Solving the Mystery of Forecasting and Gauging Demand

You’re ready to start the 4-month production process on a new game, which includes the game itself and two optional accessories. How many of each product do you make overall, for each sales channel, and for each fulfillment center?

This is the mystery of demand forecasting: Making guesses about the future based on limited, imperfect information, risking your resources (or your backers’, in the case of crowdfunding). If you produce far too many units, you don’t gain revenue for thousands of games you’ve already paid to manufacture. If you produce too few units, you can make more, but you can’t provide something customers want in the meantime, and the buzz may have died down significantly by the time the reprint arrives.

I had a great chat last week with Stonemaier Games COO Alex Schmidt, whose responsibilities include the very difficult job of predicting the future. Demand forecasting is a team effort, with input and data provided from a number of sources, and they culminate in Alex presenting his best guess to me for new products and for reprints. We discuss those numbers, reach consensus, and move forward with the agreed-upon quantities.

As the result of such a discussion 7 months ago, we invested in 100,000 units of Wyrmspan, 30,000 rubber playmats (1 per player for a 5-player game), and 20,000 upgrade packs (metal coins + wooden resources). Since Wyrmspan was a new product, we based these numbers on:

• historical data (webstore and distribution) from other games and from Wingspan
• input from distributors who we trust with confidential information and are willing to indicate a rough estimate of how many first-run units they’ll want
• gut instinct (based on how we feel about the game in relation to the current tabletop industry)

At this point you might wonder why we wouldn’t just run a crowdfunding campaign to remove any guesswork from the equation. Crowdfunding provides a baseline for early adopter quantities, but beyond that, demand remains a mystery. Your backer total doesn’t tell you how many other consumers, retailers, and distributors will want the game 6-12 months later.

This is evident through our Expeditions preorder in 2023, for which we initially received around 5,500 consumer orders in the US for the two versions of the game at the beginning of production (still time to adjust the quantity), but over the next few months that amount increased to almost 8,500, along with close to another 40,000 units for distributors and retailers.

As production of Wyrmspan was nearing completion, Joe (Director of Communications) stepped in to decide how many units of each product to send to each of the fulfillment centers to which we outsource (Miniature Market in the US, Spiral Galaxy in Europe, Aetherworks in Australia, Asmodee in Canada). This is an important calculation based on a formula to which Joe makes small adjustments after each product launch. Meanwhile, Alex gets purchase orders from direct-ship distributors: We ship these container-size orders directly from Panda to those distributors so they don’t need to make an unnecessary stop at our warehouses. This is still months before any consumer has purchased the game.

Then it’s time for the product launch. At this point, the numbers are locked in: We have specific webstore quantities at or soon arriving at each fulfillment center, and we’re hoping for those quantities to last us through the 5-day launch period and then for the next 5-6 months if there’s enough demand for a reprint.

For Wyrmspan, our demand forecasting estimates were pretty solid, with one exception: The 10,000 units of the upgrade pack designated for our webstores wasn’t enough for launch demand in the US, Canada, or Australia (we actually had more than enough in Europe, which will impact future iterations of the fulfillment forecasting formula).

Looking at historical data, we sold 3,071 metal coin packs for Libertalia (4,521 launch orders of the game) and we sold 3,270 metal coin packs for Expeditions (7,803 units of the game). Our original guess for Wyrmspan was that we would sell somewhere between 10-15,000 units of the game upon launch–in that way, accessories require guesswork that’s dependent on the forecast for the core product, so relative to that, 10,000 units of the upgrade pack seemed like more than enough for the webstore. In hindsight, that wasn’t the case.

So Alex and I discussed what we can do differently in the future to continue to improve demand forecasting for products like the upgrade pack. We had a few thoughts, several of them aided by input from my partner, Megan, who worked in demand forecasting in a different industry for many years:

• risk: The risk of making a big, expensive, and specific product like the Wingspan Nesting Box is completely different than the risk of making a small, cost-effective, versatile product like the Wyrmspan upgrade pack. We can overestimate for lower-risk products; we can’t do that for high-risk products.
• commitment timing: We lock in direct-distributor orders over 2 months before launching the product on our webstore. This is good for shipping directly to distributors, but it doesn’t leave us much flexibility. In the US, we work with GTS to serve hobby game retailers. It might be helpful in the future if GTS made purchase orders for X direct-ship units and also a buffer estimate for Y additional units that we could use for the webstore if necessary. [UPDATE: Alex is already ahead of me here–he did this for Wyrmspan (the game, not the upgrade pack).]
• backup plan: After Panda approves the English printer files and we begin production, our localization partners start working on translating their versions of the product. This usually takes around 2 months, after which we coordinate the first non-English print run (multiple simultaneous versions help with economies of scale). This is also our chance to add more English units if we’re having second thoughts about the original print run.
• launch vs next 6 months: Traditionally, we just come up with a single quantity for each webstore, but we don’t actually differentiate between the launch quantity and how many units we need for the next 6 months (the amount of time required for a reprint to arrive). We also haven’t traditionally tracked the “next 6 month” webstore quantity, compared to the detailed data I maintain for each product launch. I think separating those two quantities for both forecasting and tracking will be helpful in the future.

Finally, after the launch period and the retail release, we have to decide if we need to reprint anything to meet future demand. At this point we have even more data points to assist the reprint process:

• back-in-stock requests on webstore
• webstore wish lists quantities
• launch data
• historical data
• direct retailer data
• data from distributors about what retailers have purchased and if retailers are reordering after the initial wave

Typically then we invest our resources in a reprint; only once in recent memory have we asked customers to invest their money in a reprint up front (the third printing of the Wingspan Nesting Box). The reason was that the Nesting Box is an entire order by itself, while something like the Wyrmspan upgrade pack will likely join other products in customer orders when it’s back in stock in late summer.

Demand forecasting is as fascinating as it is frustrating. It’s very difficult to estimate the perfect number to serve our customers, and it’s a constant work in process. I appreciate how Alex tackles this problem throughout the year, and I appreciate all the input we get from the Stonemaier Games team and our distributors.

I’d love to hear your insights and questions about demand forecasting in the comments below!

Also read:

• Board Game Supply Chain Basics
• The Great Puzzle of Logistics, Demand, and Supply
• An Open Letter to Distributors and Retailers from Stonemaier Games
• How We Sell Some Inventory Directly to Local Retailers
• Lessons Learned from Quitting Kickstarter (Part 5)
• The Great Nesting Box Debate
• Eager Mechs: Exploring the Expeditions Surprise
• Teaser, Preorder, or Launch: Our Three Announcement Methods

If you gain value from the 100 articles Jamey publishes on this blog each year, please consider championing this content! You can also listen to posts like this in the audio version of the blog.

• Tweet