Finding & Using ‘Accurate’ Forecast Data

The conventional wisdom in the electronics industry is that most, if not all, forecasts — whether of sales, demand, or plant capacity utilization — are often wildly off the mark, somewhat unreliable, and useless to a large extent. Don't believe any of that. Forecasts, even the not so reliable ones, have a major role in supply chain planning, as long as you know how to use them and adjust for shifting market conditions.

Even the executives who say market forecasts are unreliable have been caught citing figures from research firms to support certain strategic actions. There are justifications for this double position and it's important to understand why companies find themselves relying on what they assume could be inaccurate numbers for strategic project planning. Market projections serve a variety of functions and a supply chain that fails to operate on the basis of certain critical assumptions will ultimately fail.

In fact, one of the key elements of a competitive supply chain is the ability to foretell with some degree of accuracy future events related to demand-supply conditions and also subjects such as the direction of the general economy, supplier reliability, natural disasters (yes!), and the weather. Preparing for the unknown has become a fact of life for many supply chain executives where CEOs expect operations to keep on running regardless of the severity of a disaster for instance. In other words, forecasts should not be focused simply on the accuracy of the expected events but on how a company can prepare for events beyond its control.

Those who discredit forecasts focus wrongly on the issue of accuracy whereas what they need to look more closely at is a company's ability to anticipate and prepare for both the expected and the unexpected. How well your company is able to gauge market trends and on this basis stake out a future position will determine its success. Interestingly, the supply chain is central to such activities. Here are some suggested actions with regard to forecast numbers:

  • Partner with reputable organizations:
  • The best research firms understand they won't always get it right but they try to come as close as possible. What they also do best is continuous market monitoring and information filtering as this comes in from partners and other sources. Select your forecast partner carefully and check their track record as well as methodology to assure a reasonable level of dependability.

  • Use multiple sources:
  • It may cost more to diversify your research sources but the gains from such a strategy will compensate for the additional expense. Research firms use different methodologies and databases that might be slightly or widely different from what rivals use. By sourcing data from more than one provider, you increase the likelihood of gaining a fuller picture of events in your end market and give the supply chain managers better tools upon which to base their planning.

  • Research beyond your market:
  • Many companies make the mistake of concentrating research spending only on their market sector and its material needs, neglecting the end users and failing as a result to capture a more complete picture of demand needs. Partner only with companies that look beyond your immediate customers when conducting their research. The electronics industry is notorious for overstating and understating demand due to the failure of company management and their research partners to delve deeper into consumption patterns of end users.

  • Use shorter-term forecasts for project management and longer-term projections for roadmap building:
  • In today's market, very few companies know what demand will really look like beyond 60 to 90 days, except for the aviation industry where deliveries are expected years from order date. Even in aviation, airlines have been known to whittle down orders or aggressively ratchet them up when required, screwing up careful planning on the manufacturers' end. The best-in-class companies carefully match short-term projections with book-to-bill to determine supply chain support requirements while using longer-term projections to set future capacity needs.

  • Learn about the most effective ways to use market data:
  • If your company is spending a lot of money on research, make sure it is also getting value for its investment. It's important to teach employees to be well versed in the most effective ways to use research materials as it can be both a useful and dangerous tool. There are numerous resources on the Web for understanding the best ways to use research data. Start with a provider you feel comfortable with but keep your BS detector on.

8 comments on “Finding & Using ‘Accurate’ Forecast Data

  1. Barbara Jorgensen
    July 13, 2012

    Great blog. Forecasts are simply a tool to help businesses get a feel for trends. Using multiple sources simply makes sense. If each source tracks fairly closely, then it is truly a trend. If not, a closer look is warranted.

    Sometimes I wonder is the tail doesn't wag the dog, though. A decline in PC sales has been forecast ever since smartphones and tablets came out. It is true they are probably related, but it may also be the PC market is saturated. I know I like to hang onto a PC for more than a year at a time. Slower growth is a natural progression. LCD TV is another example: how long are we holding on to our big screen TVs? Our first is more than 5 years old. So I guess the point is forecasts are good, but they are only part of the picture.

  2. dalexander
    July 13, 2012

    @Barbara and Bolaji, Can either of you comment on looking at Raw material supply data with regards to stock pricing as an indicator? Palladium and Tantalum stock values changed with the impact on the Electronics Industry supply and demand fluctuations. Does the commodity market factor in to forecast?

  3. bolaji ojo
    July 13, 2012

    @Barbara, All forecasts should be treated as tentative. Even weather forecasters are conscious nothing is cast in stone. Variations in weather patterns, wind movement and other variables could turn a “solid” forecast on its head. As you noted, forecasts are road markers and in determining demand and supply patterns, there are factors humans cannot compute with 100 percent accuracy — and that is humans.

    Forecasts provide general direction but by continuously evaluating market situations and having a wet finger permanently in the wind, even the weakest data can be strengthened.

  4. bolaji ojo
    July 13, 2012

    I would never use raw materials supply and demand data to determine stock pricing. I covered the commodity markets in the 90s and I can tell you supply and demand have often limited and at best only momentary impact on pricing as well as the stock price of producers.

    If U.S. consumer sentiment data dips, for instance, this could move pricing of metals, crude oil, etc. This may seem natural but don't forget speculators are in the business of shifting resources around to take advantage of perceived weakness or strength. It all comes down to perception, which often can be wrong.

    One more factor to note is related to both nature and geo-political developments. If a severe storm disrupts production of any raw materials or war breaks out in the Congo or a riot (yes) in London, raw material pricing, movement etc., could be negatively impacted. What was a 52-week price target yesterday could turn out to be a dream number today!

  5. Taimoor Zubar
    July 15, 2012

    Bolaji, how important do you think is the role a data analytics system plays in the accuracy of data forecasts? Normally, there are a range of data analytics tools available to companies and they vary greatly in terms of their features and costs. Companies have to choose a tool smartly that can enhance the accuracy of the forecast.

  6. Taimoor Zubar
    July 15, 2012

    “So I guess the point is forecasts are good, but they are only part of the picture.”

    @Barbara: Forecasts are really one way of looking at things and they cannot be completely relied on. Unfortunately, a lot of companies do not take this into account while they are making decisions based on forecasts. Multiple forecasts from different entities is a good option but even that's not a fully reliable way.

  7. bolaji ojo
    July 15, 2012

    TaimoorZ, Analytics are essential in data evaluation. I would say they are critical if a company wants to be sure it would be able to reliably trust the data generated. In using analytics, a company is putting itself on notice for regular data evaluation and simulation. It may sound like distrust but it's more appropriate to regard this as the “trust but verify” approach to data management.

  8. stochastic excursion
    July 16, 2012

    A lot of the forecasts I see are very sparing with statistics and probabilistic breakdowns.  It occurs to me that forecasters could make more use of these tools.  

    On the other hand, unlike actuaries, who have ample population and frequency data, forecasters of business trends have to deal with hidden dynamics and make educated guesses.  Like anything I'm sure getting results in this area requires a lot of experience.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.