Since I've been the CFO at Sodtt Ceramics the last two-plus years, I have worked to develop a thorough sales and operations planning process that monthly works through plans for sales, production capacities, inventory, lead times, investments, finances, etc. Sodtt makes insulators and substrates for the electronics industry as well as orifices and nozzles for industrial uses. With such a range of customers, it's important that our planning works as planned. Occasionally, everything hums along just fine from month to month, but inevitably we are blindsided.
I believe good organizational planning really starts and ends with our customers and sales forecasting, and too often this data turns out to be very unreliable. The director of sales and I have tried to install some standardization to the sales function, from the way salespeople define "a sale" to the manner by which they develop their individual forecasts. But most of the sales staff are a bunch of cowboys, shooting from the hip and grabbing their commissions (forecasts be damned). And despite some investments in demand planning and supply-chain planning tools, as a company we still rely heavily on a variety of manual techniques and legacy spreadsheets to pull our forecasts together and share. And these homegrown tools make it difficult to update the many day-to-day changes that occur.
I believe the sales forecasting component of our planning process has turned into Sodtt's Achilles heel. With the rapid ups and downs of our markets today, my operations and supply chain is constantly being whipped back and forth: either piling up unnecessary inventories (ours and suppliers'), overtime, and expediting costs to hit sales targets that eventually don't materialize, or straining to satisfy unexpected orders and racking up every conceivable quality and delivery error in the rush. I don't want a crystal ball, but I do need to reorganize our sales forecasting before it damages the company. Where to begin?
* The Challenge incorporates hypothetical persons, companies, and products and does not portray the actions of any actual persons, companies, or products.
By Stephen Brown
The situation with the sales forecast at Sodtt Ceramics is difficult but not uncommon for companies operating in a make-to-stock environment. Your company requires a fundamental rethinking of its forecasting approach. The good news is that the investments required to fix forecasting at Sodtt will deliver significant financial and customer service benefits.
Finding the Right Place
"Where to begin?" was your question: I would start with redefining who owns the forecast. Placing the responsibility for forecasting in the right functional area is a key. In most instances, sales forecasting belongs in Supply Chain. With the responsibility to manage customer service, inventory, and costs, the Supply Chain function is the most objective of the potential functions owning this process. If forecasting is owned by Sales, as is the case with Sodtt, the temptation exists to "game the system," which means forecasts are produced that maximize compensation for Sales often at the cost of increased inventory. Sales needs to participate in the forecasting, but they should not be the primary owners.
While Supply Chain is the right function to own it that does not mean that the supply chain team at Sodtt has the right experience to do forecasting well. This is a professional discipline that requires the right academic background and experience to be effective. Sodtt likely needs to augment its forecasting team with experienced resources from outside the company if it is to reach the forecast accuracy levels expected.
Mixing Art and Science
It seems evident that the current forecast process at Sodtt is lacking in definition and adherence. A well-structured "day-in-the-life/week-in-the-life" workflow that encompasses all the steps in the forecast process is required. The process needs to be defined with enough detail that participants are not making it up as they go along but also not so restrictive that it drives mindless adherence to a set of activities that are repeated mainly according to a timetable. Forecasting is a blend of art and science. Knowing what this means and how to account for it within the process can determine success or failure.
The data hierarchy along product, customer, and geography dimensions is the structure that allows the forecast to be rolled up or disaggregated to meet various needs within the company. The hierarchy should be defined with awareness of other data hierarchies that exist within Sodtt. Doing so increases the integration of data across systems and allows data and reports to be compared across functions in a like-for-like manner.
The choice of historical data series to use to generate the statistical forecast is an important consideration. Most companies choose between shipments or customer orders as their historical data with pros and cons associated with each. It is becoming increasingly common for companies to incorporate point-of-sale data into the process to add increased visibility and intelligence. The choice of forecast horizon and time buckets also need to be made with consideration to what drives the greatest business value for Sodtt. A longer time horizon can add strategic value if the forecast quality is maintained in the outer periods, but this adds complexity and cost to the information technology components. A weekly time bucket may be the most effective one for Sodtt given the volatility in your business, but this requires a business process that operates on a weekly frequency vs. a monthly one.
Leveraging the Technology
Sodtt has made "some" investments in demand planning and supply chain planning tools, but it is clear that these investments are not optimal and the forecast is produced via muscle and heavy lifting rather than automation and efficiency. Sodtt should conduct a review of the current tools to determine whether they are sufficient to support a new forecasting process. In many client instances we find that the tools are fine but they are not configured in the optimal way and the users often do not know how to fully leverage the functionality available. Sodtt needs to get back to basics, establish a reliable forecast process and then follow the "crawl-walk-run" approach to adding sophistication.
Collaboration and Visibility
Although Supply Chain should own the forecast, it should not be generating it in isolation. Input from Sales, Marketing, R&D and other functions adds vital intelligence that is often not available in the historical data. Sodtt has the forum for this input via the sales and operations planning process.
Customer collaboration provides another important input to drive higher forecast accuracy and should be incorporated into the re-engineered processes at Sodtt. Sharing the sales forecast with key suppliers and supply chain partners is a smart way to drive visibility and efficiency throughout the end-to-end supply chain.
Measuring Performance
The quantitative nature of forecasting allows for timely and accurate measurement of performance. Sodtt should measure forecast accuracy at the item-national level four weeks in advance of production for the purposes of benchmarking with other companies and industries. This is not the only accuracy metric that Sodtt should capture, but this is likely the best place to start. Measuring the forecast bias and stability at various lags is also important to identify root causes of inaccuracy.
Achieving Success
Sodtt has a significant challenge ahead, and should be planning at least a one-year roadmap to sustained improvements. By tackling it with determination and assigning the best resources, the company can significantly improve its forecasting performance. The financial and customer service benefits of improved forecasting typically dwarf the costs of delivering them, and as CFO, you should take heart that the same opportunity is there for the taking at Sodtt.
Stephen Brown is a Partner with
Deloitte's (www.deloitte.com)
supply chain consulting practice, based in Toronto. For the past 15 years he
has been
assisting Fortune 500 clients with improvements to their supply chain
in the areas of planning, metrics, logistics, organization structure, and
technology. Information on Deloitte's wide range of services and industry
practices can be found at www.deloitte.com.
Stephen can be contacted at
stephenbrown@deloitte.ca
By Bob May and Randy Littleson
Forecast accuracy is truly one of the great challenges companies in volatile, dynamic industries face. This is made even harder in times of great economic uncertainty. Since most forecasts begin with a review of past performance, you'll quickly realize that the past is not an accurate indicator of the present or future. This is true in the midst of both extreme upside and downside economic periods.
To start with, Sodtt needs to acknowledge that you'll never get a completely accurate forecast. This is important, because it acknowledges reality and ensures that you make investments in two critical areas: ensuring the most accurate forecast possible and ensuring you are equipped to deal with forecast inaccuracies and their impact. Far too many companies fail to acknowledge the latter, and so they end up investing all of their energies solely in trying to achieve the perfect forecast.
One of the first things that Sodtt should do is establish processes around collaborative forecasting techniques to improve the accuracy of forecasts. At a minimum, Sodtt should be collecting sales forecast data on a monthly basis and then consulting with marketing and product management regarding new product introductions, product end-of-lifes, promotions, discounting, competitor activity, and industry trends. One of the real challenges related to this activity is trying to minimize the time collecting and rationalizing the data, which consumes time better spent on analysis that adds real value. This usually can be solved by choosing a suitable, scalable tool that all users will accept (hint: spreadsheets won't cut it).
Generally, collaborating with customers on forecasting is something that is considered to be difficult to pull off. Thus, I would recommend efforts in this area be focused on your small set of ‘A' customers. You also need to be sure you have a forecasting model you believe in and that has shown to work because you'll often be pitching this process to a customer that is still in the dark ages of forecasting. I would recommend a gradual approach, starting with a few customers and allowing for iteration and improvements along the way.
It's also important to measure forecast accuracy properly. Some companies use a model that looks at a weighted composite of bias (actuals consistently higher or lower than forecast); error (the size of the delta between actuals and forecast); and stability (how much is the forecast changing over time), which should be applied separately to each of the collaborative inputs.
There may be a temptation to forecast more frequently. While in theory this makes sense, it is typically difficult to pull off. Forecasting is inherently a process of gathering multiple points of information and finding patterns that, when compelling enough, cause a change to occur. Rather than trying to forecast more frequently, establish processes to ensure that you are effectively communicating changes to the forecast to all affected parties as soon as they occur and only when they occur rather than forecasting more often.
In parallel to your investments to improve the forecasting process should be efforts to improve demand management and your ability to respond to forecast change and error. As noted above, this is reality, so you need processes to deal with reality. Some fundamentals to achieve this are:
- Integrated demand-supply visibility: At the heart of being able to manage demand changes and respond to errors is the ability to see, side-by-side, the interrelationship between demand and supply. You need all of your key stakeholders to be able to immediately see that a change in demand has a corresponding impact on supply and vice versa. Too many companies have created silo organizations that separately own demand and supply and, thus, are supported by disparate tools that fail to provide this integrated view.
- Rapid scenario creation and evaluation: Your front-line decision makers are going to need to be able to quickly create myriad scenarios (e.g., what would happen if demand were up by 10%, what would happen if we accepted this new order from this customer, etc.) and evaluate both their impact on your operations and your ability to execute to them. These scenarios must be quickly created, shared, evaluated, and scored so that everyone involved — both the demand and supply sides of Sodtt — can collaborate on the right course corrections to make in light of an unplanned event.
- Collaboration from customer to supplier: We've already spoken about customer collaboration in regards to forecasting. There's also the need to collaborate with suppliers. You need systems that support collaboration all the way from the customer through Sodtt and out to your end suppliers. This is critical since Sodtt is in the position of orchestrating response to change and needs to align the entire supply chain to deal with these circumstances.
Forecast accuracy is a challenging problem for many manufacturers and, as Sodtt has learned, the consequences of not adequately dealing with poor forecast accuracy can be pervasive and severe. For Sodtt to get ahead of this problem, you need to develop strategies both to improve forecast accuracy and to deal with the reality that your forecast will never be perfect.
Bob May is a senior product manager at
Kinaxis (www.kinaxis.com),
and is responsible for the demand management functionality, installations and
upgrades in the RapidResponse product. In his 20 years with Kinaxis, Bob
has fulfilled
roles in integration, design, services, support, and most
recently product management. Prior to Kinaxis Bob was an independent
information technology consultant. Bob can be reached at
bmay@kinaxis.com
Randy Littleson is vice president of
marketing with Kinaxis (www.kinaxis.com)the
provider of an on-demand service that empowers multi-enterprise manufacturers
with the integrated demand-supply planning, monitoring, and collaborative
response capabilities required in today's complex and dynamic world. Randy can
be reached at rlittleson@kinaxis.com
Step on your soapbox! Here's your chance to debate and discuss this month's
Challenge and past Challenges with other readers. Check out the
IW Manufacturing Business Challenge thread
in the IndustryWeek Forums. Add your comments, or read what others have to say.
Click here to view now
Tell us how your company has developed more accurate forecasts. If it meets IW's publishing criteria, it will be posted in
the Archives section of this website. E-mail your solution to
info@iwchallenge.com. If you wish to be
identified with the Solution, please include your name, title, company and a
telephone number where you can be reached. IWMBC selects Reader Solutions from
those submitted and reserves the right to edit submissions for length and
clarity.
Tell us about a Challenge your company is currently facing. If it passes
editorial muster, the IW Manufacturing Business Challenge experts will offer
solutions in an upcoming e-newsletter. Send your Challenge to
challenges@iwchallenge.com. If you wish to
be identified with the Challenge, please include your name, title, company and
a telephone number where you can be reached.
www.industryweek.com
www.kinaxis.com
www.deloitte.com
To unsubscribe to the IW Manufacturing Business Challenge e-newsletter, send a blank e-mail to unsubscribe@iwchallenge.com.
