Inventory forecasting is key to direct-to-consumer businesses. Here are 7 mistakes retail brands make and how to avoid them.

If there’s one thing retail brands want to know, it’s what will happen in the future of their business. 

A brand’s internal teams wonder things like:

Will customers want our new products?
How many items will sell in a particular month?
How much inventory should we order?
Should we stay lean or should we order safety stock just in case?

As all humans do, teams try to predict the future in the hopes they can prepare and in some way control the environment their business operates within.

To do this, direct-to-consumer (DTC) businesses create forecasts in order to make their operational decisions and create their strategies for a given time period. Accurate forecasting helps teams make crucial decisions with real-time data.

The problem is that forecasts are inherently difficult to create in a foolproof way. 

True customer demand drives forecast accuracy, but knowing the absolute truth about what customers want at any given point is impossible.

Brian Berger, CEO of menswear brand Mack Wheldon, recently shared on the Shopify Masters podcast how his team struggles with this, saying:

"In DTC or the e-comm world, it's particularly stressful because you don't really have any visibility into demand."

Without visibility into consumer demand, how can a brand build an accurate forecast to drive its operations?

There are many reasons why business data gets muddled and distorts a brand’s visibility into what customers want—but there are ways to fix it. 

We’ve studied and observed data from brands that work with us and picked up best practices from around the industry to share with you today.

Below, we’ll discuss the top seven mistakes retail businesses make with their data and how to proactively fix them.

#1: Keep data up-to-date to empower teams.

When using data to create forecasts and do demand planning for future sales, retail brands use two inputs:

  1. historical sales data
  2. inventory data

The timing of that data matters.

By using the most up-to-date information, businesses create forecasts that are more accurate but also more empowering.

When systems that use real-time data, teams are empowered to be more agile and proactive in their decision-making and strategies. Because data is always up-to-date, they don’t need to spend time updating systems before accurate forecasts can be produced. Teams have constant visibility into changes in trends, equipping them to take action when necessary.

Spreadsheets, on the other hand, are always reactive. Teams need to upload all the latest data to a spreadsheet before anything can be analyzed to create a forecast. Using outdated information can cause teams to make decisions that may hurt the brand’s bottom line and cash flow. If sales and inventory information isn’t updated in a timely manner, forecasts can be thrown off considerably.

In order to remedy this, teams must find a way to shorten the time it takes to update their systems with the most recent data. Ideally, brands would find a way to sync their systems so that they are always working with real-time data.

When using real-time data, forecasts become more accurate and teams become more agile in their demand planning.

#2: Never change SKU IDs on products.

Another mistake businesses make is changing the SKU ID on a product. 

For example, if Product A is sold as SKU ID ‘product-a’, but it is later changed to SKU ID ‘product-ab’, the data between those two SKU IDs is disconnected. The forecasting system considers these two SKU IDs as two separate products: essentially, Product A and Product B.

When a product’s data is inadvertently disconnected in this way, the forecast is thrown off and loses accuracy. Historic demand is broken up between two separate datasets that are isolated in the system.

There are different forecasting methods, but it’s always difficult to connect those two data streams to get the full picture and with that create an optimal forecast. Without the full picture on historic demand, there's no way to get even a semi-accurate forecast of future demand.

#3: Inventory levels must be tracked and taken into account.

Most forecasting systems do not take historic inventory levels into consideration. Instead, sales history makes up the majority of what is taken into account to create an accurate forecast. But when inventory levels are not taken into consideration, there are big pitfalls in the accuracy of forecasts that are created.

For example, if we only look at historical sales data and see that sales suddenly drop off, we are likely to interpret that sudden drop in sales as a natural decline in customer demand.

But what if customer demand didn’t drop off? What if customers did want to buy more of that product and the problem was there was no more product to sell?

When historic inventory levels are not taken into account, operations teams cannot see when a drop in sales data matches up with a stockout. The data they see isn’t truly representative of customer demand, skewing forecasts and biasing a team’s decisions. It may lead to poor operational decisions into the future: underestimating customer demand and, as a result, more stockouts.

With inventory data in hand, though, forecasts improve meaningfully over time.

Cogsy, on the other hand, provides that functionality.

In Cogsy, inventory levels of each SKU are tracked continuously and then layered in when building inventory forecasts, increasing accuracy but also improving the likelihood that DTC teams can make optimal operations decisions with those forecasts.

#4: Limited edition product drops must be identified as such.

There are two types of products that most DTC businesses sell: evergreen products and limited edition products.

Evergreen products are those that brands sell in an ongoing manner and which have an everlasting appeal. Limited edition products are sold for a restricted amount of time. Of those limited edition products, there are two types: seasonal and one-time.

One of the most recognized limited edition products is Starbucks’ seasonal pumpkin spice drink that consumers flock to purchase at the beginning of autumn each year. On the other hand, Nike shoes are famous for one-time limited edition product drops. On the other hand, some retail brands use limited editions of a popular product to sell at specific retail stores, like Caraway did with Crate & Barrel.

Brands release limited edition products such as these to create hype and excitement around their products. It works—but it can also create chaos in the operational side of the business.

Limited edition product drops can skew forecasts because of the finite number of units available to be sold. When those units sell out, inventory forecasting software can sometimes present the situation as if customer demand dried up, instead of a successful product launch where all available units were purchased. 

As a result, the inventory forecast may view the product as a failure instead of a success. The forecasting model may make recommendations for the future that won’t serve the business’s best interests.

To avoid the data from being skewed, an improved forecasting technique is to identify limited edition products as such in the inventory management system. By doing so, evergreen products can be forecasted separately than products that drop at different intervals.

This allows the inventory forecast to create an accurate recommendation of an optimal purchase order for the future.

#5: Link demand for all versions of the same product.

When retail brands release new versions of an older product, it can skew forecasts.

For example, if Product A’s Version 1 sold well, the team will want to link that demand to Product A’s Version 2. At the SKU level, they should be linked in order to infuse some of Version 1’s historic demand into Version 2’s initial forecast for Version 2.

Once Version 2 builds up enough data of its own over time, you can unlink the versions and create its own forecast for that SKU.

For a team to do inventory planning and create a forecast going forward, they’d want to know where new versions of products were introduced in the past.

It’s possible to change those transitions from one SKU to another where a team might replace a product either with a newer version or an alternative product altogether. In both cases, it’s inferred that the demand for those products is similar.

By linking the demand for all versions of the same product, forecasts for the future become more accurate. In addition, allowing teams to see where those transitions to newer versions happened and how they influenced sales will help make better operational decisions into the future.

#6: Analyze each channel separately.

Different channels have different demand curves and should be treated separately.

If you compare DTC to wholesale, for example, they have different intervals at which products are ordered and need to be analyzed separately in order to be correctly understood.

When different channels are analyzed together and data is homogeneous, teams might see spikes and fluctuations in sales trends and interpret them as some kind of seasonality, when in reality it may be due to periodic wholesale orders.

To normalize those demand curves, each channel should be identified as such and analyzed separately. DTC orders and wholesale orders should be isolated to not skew the forecasts or the business’s cash flow.

Speaking particularly about wholesale orders, the one metric that positively influences forecast accuracy is the sell-through rate, or how much of the inventory a brand sends a wholesaler actually gets sold.

As mentioned earlier, what drives forecast accuracy is true customer demand.

In terms of working with wholesale, the sell-through rate is what provides a brand with data to determine that true customer demand.

For example, if a brand sells 10,000 units to a wholesaler every quarter, that data isn’t useful for demand forecasting. The data that operations teams need to know is the sell-through rate in order to predict just how much of their products customers want to buy and how they should prepare accordingly.

Data on sell-through rates is hard to come by, mostly because it takes time and effort to relay that information from the wholesaler to the brand. The sell-through rate data could be outdated, incomplete, incorrect, or costly. 

But it’s worth pursuing because of how much it improves forecast accuracy.

#7: Keep human error to the bare minimum.

Most retail brands rely on spreadsheets to manage their inventory and do demand planning, but accurate inventory forecasting is hard to do on a spreadsheet.

Spreadsheets create various problems for operations teams.

First of all, spreadsheets only ever reflect a single moment in time. They are not dynamic. When a team consults a spreadsheet for a forecast, it may already be out-dated.

Secondly, they are prone to human error. One inadvertent flick of a finger can skew data exponentially. Add to that the many team members who may be working on the same spreadsheet and the potential errors multiply.

Spreadsheets (and their errors) don’t empower teams to make proactive decisions.

When possible, teams should rely on real-time synchronized data, essentially eliminating human error, in a product like Cogsy.

With inventory levels and purchase orders streaming in from inventory management software and sales data coming in from an e-commerce provider (like Shopify or even a proprietary website), a brand’s internal teams have a clear view of what is happening in real-time. 

The supply chain and product lead times can then be linked to the sales forecast to give teams the full picture of their inventory turnover and help them get closer to optimization.

Better data, better forecast, better business. 

The more accurate data businesses have on hand, the better the decisions they can make.

By infusing better data into their forecasting process, DTC retail brands will have a better picture of where they stand and where they can expect to be in the future. With that information in hand, they can chart a path to get there.

To do accurate inventory forecasting, remember:

  1. Shorten the time it takes to update data in your systems to empower teams.
  2. Avoid changing SKU IDs for the same product.
  3. Take inventory stock levels into account when doing demand forecasting.
  4. Identify limited edition products to interpret their data accordingly.
  5. Link demand for all versions of the same product.
  6. Analyze each channel separately.
  7. Avoid using spreadsheets to minimize human error.

There’s no way a DTC business can create an accurate forecast with skewed data. Avoid these 7 mistakes to empower retail brand teams to make the best operating decisions possible.

Marcella Chamorro
Head of Marketing at Cogsy · Writer and podcaster on personal growth, marketing and tech