Building a Forecasting Model Using the Six-Pack Metrics

Forecasting is the most important exercise you can do to understand how your business is working, anticipate challenges, and test what levers you can pull to hit your revenue and sales goals.

Remember the six-pack metrics? That set of six key eCommerce metrics we recommend monitoring daily to keep your finger on the pulse of your business? That handy bunch of metrics is also useful for another business essential: forecasting. 

Forecasting helps you make smarter decisions about your business. Yes, it’s an educated guess about how your business will do next year, but it’s also the most critical exercise you can do to understand how your business is functioning, anticipate pain points, and hypothetically test what levers you can pull to hit your goals.

Forecasting models also provide you with a benchmark against which to measure performance—which can be essential if you’re a start-up and your year-over-year growth was so wild that you can’t depend on those numbers to inform what the business will do in the following year. 

Go in the right direction: bottom up or top down? 

How do you build a forecast? For example, let’s say you think you can double sales next year. How did you arrive at that projection? Is it realistic? How are you going to achieve it? There are two ways to build a forecast: bottom up or top down. We recommend a bottom-up forecast for several reasons. First, it’s more accurate. Second, it helps you explain to anyone who wants to know—like your boss or an investor— how you determined it. And third, it lays out pretty nicely the key levers you have at your disposal to adjust daily, weekly, and monthly to keep your business on track. 

However, it doesn’t hurt to also use a quick top-down calculation to double-check that your bottom-up forecast is in the ballpark. A top-down forecast is a great way to get a quick approximation of your sales/revenue goal based on an estimate of your total customers and their total spend in a year.

Bottom-up forecasting models derive the sales/revenue goal through estimating how much traffic you can drive to your website (via marketing), and how many consumers you can get to convert to purchase and spend at a certain amount (maybe via improvements to user experience). If the top-down and bottom-up total sales/gross revenue numbers aren’t relatively close, there’s likely something off with your inputs for one or more of the six-pack metrics; for example, you may need to adjust your marketing or sales channels to increase your traffic and conversion rate.

Do the math: how to forecast

A simple bottom-up forecast using the six-pack metrics would look something like the following. 

First, you’ll need to make several assumptions, based on historical data (if you have it): 

  • How much traffic you’ll have on your website next year (to derive this number, consider if traffic will be organic or if you’ll need ads to drive traffic)
  • On average, how many customers will convert to purchase
  • What you expect your average order value (AOV) will be
how to forecast in three easy steps
Here's how to forecast and check your forecast.


Take forecasting to the next level

These simple calculations above are just the starting point for what you can do with forecasting. You can use this high-level forecasting model to help you think through any high-level changes you’ve made or plan to make that will impact your business in the next year (e.g., new product line, website improvements, etc.).

In addition, you may want to consider breaking the high-level forecast down further for a more granular view that provides more opportunities for optimization. For example, you could forecast performance monthly and/or by sales channel, such as by Amazon and Shopify. Or forecast performance at the category or product level to help you plan which products you should feature in marketing campaigns. 

Sales forecasting

Sales forecasting is predicting actual sales for a specified period of time—typically weekly or monthly. While you can use historical sales data as your baseline, making a forecast weekly/monthly for the next year should incorporate multiple variables, such as: 

  • Will you run the same/different promotions? 
  • Will you be using the same/different sales channels? 
  • Will you be using different marketing channels, reallocating marketing budget, etc.? 
  • Is the holiday calendar different this year (which could affect your monthly forecast)? 
  • Will you have new products/services? 
  • Will you be changing pricing? 

Taking a close look at each of your marketing and sales channels and optimizing them to focus budget where it’s most effective gives you information about how much you need to spend to bring in a certain number of orders per week or month—which then you can roll up into your total sales estimate. 

Demand and inventory forecasting 

Demand forecasting helps you determine how much demand there will be for products/services during specific periods, which helps you optimize inventory—and fulfillment planning. There’s nothing worse than running out of a product and missing potential sales or investing in too much of a product that doesn’t move. Doing this forecasting involves looking at products at the category or SKU level. Demand can be impacted by multiple factors, including: 

  • Any changes in customer purchasing patterns
  • Changes in the holiday calendar or other events that triggered demand last year
  • Changes to marketing or sales channels
  • Spikes/reduction in sales of complementary products (e.g., if we sell more product A, will we also sell more product B?) 

Test impacts of different levers

What happens if your conversion rate goes up? What if you increase AOV? As you build out the forecast(s), experimenting with the different six-pack levers can help you think through different scenarios of how you might optimize your business—and what else might be affected. For example, if you want to increase conversion by .1%, it will increase revenue by X, but… what has to happen to increase conversion? How will you achieve that goal? More marketing? Changes to the website? 

Watch out for common pitfalls

A forecast is only as good as the data you put into it. Therefore, it’s good to follow a few best practices to make your data as accurate as possible. Following are a few common mistakes that can really mess up a forecast. 

  • Not keeping a close eye on the dates: When forecasting month-to-month (using historical data as a baseline), consistency in what you’re measuring is key. Be aware that holiday dates can shift year to year. For example, Cyber Monday can sometimes fall in November or December, which would shift demand and affect your monthly estimates and subsequently metrics like your ROAS and customer acquisition costs (CAC).
  • Not asking for enough input: Because forecasting can be impacted by so many factors, it’s a good idea to ask for department-specific forecasts or run numbers by the team members that manage marketing, inventory, fulfillment, etc. day to day. If the numbers aren’t coming together, you have to have the hard conversations to find out why. 
  • Not using historical data: Last year’s data is the place to start to determine what promotions you might want to run again this year or what potential issues you might have to address in order to come in at or above plan.  

Keep looking forward

If there’s one piece of advice we have it’s this: do forecasting. It does take effort. You may feel like a forecast will lock you into numbers you can’t achieve. But the reality is, if you’re a DTC brand and your business isn’t hugely complex, doing a bottom-up forecast might take the team just one to two weeks to complete. If you’re a young start-up without much historical data, forecasting will be even faster because you’ll depend mostly on the simple top-down method. And, if you find your forecast seems way off the mark, redo it!

It’s better to admit you were off and fix it than continue to operate in the dark. Your business will thank you.