As your company grows, effective forecasting techniques ensure that your team is always prepared to meet customer and business needs without overspending—so, it’s a necessary process!
Forecasting early allows you to invest proactively in things like:
- Number of agents
- Management
- Business process improvement
- Technology implementation
Healthy forecasting will lead to better service levels, higher customer satisfaction, and optimized operational efficiency.
So, let’s all get better at forecasting. This article maps out a bunch of key concepts with people who have done it before—including leaders at NEOGOV and Snowplow—and shares a simple forecasting model at the end to get you started.
First, Some Context and Where to Start
You need to consider existing service level expectations as well as your company’s unique targets and values.
Tracey Virtue, President, High Line Software (a NEOGOV Company) and Chief Customer Success Officer (NEOGOV) shares an excellent example:
The factors you consider in your forecasting model need to reflect your company’s core values. For example, if customer-centricity is a core value, your Average Handle Time (AHT) needs to factor in time for nurturing relationships.
Here are four more considerations.
Customer Region
If the scale is global in nature, coverage and language might be a priority. Catering to a global customer base often requires support availability across multiple time zones. This adds complexity to forecasting since this will require shift work or hiring localized teams with minimal or no overlapping work hours. The complexity increases even more if you choose to offer multiple languages.
New Markets
If you are expanding into a new market segment, you may have different demographics or customer expectations and demand in that new area. This could include new support channels, longer operating hours, or even new types of support services such as a premium offering.
Product Complexity
As the complexity of customer queries increases with business growth, so does the need for specialized skills and efficient escalation processes. Forecasting helps in planning for the recruitment and training of support agents with specific expertise, and in designing escalation paths that minimize resolution times for complex issues.
Seasonality
If the scale is temporary, such as seasonality of your product or service you also need to consider rapid ramp up and ramp down trends. How you staff most cost effectively year round might have to change based on the trends you observe.
Forecasting Methodologies
There are many ways to forecast, but most importantly, you need a model or methodology.
Charlotte Ward, Director of Support at Snowplow, has a global support organization and has seen the value of forecasting first hand. She sums up the importance of forecasting well:
Whatever data you apply and how you choose to apply it—if you don’t have a forecasting model, then you’re shooting from the hip when it comes to scaling your team.
The down sides are incredibly impactful. She continues, “a poorly scaled team is unsustainable: individuals suffer from boredom or burnout. Under or over staffing are each a commercial risk to your organization, and both also deliver a poor customer experience in the long term.”
Here are a few concepts and methodologies to avoid those negative impacts.
A Simple Forecasting Model—Capacity Planning
Capacity planning is a great way to forecast for relatively small teams, with low complexity. For example, one language, support tier or geographical region. Always start with understanding your current productivity.
Step 1: Productivity
Productivity measures how efficiently your support team can handle incoming tickets. This is essentially the number of productive hours an employee has in a given month. Often this is calculated by taking the total number of hours in a period of time on average (for example 40 hours per week, or 168 hours per month) and then dividing that into the number of tickets in that same period to get a sense of the number of tickets per hour an agent can handle. For most teams it is as easy as three steps:
| Metric | Value |
|---|---|
| Volume in Period on average | 950 tickets/month |
| Current # of Agents handling volume | 5 agents |
| # of working hours in period | 21 days * 8 hours = 168 hours |
| Average Handle Time | 30 mins |
Step 2: Time spent not taking tickets
Next we must account for shrinkage and occupancy which are times where an agent is not taking tickets. Shrinkage accounts for the time agents spend on breaks, training, vacation, sick leave, meetings, and other non-productive activities. Add all the average times of any time spent outside of core support duties together to get a number of shrinkage hours to reduce the total available hours by.
Occupancy refers to the percentage of time agents spend actively handling queries. There is no right number for occupancy. In some heavy volume call centers, it might be close to 100%. Meaning that the user finishes one ticket, and is immediately presented with another. In others it might be closer to 75% where there is a gap between each new ticket. You can find occupancy by reviewing your incoming ticket patterns and trends. If occupancy is too high, burnout from over-occupancy is likely and too low may mean you are overstaffed. A good industry standard is 85%.
| Shrinkage | Expected Allocation | Average Hours per Month |
|---|---|---|
| Planned Time Off (Vacation) | 15 days/year | 10 hrs/month |
| Unplanned Leave | 5 days/year | 4 hrs/month |
| Internal Meetings | 1 hr/week | 4 hrs/month |
| Process Training | 2 hrs/month | 2 hrs/month |
| Professional Development | 2 hrs/week | 8 hrs/month |
| Breaks/Lunches | 1.25 hr/day | 27 hrs/month |
| Total | 55 hrs/month |
Step 3: Putting it all together
The two tables above allow us to understand our Productive Hours in a given period per agent.
One way to look at this data is to calculate how many agents you need based on your desired handle time. In our example, our desired ticket handle time is 30 mins.
This shows that each agent can handle 192 tickets in a month. At the simplest form, you can then divide that into your forecast for any month. For example, if you think you will grow 20% next month and we know current volume is 950, that would mean 1140 tickets (950*1.2) are forecasted.
This relatively simple calculation is easily aided by the metrics from your helpdesk tool. Usually, you can easily report on tickets by a given time period which allows you to do rough calculations for headcount needs and shift schedules and forecast for any amount of volume expansion or volume spikes.
Further Reading: The Erlang Model
For more complex teams, the Erlang model is often a best bet for forecasting headcount needs. It is used heavily in call centers and works by combining historical data with anticipated ticket trends. One of the key differentiators above basic capacity planning is that this model adjusts the predicted number of agents required based on your desired service level. It accounts for variables such as average handling time and incoming query rates, providing a robust foundation for staffing decisions.
While the Erlang model is significantly more complex to calculate, it does provide a high level of accuracy for forecasting staffing needs based on historical data and anticipated ticket trends. It can also scale for very large businesses with 100’s of agents all while optimizing resource allocation so that overstaffing is reduced.
To get started with Erlang, you can find calculators online or add-ins to Excel. Most of these calculators and software tools can simplify the process. All you need to do is input your AHT, query rate, and desired service level to get the required number of agents.
Workforce Management Tooling
Workforce Management Tools (WMT) are software packages that can help with real-time coordination of your agents between channels, staffing peak and valley times, organizing and reporting on vacation or adherence to the schedule, etc. Tools in this market can automate a lot of the capacity planning and calculations for you.
Cost Control
Another factor in planning headcount and support costs are the tools and staffing models you use to augment your team and provide the service levels your customers expect.
Automation
AI-enabled CX tools and other automation techniques are rapidly expanding. If you want to increase agent productivity or reduce headcount needs, understanding where your inefficiencies are and what your customers expect of your services is vital. You can often save money by implementing a tool that reduces the time it takes to update or respond to a ticket. You can use conversational AI to answer tickets directly to a customer even when your team is busy or out of office.
A healthy consideration during forecasting is to identify the types of queries that can be automated, the volume of such queries, and the potential cost savings. This allows for strategic investments in automation technologies that enhance efficiency and change your capacity planning assumptions.
Outsourcing
Using an outsourcing partner is another cost control measure as you scale rapidly with your business. Outsourcers are set up to ramp up and down staffing faster and hire across multiple languages and timezones more easily. They also have top-tier experience in forecasting and headcount planning to help you make cost-effective decisions. Partnering with an outsourcing company can help you achieve more efficiency at lower cost and comes with the expertise to recommend tactics at any stage of scale.
The Importance of Forecasting
While forecasting is ultimately a mathematical exercise and will always have a degree of inaccuracy, it is a required activity.
No matter which method you choose, you must make a reasonable decision and test multiple methods to see which works best for your business.
The best way to forecast support cost and headcount is to make it a frequent and consistent exercise. So, start now, even if you aren’t thinking about a scale up period yet.
Get Started with Forecasting Support Headcount
Editor’s note: If you’re looking for more tips or if you have any questions about forecasting headcount, contact Craig directly on Linkedin (Craig is an independent CX consultant and former head of CX), or get in touch with a partner manager at Influx who can help.
