A 5-Step Process to Forecasting Sales

Dear Dr. pROfIt: I'm a web analytics expert, but my boss recently asked me a question that has me stumped. She asked me to predict what our e-commerce sales will be in 2015. We currently generate $35 million in online sales. Should I just increase sales at the rate of inflation? Should I purchase a research report that outlines e-commerce growth over time? Honestly, I have no idea how to approach this.

Dr. pROfIt: Fortunately, this doesn’t have to be a difficult question. Use your existing web analytics software tools, if possible, or have someone in your IT department run a custom query for you:

1. Count your number of customers from 7/1/08 to 6/30/09.
2. Count the number of customers in step one who purchased again from 7/1/09 to 6/30/10.
3. Among the customers who purchased again in step two, calculate the average order value for these customers.
4. Count the number of customers who purchased between 7/1/09 and 6/30/10 but didn't purchase between 7/1/08 and 6/30/09.
5. Among the customers who purchased again in step four, calculate the average order value for these customers.

These five steps are all you need to produce reasonable forecasts into the future. Here’s how:

• Let's say you had 100 customers purchase last year.
• Assume that 40 customers (40 percent) purchased in the next year.
• Assume that these 40 customers spent $200 each in the next year.
• Then let's say 70 customers purchased in the next year, but didn't purchase in the past year.
• Let's say those 70 customers spend $100 each in the next year.

With this data, you know that you generated $8,000 from existing customers (100 * 0.40 * $200 = $8,000). All other customers generated $7,000 (70 * 100 = $7,000). You generated $15,000 in sales last year.

The following year, you have 110 customers to start the year (100*0.40 + 70 = 40 + 70 = 110). If everything else remains the same, you'll generate $15,800 in sales (110 * 0.40 * $200 + 70 * $100 = $15,800) from 114 customers (44 + 70 = 114 ).