If there’s one thing retail brands want to know, it’s what will happen in the future – especially regarding their business. That’s where inventory forecasting comes in.
Inventory forecasting attempts predict:
With these answers in hand, retailers can theoretically prepare for future sales and control the environment their brand operates within. But not every retailer gets their forecasts right.
Luckily, you can. Here’s how.
Inventory forecasting – also known as demand planning – predicts how many units a retail company will need to meet customer demand. These projections are based on historical sales data and market research.
Accurate inventory forecasts offer retailers more inventory control and inform operational decisions within the given forecast period. However, the problem with these forecasts is they are inherently difficult to create.
Here’s why: True customer demand drives forecast accuracy. But knowing the absolute truth about what customers want at any given point is impossible.
Take Brian Berger, CEO of menswear brand Mack Wheldon, for instance. He’s been in retail for decades. But in a 2021 episode of Shopify Masters, he shared how his team still struggles with getting sales forecasting right.
“In a consumer business, you can’t grow; you cannot make it work if you’re not prepared to own the product, to meet the demand,” Brian shared. “It’s particularly stressful because you don’t really have any visibility into demand.”
Forecasts help provide this visibility that most consumer brands lack – even if it’s not always 100% accurate.
The goal for best-in-class retail brands is operational excellence. But establishing operational excellence depends on maintaining optimal inventory levels with the help of precise inventory forecasting. Doing so can minimize stocks, reduce holding costs, and increase customer satisfaction.
|🤿 Dive deeper: A complete guide to inventory optimization.|
Stockouts cost retail brands ~$1T every year. When items go out of stock, not only do retailers miss out on sales opportunities, but it puts customer retention at risk when these folks seek the same products elsewhere.
To compensate for stockouts, most brands order more than they need. Many will even pay extra to expedite their order lead times and restock sooner.
However, all of these endeavors tie up capital. And if not informed by a reliable inventory forecast, they can lead to cash flow problems down the line.
There are, of course, measures to alleviate the challenges that come with stockouts, like selling on backorder.
Brands are in uncharted territory when it comes to predicting what demand will look like next year.
Between a fluctuating market, quick-shifting consumer interest, and evolving DTC trends, anticipating trends is the differentiator between the Nikes and the Pelotons of the world.
Sure, Peloton was on top of the consumer world for part of the pandemic. But it didn’t last. As I write this, the DTC exercise bike brand has less than 7 months to prove it can be a viable company, per The Verge.
Its mistake? Assuming pandemic consumer trends like exercising at home – which worked wildly in their favor – would last forever.
This, of course, was not the case. Gyms eventually reopened, and demand for home exercise equipment dropped.
Meanwhile, Nike is almost 60 years old and dominates the global athletic footwear market with a 39% share. It is also still beating Wall Street Expectations despite consumers’ slowing spending on consumer goods.
Why? Because the athletic giant forecasted for the worst case scenario and planned its inventory accordingly.
Luckily, brands can avoid unnecessary holding costs (and increase their profit margins) by ordering just enough inventory at just the right time. Nothing more, nothing less.
However, you never want zero holding costs (after all, $0 in carrying costs means no inventory in stock). Inventory forecasting can ensure your inventory levels remain in this Goldilock zone, where you reduce holding costs while preventing stockouts.
No retailer wants to waste resources on inventory that won’t ever sell (or don’t sell at a profit). Yes, we’re talking about dead stock.
A healthy business generally has 15% dead stock (or less) in its active inventory. But for direct-to-consumer (DTC) brands, that number typically creeps up toward 33% due to poor inventory forecasting practices leading to overstock.
For instance, most brands that overcompensated for 2020 stockouts are now stuck with excess inventory. The longer these items sit unsold in storage, the more likely they turn into dead stock and the more expensive they become.
Almost 20% of customers quit on brands after 1 bad experience. While this might not affect your revenue today, it will tomorrow. As such, brands that prioritize the customer experience are 60% more profitable than those that don’t.
Which category you fall into (those that exceed customer expectations and those that don’t) depends largely on your CX philosophy.
Old-school retailers seek to quickly fix problems when they arise (like when you accidentally oversell a SKU). Whereas new schoolers aim to prevent those problems from happening in the first place.
Accurate inventory forecasting is just one way modern retailers do that.
That’s because when all your products are adequately stocked, customers can order the products they want, when they want them, and receive them quickly. AKA, you can consistently deliver on your customers’ expectations.
Whether you are learning how to predict inventory needs or are looking for a fresh approach, there are a few inventory forecasting methods that reign supreme:
Quantitative forecasting looks at measurable, historical data (including sales, revenue, financial reports, and digital analytics) to inform predictions. This method then uses statistical modeling and trend analysis to estimate future seasonality.
A quantitative approach is best when predicting demand levels for products with a substantial sales history. That’s because this historical data can often be a key indicator of seasonal patterns.
But, this method can be challenging for new companies with little historical data or those whose seasonal demand is not tied to the traditional calendar year.
Unlike quantitive forecasting, qualitative forecasting relies on information that cannot be measured or quantified.
Instead, qualitative forecasting considers the wider economic climate by pairing historical data with industry expertise (like product life cycles and predictions about future changes to a business’ specific industry, for example).
The strength of this method is that it accounts for a wider breadth of variables. However, it requires extensive resources and relies on lots of future projections. As such, this method is most helpful when you lack historical data (such as with new products).
Trend forecasts anticipate patterns in historical sales trends to predict future demand.
While trend projection is pretty straightforward, you’ll still need to adjust your calculations for seasonality or other anomalies in your sales patterns. For example, you might’ve had a surge in sales because you went viral on TikTok.
As great as this surge was, it’s not something you can bank on happening again — making it an outlier within your data set. By noting these unusual factors in your historical data, you can keep them from influencing your trend projections.
|🤿 Dive deeper: Jones Road Beauty on balancing inventory and TikTok virality.|
Time series is an inventory forecasting technique that relies on time-stamped data to make specific predictions about future trends.
The data is collected at consistent intervals within a designated period (AKA, not randomly) to reveal how demand for your products fluctuates over time.
However, time series analysis will be less accurate if you have a lot of variables within your data (like those caused by the pandemic). So, mark any outlying variables to keep them from affecting future forecasts.
Generally speaking, 7 totally preventable inventory mistakes throw off retailers’ forecast accuracy. Often, these mistakes go unnoticed.
However, when things outside the retailer’s control (like the supply chain) go wrong, their effects are exacerbated, making their consequences unignorable.
“A [forecasting] system is only as good as the data that’s being put in,” Archie Durfee, Associate Director of Supply Chain at Ro, recently said on The Checkout. Most retailers operate under a similar belief.
But while we can probably all agree that not all data is created equal, having a few good data points doesn’t inherently mean your forecasts will be reliable.
That’s because the problem isn’t about data being “good” or “bad.” It’s about retail brands not having all their data in one place.
For instance, you might manage your cash flow in a tool like Settle, production in Anvyl, fulfillment in ShipBob, and returns in Loop. And your ERP might act as a sort of middleware, connecting most of those data sources.
But no ERP is tech stack androgynous. Meaning, with the proliferation of sales channels and new operational tools, brands will need to invest in developing custom integrations to get the rest of their operational data into their ERP. This, of course, is not a viable pursuit for most brands.
As a result, reliable data remains hard to come by.
When you can’t programmatically pull data from a single source of truth, you’re left with data gaps. At that point, it doesn’t matter if you have a few really, really good data points.
Sure, by managing your inventory in Google Sheets or another spreadsheet tool, you can finagle whatever data points you find to fill these gaps. But similar to shoving a round peg in a square hole, the final forecast won’t look right.
Meanwhile, a single source of truth not only ensures you have no data gaps but that you’re using the most reliable data available (rather than using whatever nuggets you can find). As a result, you create a better forecast.
Changing the SKU ID of a product is like renaming your dog halfway through its training: needlessly confusing for everyone involved.
For instance, say you sell Product A using SKU ID “product-a.” But you later change it to SKU ID “product-ab.” When that happens, the data between those 2 SKU IDs disconnect.
From there onward, your inventory forecasting system considers these 2 SKU IDs totally different products (essentially, Product A and Product B).
This, of course, throws off your forecast’s accuracy because the data is split into multiple, isolated datasets.
There are different forecasting methods to synthesize these now-separate data sets. But connecting the 2 streams is extremely difficult and rarely offers the complete picture.
Without this full picture, you can’t hope to create an accurate prediction of future demand. As such, the best practice is to keep that SKU ID the same in the first place.
However, say you’ve already made this mistake (the best of us have). Then, you can link these SKUs the same way you would different product variants to recreate a near-complete data set (more on this in a second).
During periods of fast growth, retail brands will typically throw historicals (past sales and inventory levels) to the wind.
Why? Because when brands see exponential growth (like when ecommerce sales jumped 39% in Q1 2021), your historicals no longer seem relevant.
But even the most mature brands often only consider historical sales when forecasting – not historical inventory levels.
When supply satisfies demand, the consequences of this go unnoticed. But all it takes is one stockout to compromise your inventory forecasts.
How so? You might see no sales for a certain product on your sales sheet.
But without historical inventory levels adding color, this can be misconstrued as a natural decline in customer demand (perhaps a new seasonal trend?). In reality, you didn’t have enough safety stock or you were simply out of stock.
As a result, your forecast will wildly underestimate how much replenishment you need for that period.
Or, the opposite could be true. You could assume you were stocked out when customers simply lost interest in the SKU, leading to overordering.
There are 2 types of products that most DTC businesses sell: evergreen products and limited edition products.
Evergreen products are those that brands sell continuously and have everlasting appeal. Meanwhile, limited edition products sell only for a short period of time.
For instance, Starbucks’ pumpkin spice drinks are only available seasonally, and many Supreme products are only available in limited edition drops. As a result, Starbucks continuously breaks fall sales records, and Supreme regularly sells out in seconds.
But if these products perform so well, why not make them available all the time? Mainly because the limited-ness creates hype around these products and subsequently increases customer demand for these products.
That said, this strategy can also create chaos on the operational side of the business. How so? Limited edition product drops can skew forecasts because only a finite number of units are available to sell.
So, when those units sell out, your demand planner might assume that customer demand has dried up and deem this product a failure. In reality, this product drop was a raging successful, and all available units were purchased.
Still, many forecasting tools won’t leverage this context. So, its restock recommendations will be overly conservative and lead to miss sales opportunities moving forward.
To avoid this pitfall, identify limited edition products as such in your inventory management software. By doing so, you can forecast evergreen products separately from 1-time drops.
No company sells the same exact product forever. You’ll eventually introduce a new version, whether it be a replacement (like when Apple releases a new iPhone) or another offering (like a new color of the iPhone).
When this happens, you might be tempted to forecast this new variant on its own. After all, it is technically a new product with its own demand curve.
But without historical data, these forecasts are wild guesses at best. And they’re likely to leave you overstocked or stocked out.
The better strategy is to link Version 2.0 with its predecessor. That way, you can leverage Version 1.0’s historical data to increase the accuracy of both SKU variants’ forecasts.
After all, neither variant sells in a vacuum. So, by linking these similar products, you can factor in total demand (for all versions of the product) and account for product cannibalization.
Take new iPhone models, for instance. The historical data will show that sales for older versions dry up when a new model is released. But total demand for the iPhone (all versions) remains roughly the same.
So, you can safely assume that you no longer need to stock older models at the same level when the next model is released – just the new model.
Similarly, when Apple introduces the existing iPhone model in a new color, you can infer that the total demand for all variants will remain roughly the same. But the new color will cannibalize some sales for older options.
By linking these variants, Apple can adjust order quantities for each color offering to reflect consumer preferences.
Once each SKU variant has built its own historical data set, you can unlink them and forecast future demand for each variant separately.
However, keeping them linked creates a more in-depth data set for when you introduce the next version of the product. Meaning, better forecasts straight out the gate.
Common sense says that if your inventory supports all your channels, you should be able to forecast for those channels together. But brands that do this rarely get their projections right.
That’s because forecast accuracy is based on actual customer demand – not sales. But when channels are analyzed together, your forecasts need to assume that total sales equate to total demand. This is not always the case.
Say you sold 20,000 units via your direct-to-consumer channel every quarter for the past year. Meanwhile, your wholesale partner has ordered 10,000 units for the past four quarters.
Based on this information, you’ll need 30,000 units to meet next quarter’s demand, right? Maybe not.
Where DTC sales directly reflect consumer demand, wholesale purchases do not. After all, that wholesaler might not be selling through as much inventory as they’re ordering.
This would create a small stockpile of your products at their warehouse, driving up their unit economics. To respond, they’ll likely purchase fewer units or replenish less often.
But your forecast won’t see this coming. So, when this wholesale order doesn’t come through as anticipated, you’ll be the one left overstocked.
The smarter approach? Forecast each channel separately. Then, add each channel’s inventory needs together. That way, you can factor in channel-specific data points (like your wholesale partner’s sell-through rate) to increase your forecast’s accuracy.
Sorry to break it to you, but manual forecasting with Excel (or any comparable tool) doesn’t work. Sure, it offers you a bit more control. But it’s reactive and prone to human errors that could easily be avoided with automation.
For instance, most brands that forecast their inventory needs manually use spreadsheets. But forecasting, by definition, means looking ahead. So, why would you use a tool that’s always looking backward?
Static spreadsheets, by nature, only ever reflect a single moment in time (one that already happened). And to use it, you first have to update it.
When you finally get around to consulting that data, it’s already outdated. So, you’re reacting to whatever demand increase or supply chain issue just happened rather than working ahead to prevent it. and you waste a lot of time updating the spreadsheets to make this happen.
Not to mention that one silly human error (after all, to err is human) can skew the projections exponentially. (Just imagine forgetting a digit.)
Both these shortcomings lead teams to make decisions (like over-ordering) that harm their cash flow and bottom line – despite the data “supporting” their decision.
Cogsy is an inventory forecasting software that supercharges how you meet customer demand. And it’s built to be the source of truth for all Shopify and Amazon brands.
With Cogsy, brands can place orders faster, reduce the frequency of stockouts, and minimize the amount of excess stock hanging out at their warehouse.
That’s because the Cogsy platform provides a godlike view of your stock levels, sales history, restock needs, incoming purchase orders, and upcoming marketing events. All in one place. So you always know how much inventory you have available (what’s just sitting there and what’s approaching its reorder point).
Cogsy then uses that data to build 12-month demand forecasts with pinpoint accuracy. You can even run “what-if” scenarios within the tool to find your best-case, worst-case, and most probable inventory strategies.
Best part? As new information becomes available, your inventory forecasts update automatically, improving your forecast accuracy.
That way, you can make smarter, more informed decisions, respond faster than your competition, and hit your most audacious revenue goals.
Plus, rands that use Cogsy, like Caraway, generate 40% more revenue and save 20+ hours a week on average.
But don’t just take our word for it – try Cogsy free for 14 days.
Inaccurate inventory forecasting can cause a business to overstock the warehouse with products that lack demand. To offload excess stock before it wrecks your profit margins, try the following strategies to increase inventory turnover:
Quantitative forecasting looks at measurable, historical data to inform predictions. Meanwhile, qualitative forecasting relies on information that cannot be measured or quantified. Quantitative forecasting is best when predicting demand levels for products with a substantial sales history. However, you should use qualitative forecasting when you lack historical data (such as with new products).