Get to know Mike Neal, CEO and Co-Founder, DecisionNext
What’s the Biggest Decision-Making Challenge Facing Commodity Industries?
Without a doubt, market volatility. And over the past decade, the volatility has only gotten greater. In the face of this uncertainty, commodity companies are routinely making multimillion-dollar decisions based on expected future prices with what amounts to gut feel and educated guesses.
We formed DecisionNext to address this problem. To help companies make sound, unbiased, empirically- based business decisions in volatile environments.
How Does DecisionNext Improve Commodity Pricing Decisions?
By implementing a rigorous, analytics-driven forecasting process, companies get not only more accurate forecasts through best of breed approaches, but also a repeatable process around the forecast that strips out as much human bias as possible.
It’s worth noting that Gartner redefined the analytics space a few years ago when they said that beyond predictive analytics is prescriptive analytics. And prescriptive analytics is what we do. We don’t just use analytics to predict the future. We use it to prescribe answers, or decisions, based on what the future holds.
“We use analytics to prescribe answers based on what the future holds.”
Of course, there is a huge data component to what we do. But what sets us apart is that when we comb through data we are not searching for questions or relationships we hadn’t thought about. We already know the questions. It’s pretty straightforward: What will price be next year? What will supply be? What product mix should I produce? How far forward bought or sold should I get?
We use analytics to give our customers the answers — the most accurate answers so they can make highly profitable decisions.
Can You Offer an Example of Improving a Company’s Pricing Decisions?
I’ll give you an example in the natural resources industry. Let’s say there’s a cargo ship loaded with iron ore leaving Australia heading for China. It takes 10 days to get to China.
The price of iron ore ten days from now will almost certainly not be the same as it is today given the tremendous volatility we’ve seen.
But because the natural resources industry has very little technology around forecasting iron ore prices, they usually pick a fixed time interval in advance of delivery to set a price – say, five weeks. A normal practice is to consistently price these loads using the reported prevailing price at the exact same time interval ahead of delivery.
Over time when you do that, you end up achieving, by definition, the average price. Sometimes it’s higher than the price at delivery. Sometimes it’s lower.
But what if you had a more accurate price forecast? What if you didn’t have to achieve the industry average price? With our forecasting solutions you could flex the time interval to your advantage to price the iron ore more profitably. What the model is telling you there’s a high probability prices will be higher than the consensus forecast, you shorten the window to the extent you can, and vice versa.
That’s just one example among many of how we help a company optimize its decisions, in this case what price to charge and how far forward to charge. But our solutions optimize a whole range of decisions whether it’s purchasing, logistics, production or sales. There is a big opportunity available to those who choose to invest in analytics.
Compare Current Forecasting Techniques With Your Analytics-Driven Approach.
Today the prevailing technique for best in class forecasting in mining is cost curve intersect analysis.
If we compiled a list of iron ore producers in the world and sorted the list by how much it cost to produce and deliver a metric tonne of iron ore, you would get numbers anywhere from as low $20 a tonne all the way up to well over $100 a tonne.
“Cost curve intersect analysis is a very logical and well understood approach…… But there’s one big problem with this approach. It just doesn’t work.”
Let’s say I want to price a load of iron ore today for delivery a year from now. One way to calculate a reasonable price per tonne is to use this “cost curve intersect” approach. That means we would estimate how much iron ore the world will be consuming a year from now. I would then go down my list of producers from the lowest marginal cost per tonne to produce, moving to those with greater and greater costs to produce. I would eventually get to the one producer where the total cumulative tonnes equals world demand. Whatever that producer’s cost to deliver is will be the cost that determines the prevailing price for the industry.
It’s a very logical and well understood approach. It’s where the supply curve and demand curve intersect. But there’s one big problem with this approach. It just doesn’t work.
It doesn’t reflect the real world. There are too many factors that affect the prevailing price beyond just the marginal cost to produce for the “last” company on that list – the last company who enables us to meet world demand.
“Adopting analytics-driven forecasting isn’t just a matter or profitability. It’s a matter of survival considering how companies are making decisions in the 21st century.”
It makes more sense to use a rigorous, analytically-driven, forecasting approach that is inherently more accurate in predicting prices.
When you think about it, adopting this approach isn’t just a matter of profitability for commodity companies. It’s a matter of survival considering that this is how more and more industries are making decisions in the 21st century.
Is the DecisionNext Forecasting Approach Purely About Analytics?
No. In fact the key to our price and supply forecasting approach is that it blends the best of humans and technology.
Both humans and technology bring something unique and important to the forecasting problem. But both bring problems too. The trick is to bring them together efficiently and to capture the best of each.
“By combining the best of humans and technology we deliver more accurate results than either could ever achieve on their own.”
While an econometric model can be built that reliably finds an unbiased point estimate of price 6 months from now, it cannot take into account the news story that broke this morning, completely changing market sentiment and momentum.
Of course, the model would likely catch up tomorrow, once an additional day’s actual prices are ingested by the system. But it’s much better to give an expert user the ability to tweak model inputs – which is what our approach allows. By combining human with technology we deliver more accurate results than either could ever achieve on their own.
What Kind of Experience Does Your Team Bring to the Table?
For over 15 years our experts have worked across multiple industries on the specific problems we’re solving today which has allowed us to refine our analytics-driven approach significantly.
“It has taken years of learning on our part to get this right. And the results our clients are achieving in terms of revenue and margin increases, prove that we have.”
What we do is about much more than formulas. It’s about building formulas into processes. It’s about supporting those processes with bullet-proof software. It’s about working with people and teaching them to use the software to achieve their business benefits.
It has taken multiple skill sets and years of learning on our part to get this right. And the results our clients are achieving in terms of increases in revenue and margin prove that we have.
What Does the Future Look Like for Analytics-Driven Decision Making?
The future is here today. There are at least a dozen industries today where advanced highly focused analytics have changed the game. It started on the revenue side of businesses in the early 1980’s with airlines. Today you cannot be competitive in the airline industry without analytics-driven decision -making, referred to in that industry as “revenue management”.
Then we saw analytics take hold with hotels and financial services. And then in supply chain, then retail. In fact, it was in retail that one of my earlier analytics companies, DemandTec, made a difference. I co-founded DemandTec with Hau Lee, a Stanford Professor and it is now part of IBM. Interestingly Bob Pierce, my co-founder at DecisionNext, was a scientist for Khimetrics, our direct competitor at DemandTec.
“With the proven results of analytics-driven decisions, there is no excuse today for commodity companies to leave money on the table by using gut feel in today’s volatile markets”
Looking ahead, I believe given the enormous dollar amounts at stake, and the current lack of analytics in commodities compared to other industries, adoption is just a matter of time. Commodities players will begin supporting their teams with rigorous models, and just like airlines, financial services and retail, we will see this become table stakes – you won’t be able to compete effectively without it.
With the proven results of analytics-driven decisions, there is no excuse today for commodity companies to leave money on the table by using gut feel in today’s volatile market.
For more information contact: