Forecasting eCommerce demand is one of the biggest challenges in eCommerce. Find out why marketing plays a bigger role in demand forecasting than you think.
What is the number one factor impacting conversion rate on your website?
It isn’t copy. It isn’t load time. It’s the product you’re selling.
That’s why marketers should learn a little bit about how customers interact with products and how the product lifecycle tends to play out. And that’s why merchandisers who were trained in the ways of physical retail need to learn how product and marketing channels interact.
Forecasting eCommerce demand is one of the biggest challenges in eCommerce: how much of this thing can I sell and when will I sell it? Get it right and the big money will follow. Get it wrong and you’ll find a lot of your working capital tied up in widgets.
That’s why good planning is essential. Planning is timing the purchase and delivery of inventory to maximize turnover and free cash flow. Simply: what should I buy and when.
Let’s look at the difference between planning in the TOGU (Traffic Only Goes Up) era vs planning for an eCommerce business today.
Demand Forecasting When Traffic Only Goes Up
For much of recent history, retail meant physical stores, a booming consumer economy, and foot traffic that was always on the rise.
The physical store provided the audience. If the clientele was local, the demographics, tastes and preferences of the audience were fairly predictable because they mirrored the surrounding community. Even if the clientele was tourist-driven, they were probably looking to buy a piece of the local flavor–also a relatively predictable scenario.
That meant that forecasting year on year trends was pretty straightforward, especially for more basic items. The result: predictable inventory turnover and fewer clearance sales.
Demand Forecasting, When No Traffic Is Guaranteed
For the last five or so years, retailers have been 100% accountable for generating their own traffic both in physical stores and online. For more about why and how that happened you can read my Trash Bear piece .
But the net effect is that inventory planning is harder to do predictably. And the less mass awareness a brand has, the harder the planning process becomes.
A young brand with limited awareness essentially builds its customer base up from zero each year until they’re able to develop a meaningful customer base. But those customers don’t come from a geographic area. They probably come from a marketing channel like Facebook.
Complicating matters–in many organizations, marketing and planning are almost completely siloed. Planners, trained in the TOGU era, base this year heavily on last year. And Marketers, also trained in the TOGU era, indiscriminately fire traffic at the website because “a dollar is a dollar”.
The result: less accuracy in forecasting eCommerce demand and much, much more clearance activity.
Stop The Insanity! How Do I Forecast eCommerce Demand?
If you want to know how much inventory to buy, you need two acronyms: TAM and CAM. Note: I covered TAM and CAM in more detail here.
TAM = Total Addressable Market
This is anyone on earth who would possibly buy your product. Your levers are total population, demographics, household income, tastes, preferences & values
CAM = Current Addressable Market
This is anyone on earth who could reasonably buy your product today. Your levers are awareness, distribution, ability to fund both profitably.
You can think of both TAM and CAM as amorphous clouds with a boundary that is shifting daily. You can make reasonable estimates of the size of each cloud, but its true size is impossible to nail down precisely.
If you had an unlimited budget and no concern for profitability you could get your CAM to match your TAM pretty quickly. You could blanket your target markets in advertising, open your own stores to scale distribution, and run aggressive promotions to broaden your audience.
But most businesses are looking to turn a profit. So forecasting eCommerce demand hinges on understanding how much CAM you’ll be able to reach profitably in a given period of time. And that’s really a marketing question. So planners and marketers should work together.
Each product you sell has a lifetime value, in the same way a customer has a lifetime value. A product’s LTV curve will vary based on the nature of the business and how the product is marketed. Here are three common product LTV patterns:
Forecasting For A 10 Year Old Fashion Brand
Example: a brand like Everlane that releases a mix of evergreen replenishment product and fashion product each year.
Everlane has been in business since 2010. They built up a base of loyal customers who subscribe to emails and purchase from the brand regularly. They also received enough press coverage to become part of the cultural zeitgeist, at least for certain segments of the population. So there is also an audience who checks in to see what Everlane is up to, even if they haven’t made a purchase in a while.
Everlane can generate some decent demand by selling new products to this existing audience. But this isn’t a sustainable strategy because that retained audience pool isn’t stable–people are dropping out all the time.
Everlane sells a complex, multi-SKU and multi-category assortment. With each new product launch they have a choice: is this a star or a supporting character? “Star” products are marketed heavily and are more likely to be new customer acquisition drivers. Supporting characters help build AOV–they’re products you add to cart on impulse after the “Star” drives you to the website.
For a brand like this, the launch year is typically the peak demand year. If the product is brought back for multiple years, total demand will decline each year.
The exception to this rule is a “sleeper hit”, where a product goes viral after its first year on the market. Maybe the product is discovered by a Tik Tok influencer and gets exposed to an audience beyond the brand’s core customers.
Moments like these are hard to engineer, so they’re almost impossible to plan for. But after the trend-driven bump, demand will start to fade off again.
Forecasting For A 4 Year Old Cosmetics Brand (Replenishable)
Example: a brand like Three Ships, selling a broad-ish assortment of beauty products and releasing new SKUs each year.
Three Ships is a younger brand, so their roster of loyal customers will be smaller than Everlane’s. They don’t yet have the awareness to drive a high volume of direct visits from prospects…yet.
An interesting twist is that Three Ships’ products have a built-in replenishment cycle. If someone tries a skincare product and likes the results, they’ll buy it again when they run out. So part of forecasting demand is understanding how many customers that try a product for the first time go on to purchase it regularly.
Forecasting demand for a beauty product is closer to forecasting sales for a subscription service. Some percentage of the customers you acquire will churn, but there will be a stable base.
Skincare is also a relatively evergreen category compared to apparel so there are multiple opportunities to drive new customer acquisition post-launch. If a product has good traction at launch and there are new marketing pushes each year, you can actually expect demand to increase slowly but steadily over time.
Forecasting For A 3 Year Old Consumer Product Brand (Non-Replenishable)
Example: a brand like Our Place that launched with a signature product like the Always Pan that is use case specific and does not need to be replaced frequently.
Our Place is a relatively young brand bringing a unique angle to a sleepy category: what if, instead of buying an eight piece cookware set, you could buy one pan that covers all eight functions?
The cookware category kind of has a replenishment cycle, but it’s a long one. Pans do wear out, but many people buy a set of pans for their first apartment and let them live on for far too long. So there are really two sub-audiences in this market: cooking enthusiasts (aka people who would replace a pan when it wears out) and everyone else who needs pans.
Demand for this pan will probably surge for 2-3 years as the brand’s marketing efforts reach an increasing percentage of the enthusiast market. Enthusiasts may even buy multiple pans. There will also be some supplementary demand from casual pan buyers during this time.
But when the majority of the enthusiast market has either purchased the pan or considered and rejected it, demand will be harder to come by. So Our Place will need to release new products to appeal to the enthusiast market, while continuing to produce the pan in smaller quantities to meet demand for casual buyers.
Why Demand Forecasting Matters for eCommerce Marketers
In the post-TOGU world, marketing and planning need to work together to forecast eCommerce demand. In the real world, these functions are often siloed.
This means that, as a marketer, you will often walk into scenarios where you’re being asked to fight against massive headwinds caused by the nature of the demand curve for a given product.
Imagine if you were hired as a growth marketer for Our Place in year 4. The brand wants to accelerate growth, despite the fact that the cooking enthusiast market is tapped out. Customer acquisition costs are rising, and they fired the last growth marketer because that person was unable to reverse the trend.
You’re essentially being asked to bend the laws of space and time to “make it work”. You think you can do it, because you did it at your last job…a brand at a different stage in the demand curve. You step into the role and nothing in your old bag of tricks works. Your boss is angry. You’re getting emails about Facebook Ads at 3am. No bueno.
This is one of the big reasons that the CMO is the executive role with the highest rate of turnover. If you want to have a successful career as a senior marketer or eCommerce leader, you need to understand the whole pie, not just the slices that are directly under your control. And that’s why you need modern lifecycle marketing.