So, what factors would one consider to forecast demand? Is there a way to improve forecast accuracy?
Very good questions. To recap, my previous post articulated the forecasting challenges faced by a marketer (presumably in Product Line Marketing) who was responsible for party supplies (paper cups, plates, spoons, forks, napkins etc).
Let’s look at this situation more closely starting from the various sales channels and work our way in. The marketing manager is responsible for marketing the product line via retail and wholesale channels. While I am not intimately familiar with the party supply business, but my guess would be that large big-box warehouse clubs such as Costco, BJ’s, Sam’s Club or even national/regional distributors would constitute the wholesaler market. Smaller retail stores such as Party Central, Target, grocery chains (Kroger or SafeWay’s for example), mom-pop stores etc would be the retail channels. I am only considering the US market in this case.
The marketing manager would be intimately familiar with these sales channels and their sales history. The buyers in these channels would have a certain established cadence of placing orders. From a gross perspective, that’s one part of the marketing manager’s demand, what I term as “pull” demand or “channel demand”. This “pull” also depends on a number of factors such as pricing, seasonality, product quality, packaging, customer feedback/preferences, regional markets, sales relationships, competitive pressures, innovation etc. But, let’s assume that this pull is going to be there.
The other factor is what I call the “push” demand. The marketing and sales teams would have their sales target that’s typically keyed down to the quarter. These targets always tend to have an upwards trend. No CEO in his/her right mind will reduce sales targets (unless this is a shrinking market or there is a reason to exit). The marketing manager now has to contend with these factors as well and determine how to reconcile the various elements that would constitute the demand forecast.
Normally, existing channels have the ability to soak up only so much supply. More often than not channels want the flexibility of having primary and secondary suppliers. This allows them to keep a number of things in constant consideration (such as customer preferences, pricing etc) and basically keeps their suppliers on their toes. On the other hand, suppliers have lead channels and secondary channels as well. As sales targets increase, the manager (hopefully in consultation with his sales team) has to determine which channels would have the propensity to take in the push.
“Stores X and Y are my primaries. Can they taken in more this quarter? Or maybe I should work with the sales team to expand into my tier-2 or 3 customers? Or what specific actions will help me drive demand to the levels I expect?“- All of these thoughts are part of the managers marketing strategy for that time period.
Long story short, the marketing manager has some numbers to work with. And, where there are numbers there’s always a pretty graph :-). Below is a made up example.
Most businesses forecast revenues first. That’s because it’s much easier to quantify sales that way and financial reporting is always based on dollars (not units). Revenue forecasts will drive the unit forecasts. The below figures articulate the demand forecasts for the entire product line.
In my experience, there will always be a gap between forecast and customer (“pull/channel”) demand. Forecasts will almost always be greater than “demand”, unless there is a deliberate strategy is to starve the market of a “hot” product in order to keep prices high (as an example).
The below graph articulates a common scenario (or forecast gap) for the “plates” line of products.
“True” market demand is incredibly hard to forecast even for stores. So, then how can forecasting accuracy be improved? While it is impossible to predict the future accurately, what can be done is to use the past to help guide us with predictive analysis. There are some standard factors that like seasonality, events, advertising, marketing campaigns, promotions, placement of products, social media, sales incentives, competitive actions, market research etc. that can (and should) be considered. These factors should be regressed with sales data to determine the various weights of causality, confidence intervals and correlations among other relevant stats. Often, to help with the analysis new factors need to be created to help tease out hidden relationships. Hypothesis testing may also be used among other advanced statistical modeling. My point here is that it is possible to predict with some degree of certainty what future demand is going to be. This is an iterative process that needs to be refined over time until a good working predictive model can be obtained.
The bigger question in this context is – “Does the marketing manager and peers in other departments such as supply chain have the time/band-width to purse this option?” The ROI is certainly there and this methodology is infinitely better than making guesses during each forecasting cycle.
In summary, while demand forecasting is a complex and critical undertaking, the financial impact to the business due to improper forecasting can easily be mitigated to a great extent by creating models, metrics and by increased collaboration between business leaders, sales, marketing and supply chain. In my opinion, product marketing should take a proactive lead in this process. Mistakes will be made and should be used as an opportunity to learn, fix/improve and move-on, rather than an excuse to play the blame-game. In the end, all stakeholders should help make forecasting a really fun activity.
Hmmm..is that truly possible?….. “Forecasting and fun…?”……. What d’ya say?