This is particularly challenging if your company is growing significantly year-over-year. How can you predict what your holiday sales will be? And what impact will that have on your volume of customer support tickets over the holiday season?
We recently went through this process with one of our clients, an e-commerce company, and wanted to share our five-step approach.
1. Start planning your holiday customer support staffing early.
Starting early is critical for two reasons. First, training: You need to build in time to make sure your new agents are fully trained by November. Second, you need to build in time to change your plan. If customer support ticket volume seems to be growing by more or less than expected, you want to make sure you can adjust your plan accordingly.
At this client, we started planning in August, and hired our first round of agents in September.
2. Develop a holiday sales projection.
First, you need a sales projection for the holidays.
Here’s the key: Ideally, you won’t just have a dollar projection, but a projection for the number of units sold, number of orders, and the number of customers. Why? These sometimes grow in line with each other, and sometimes don’t.
Maybe your customers are buying more items and spending more money per purchase compared to last year. If that’s the case, the number of customer service tickets may not rise in tandem with dollar sales, but with the total number of customers.
3. Find the right metric for predicting ticket volume.
Next, you need to translate your sales projection into a projection for customer support ticket volume. To do that, you need to find the right metric. Does your customer support ticket volume tend to grow in line with dollar sales? With unit sales? Or with your total number of customers?
For this client, we started with four simple calculations:
- Customer support tickets / unit sales
- Customer support tickets / dollar sales
- Customer support tickets / purchase
- Customer support tickets / customer
It turned out that customer support tickets per unit sales had the lowest variance over a 16-month period, so we decided to use that metric for predicting holiday volume.
4. Develop a hypothesis for your holiday customer support team size.
Based on those calculations, we came up with a prediction for holiday team size. The initial prediction showed that we needed to scale the team by 3x to meet holiday volume.
This seemed high. We didn’t want to end up under-staffed, but we didn’t want to spend more money than necessary. We also didn’t want to over-tax our trainers. So we decided to take it slow, hire a few people, and re-evaluate our hypothesis a month later when we had a little more data.
5. Test your hypothesis.
We tested our hypothesis in three ways. First, we developed a robust customer support queuing model. This type of model takes into account not just total ticket volume, but when tickets come in. If all your calls and chats come in between noon and 2 p.m., you need to hire more people than you do if tickets come in a steadier flow throughout the day.
The model indicated that our initial hypothesis would be adequate to manage the holiday volume of customer support tickets.
Meanwhile, we looked at ways we could stretch the capacity of our existing team. We considered allowing a higher first response time, and reducing the availability of online chat.
Finally, we evaluated new data that came in during August, September, and early October. Were the initial sales projections accurate? Was the number of tickets rising in tandem with sales?
We ended up deciding to scale the team significantly, but slightly lower than the initial estimate.
Of course, the real test is only just beginning! But we hope we’ve laid a strong enough foundation to adjust our plan in November and December if need be. The bottom line: If you run a high-growth company and you’re scaling your customer support team for the holidays, you have to be flexible. And err on the side of over-staffing — having a few idle agents is better than providing sub-par support to your customers.