All Transaction Channels

Banks typically forecast all their transaction channels independently. However, it’s much easier to get an accurate forecast and understand customer patterns by looking at all transaction channels combined (in-person, ATM, mobile, and online). This is especially important now, as so much volume has temporarily shifted away from the branches.

Changes at the branch level make much more sense in the context of overall transaction migration. Total transaction volume and the percent makeup by channel are generally stabler and easier to forecast.

Three: Build Potential Traffic Disruption Into Your Models

Even before COVID-19, disruptions to branch traffic (including natural disasters, road closures and the like) all were unfortunately common. Post-COVID-19, you can explore the creation of new variables as part of the forecast to help your predictive model better understand the underlying customer patterns.

Examples include: “days since shelter in place lifted” or “percent of neighboring business open.” These variables are closely tied to customer sentiments and the likelihood of customers visiting a branch in-person.

Four: Fix your Forecast Models on the Fly

Quickly tweaking your forecasting model is crucial. Using an accessible platform—where business users can directly control the variables and parameters, is paramount. But if you are dependent on your software vendor for assistance in variable selection or for creating new variables, you lose both flexibility and agility. Managers should be capable of doing such tasks without vendor or IT support.

Five: Monitor Exceptions

Once a forecast is live, monitoring for exceptions becomes the core duty. To quickly and efficiently identify what areas to review, it’s crucial to have robust dashboarding and business intelligence capabilities. For example, reports that consistently highlight your top branch forecast variances accelerate corrective action. This visibility also simplifies communicating and sharing business exceptions with executives and field leadership. Ultimately, exception-based management can free your team’s time for more strategic initiatives.

Good forecasting, in conjunction with smart workforce planning and mobile-first scheduling, enables you to get the most out of your associates. Instituting these five activities will help you ensure the highest possible forecast accuracy in this dynamic and variable environment.

To learn how Reflexis can help you achieve optimized forecasting and scheduling, reach out today or join our upcoming webinar: “How Superior Branch Execution Can Drive Customer Engagement & Efficiency.”