Everything you need to know about forecast accuracy
Responsive, human support is needed now more than ever. Creating a support environment that fosters this spirit day in and day out is a combination of culture and—more mundanely but just as importantly—staffing calculations derived from forecasted volume. Accurate forecasts allow managers to plan for the future more consistently and allow agents to focus on delivering the best possible customer experience without worrying about a blinking red queue. As defined in our Essential Workforce Management Glossary, forecast accuracy is the degree to which your prediction of support volume reflects the reality of observed support volume. In other words, how well were you able to predict the future?
Who needs a forecast?
You may be asking yourself, why bother forecasting volume in the first place? We're not able to predict human behavior with 100% accuracy meaning that any forecast will be wrong anyways, so why does it matter?
It’s true that we can’t predict the future with 100% accuracy (if you are able to reliably predict incoming support contacts with 100% accuracy, go out and buy a lotto ticket today or come join us at Assembled and tell us your secret!) But a forecast doesn’t need to be flawless to be helpful. Improving the accuracy of your forecast, even marginally, has deep implications on your organization’s cost-effectiveness since it is ultimately what your staffing projections should be based on. This, in turn, informs hiring plans and fosters predictable response times. A good forecast captures trends in your arrival pattern and allows you to plan headcount rigorously against customer demand rather than responding reactively to periods of over or under-work.
In fact, an accurate forecast of your team’s support volume is at the base of an efficient and human support organization. Don't underestimate the power of planning, or the stress that lack of planning can lead to—an understaffed team with a chronic backlog will show it in their day-to-day interactions. This calculation also has downstream effects on the team's handle time and backlog within a given day. The further you are from reality, the more wait time for users and the more idle time for agents.
Since forecast accuracy is ultimately about finding the Goldilocks point within the set of possible outcomes, there's no distinction made between an over-projection and an under-projection in the calculation:
forecasted volume - absolute value of the deviation
= ━━━━━━━━━━━━━━━━━━━━━━━━━
forecasted volume
A forecast accuracy of 100% means that you received the exact number of support requests you expected to during a measured interval. For example, if your expected chat volume was 100 users during an hour and you received 100 chats in reality, your forecast accuracy would be 100-|0|/100 or 100%. We’ll take a look at both over-projection and under-projection below and explore their implications and solutions but first, here's a template to illustrate how forecasts and actual volumes interact. Preview it below or click here to make a copy of the template:
Over-projection
A 5% forecast over-projection means that you received 5% less contacts than you expected to during a given interval. For example, if your expected chat volume was 100 users during an hour and you received 95 chats in reality, your forecast accuracy would be 100-|5|/100 or 95%.
From a customer perspective, over-projection is not necessarily a bad thing. All things equal it should result in faster response times and the possibility for longer handle times yielding more human support. For your organization though, over-projection can be costly in terms of idle time for agents.
Short-term fixes for this include allowing agents to flex into non-support work when volumes drops or offering intraday staffing flexibility measures, such as voluntary time off during these times. These measures ensure that your support organization is running efficiently on a given day but if the trend continues, some long-term solutions to consider include reducing headcount or creating additional time-insensitive project work. Dips in volume can be great times to redirect attention to goals like tagging data sets and creating or updating canned responses for the team’s future efficiency.
Under-projection
A 5% forecast under-projection, on the other hand, would mean that you received 5% more contacts than you expected to. So, if your expected chat volume was 100 users during an hour and you received 105 chats in reality, your forecast accuracy would also be 100-|5|/100 or 95%.
A forecast under-projection leads to longer wait times for users when left unchecked and a potential for more rushed-feeling, transactional support. For agents, the situation isn’t much better. Backlogs create stressful workplace environments, particularly if the root cause is not addressed.
Immediate solutions include an all-hands-on-deck approach, pulling anyone and everyone who can help back into the queues. In small doses, coming together like this can have a galvanizing effect on the team. But if the root cause is not addressed and it becomes a chronic backlog of emails from exasperated users without an end in sight, the team can easily become dispirited and rightfully so. Long-term solutions focus on the root of the problem, which is but there's more demand for support than there is supply. Hiring additional headcount internally or working with a BPO vendor can help to smooth these spikes.
But it’s not just about headcount. Process automation can free up the existing team to handle the same number of contacts more efficiently. Similarly, if there's a particular question or concern that pops up disproportionately, adding an answer to an FAQ page on your website or better yet, working with Product to change the underlying issue will reduce volume upstream and minimize the frequency of these moments.
—
In a perfect world, times of under-projection and over-projection can work together in harmony to create a more efficient workplace. Adding to an FAQ page is a worthy goal but not necessarily when the queues are red. In the times of over-projection when agents would otherwise sit idle, the team can work on these gains in efficiency that can that be used in the times of under-projection.
Related content
