As we all know, improved forecasting can help a company serve its customers better, and in a more cost-effective way.
In the world of physical products at DRY Soda Company (where we deliver our goods via trucks, not electrons), we see this every day. If the forecast is off, we might have inventory of our sodas in the wrong location. That means we’ll need to pay higher costs for expedited transportation, to fulfill a customer order on time and in full—and it’s also a drain on our team to chase these errors around!
With a little more upfront focus on forecasting, we can avoid those expenses—while also reaping the benefits of lower inventory. Inventory not only ties up valuable cash (which is critical for a small company), but it also costs money to store in a warehouse.
This year at DRY Soda, we’ve been able to improve our inventory turns by 33%, in part by improving our forecasting processes. Not only does this save us money, but it ensures that our customers get to enjoy the freshest soda possible. Good forecasts mean that we don’t make anything earlier than our customers are ready for it, and that we rarely have product expire, which can be very expensive.
Here are our top 5 tips on how to improve your forecasting game:
1. Find the sweet spot: 50,000 feet or in the weeds?
One of the most important forecasting decisions is the level of granularity that makes sense for your business. This should be done across three dimensions: time, product and customer. There is no right answer for how granular your forecast should be—it really depends on your industry and business.
At DRY Soda, we forecast on a monthly basis, at the individual product level. For example, we estimate how many cases of Lavender DRY Sparkling we plan to ship in September 2018.
On the customer dimension, we group some of our customers together (such as “Natural Channel East” or “Direct Store Delivery Distributors—West”), while we give our larger customers their own forecasts. For example, a customer like Target moves through a massive amount DRY Sparkling bottle 4-packs, so we need to have enough of each SKU that they carry to fulfill their orders.
2. Look both ways! Balance what you know about history and the future.
Historical performance can be a great place to start forecasting, but in all but the most staid businesses, changes abound. Blending both a historical and future-oriented perspective is especially important when a few customers drive a significant portion of the volume.
When DRY Soda had a lot of small customers, historical volumes and flavor splits were accurate enough to support inventory and production planning. Once Kroger, Target, and CVS came on board, it was essential that we started to factor in future plans, because those customers can each move the needle drastically. Even in those cases, a sense of history still helps provide insight into unique customer behaviors.
3. One is the loneliest number: Get the right people involved.
Where should forecasting live at a company? Operations, Sales, Finance? The right answer is D: All of the above. With each group approaching the forecast from a different viewpoint, the consensus forecast will be much more robust than any solo efforts.
Our Operations leads the forecasting work each month by gathering all of the historical data and running the statistical model to output a first pass at the forecast. Then, the Sales team overlays valuable insights about new customers, new distribution, and new promotions, which the statistical model couldn’t have anticipated. Finally, the Finance team marries the bottoms-up forecast with the top-down revenue forecast as a point of reference and validation.
4. Don’t be an incessant adjuster: Major changes only.
There is a fine balance to strike between including valuable business intelligence that the statistical forecast couldn’t have known (such as a new customer or new distribution) and making minor adjustments to the output of the model. And there are a lot of studies that says that the latter almost always makes the forecast worse than doing nothing at all! In those cases, it’s best to take a hands-off approach.
Any time we add a “business intelligence” override to our statistical forecast at DRY Soda, we make a note about why we’re making that change and what data it’s based on. Then, we can revisit these overrides each month to ensure they’re still needed and wouldn’t otherwise be captured by the statistical model. In most cases, we push ourselves to see which overrides we can remove so the model can work its magic.
5. The proof is in the pudding: Measurement is a must!
Measuring and communicating forecast accuracy can be difficult and confusing. Do it anyway. At a minimum, the facilitator of the process needs to understand if the team is getting better at forecasting over time or if there are any biases that are taking over.
We measure forecast accuracy at multiple levels of granularity to provide different insights into the business and our process at DRY Soda. These results are reported out monthly in conjunction with other key operations metrics, like case fill and inventory turns. Together, these three measures tell the story of how we’re performing and where to focus next!
Another important element to forecasting is having a way to look at data to make these optimizing decisions. Our founder, Sharelle Klaus, explains: “We have to bring in sales reports from literally hundreds of different distributors and retailers, so it’s about putting all that together in one cohesive package. We use a lot of Excel spreadsheets to do that because we’re allowed to import, do pivot tables, and all sorts of things to manipulate that data. That’s key because different people need to look at different aspects, so understanding where all those sales are coming from helps our sales and operations teams with the forecasting process.”
I hope you find these tips useful for your own forecasting process. You might even discover that the best benefits of forecasting are the unexpected ones—like seeing your employees get more engaged. At DRY Soda, when our Sales, Marketing, Operations, and Finance teams share information through forecasting, it helps everyone do their jobs better and get a deeper understanding how each part of the work contributes to our overall success.
