What is decision analysis?
Decision analysis (DA) is the formalized, practical application of decision theory in real-world business scenarios. It helps decision-makers and business leaders make optimal decisions when presented with different choices. In fact, most people use decision analysis methods on a daily basis for relatively straightforward decision making without realizing it.
For example, whenever a manager has to assemble a team for a new project or a vendor presents a procurement officer with various offers, informal decision analysis usually determines the optimal solution. Because these decisions are relatively straightforward, a more formal analysis may be unnecessary to arrive at the “right” decision.
But when the decision in question involves complex—and sometimes competing—multivariate choices, analysts employ systematic and quantitative decision analysis models and techniques to arrive at the most optimal solution (ideally the one with the lowest potential risk and the highest potential upside).
History of decision theory and decision analysis
According to Harvard Business Review, “decision making” was not even a common phrase until the middle of the 19th century. Before that, “policymaking” was the de facto mode of thought for businesses. But because policies can be debated endlessly without results and decisions are final, “decision making” became the new modus operandi for business people.
While this might sound like a silly distinction, it was really innovative at the time. After all, figuring out how humans make optimal decisions in a quantifiable way was (and is) no small feat. The study of decision making has led to groundbreaking research that completely revolutionized the way we do business, including decision theory and game theory.
The term “decision analysis” was first used by a professor of Management Science and Engineering at Stanford University in 1964. In the 56 years since, millions of decision trees, matrices, and SWOT analyses (among many other decision models) have helped countless businesses make the best decisions they can in an uncertain world.
Much more recently, with advances in computing and artificial intelligence, decision analysis is faster than ever. What once took a team of analysts months to do with pen and paper can now be automated by software programs in a single day.
Simply put, it’s never been easier for an enterprise-level company to easily and affordably analyze every possible outcome of a decision before making it.
What are decision models?
Just as there are countless day-to-day decisions to be made in business, there are numerous potential decision models that can be used by analysts depending on the situation in question.
Simply put, a decision model is a visualization of the various choices, risks, and payoffs for any given business decision. Some decision models have become well-known bywords in the world of business (think SWOT analysis) and are fixtures in MBA courses worldwide.
The most tried-and-true decision model that we’ve all used at some point in our lives is the two-column list of pros and cons we were taught as kids. Obviously, while that decision model might work as a visual aid for mundane decision making, it’s inadequate for complex, multivariate decision analysis.
The most commonly used and popular decision models in business include:
● decision trees
● decision matrices
● influence diagrams
● SWOT analyses
Each of these decision models is useful in its own way. Some decision models are more appropriate for certain jobs than others. Selecting the appropriate decision model is mission-critical for optimal decision analysis.
For example, decision trees are often the easiest to visually interpret and understand for all stakeholders. Decision matrices are “familiar” because they adopt a standard row-and-column format. Influence diagrams are if → then visualizations that are easier to customize (although they can become unwieldy). Finally, SWOT analyses are a well-worn, “textbook” approach to decision analysis.
How to “do” decision analysis
No matter which decision model you choose, the goal of every decision analyst is to identify all possible uncertainties, risks, and payoffs that result from the choices on the table, and to assign a probability to each outcome. In the case of competing payoffs, the utility (or usefulness) of each payoff is measured against the others.
Granted, there is no one-size-fits-all approach to decision analysis. But generally speaking, the same step-by-step process is often followed:
- Create/decide on a decision model for the decision in question.
- Identify all possible choices and alternatives.
- Determine the likely future outcomes of each choice.
- Ascribe quantifiable, numerical values for each outcome (i.e., the risks and payoffs). This is usually expressed as an actual or assumed monetary value.
- Determine the uncertainty of each outcome (i.e., the probability of that outcome occurring).
- Analyze every possible combination of choices, alternatives, and outcomes. Each set of choices should eventually terminate into a set of results that can be simplified into a single monetary value when weighted for uncertainty.
- Make a decision based on the results of the analysis.
Critics of decision analysis point out that when decision-makers are presented with too many options, “analysis paralysis” can stall any viable decision making. Today, this objection is largely irrelevant. Cutting-edge decision analysis solutions have all but done away with analysis paralysis by automating most if not all of the heavy mathematical lifting.
The biggest risk with decision modeling in the 21st century is human error. No matter how sophisticated your models are, they’re only as good as your assumptions and inputs (e.g., whether you’ve fully accounted for all possible outcomes and whether your probabilities are reasonable, among other factors).
Why is decision analysis important for modern businesses?
If you don’t think through the consequences of your actions, you’re likely to make bad decisions. Just about anyone who has ever had a job would agree with this simple, basic truth of risk management.
Yet despite universal agreement that risk management is essential to any successful business, not all organizations give decision analysis the same weight.
As an example, most enterprises are willing to invest in human talent (i.e., they’re willing to pay for the best analysts they can hire) but are not as willing to invest in technology (e.g., those analysts are still making decision analysis models using Excel and PowerPoint).
How decision analysis and game theory revolutionized Wall Street
This disconnect was perhaps most glaringly obvious on Wall Street in the 1980s and 1990s. Investing and trading are mathematically demanding disciplines. Yet many of the famous “Market Wizards” were known to make multi-million- and even multibillion-dollar trading decisions mostly by gut feel.
You don’t have to be a rocket scientist to realize that “gut feel” probably doesn’t lead to optimal decision making. While some humans are certainly more intuitive than others, which can lead to much better decision making, this type of intuition is not really quantifiable or measurable.
Naturally, as the years passed, some of the market wizards were never heard from again.
Meanwhile, other asset managers like James Simons (who started out as a math professor) were leveraging early computers to pioneer the fields of technical analysis and algorithmic/ high-frequency trading by mathematically modeling the markets—something that had never been previously attempted.
They realized that the best way to make money from the markets was to model them as accurately as possible. They had to be able to ask their model questions and get realistic answers before entering any trades. Only by doing so could they attempt to quantifiably analyze every possible outcome of a trading decision.
The rest is history. Today, high-frequency trading (HFT) is the norm for competitive investment banks and hedge funds around the world. And the little hedge fund that Jim Rogers started, Renaissance Technologies, has the best trading record in history: since 1988, their flagship Medallion fund has generated an average annual return of 66% before fees.
Those returns are literally inhuman. As far as outsiders can tell, the highly secretive Medallion fund is run entirely by computers with some of the most refined decision analysis algorithms in the world. It’s over $100 billion of 100% automated, high-frequency trading returns.
The problem with conducting decision analysis in Excel
Granted, Renaissance Technologies is an extreme outlier that exists on the far end of the decision analysis spectrum. Their team had over 50 years to refine and perfect their algorithms, models, and processes. Their success isn’t so much a goal as it is a case study of what is possible with optimized decision making applied at scale on a constant, daily basis.
Fast-forward to today, when most enterprise-level organizations are still using Excel to conduct high-level, complex decision analysis (often with millions and even billions of dollars on the line).
What’s going on here? Why is such a high-stakes, high-ROI process still conducted with outdated, cumbersome technology?
Part of the issue is that even with significant advances in computing power, modeling complex, multivariate business decisions remains difficult. There is no simple, one-size-fits-all solution for every business out there. DIY solutions like Excel, while cumbersome, will get the job done for analysts with enough patience.
But the problem with conducting decision analysis in Excel is that any model simulated in Excel is static and unchanging. In other words, you probably need to make a different Excel model for every important question you need to answer. Let’s not even get into the security and version control issues that might lead to corrupted or incomplete spreadsheets.
At the enterprise level, this is not really a scalable solution. For real-world, multimillion-dollar decision making, the best solution is a queryable data-sf-ec-immutable="" model that’s flexible enough for analysts to get all the answers they need in one place.
With the right setup, a queryable model can save you thousands of hours of manual modeling in Excel. More importantly, it can give you the answers you need much faster, too.
Real-world decision analysis example
As a small, rural telecommunications company, Hart Telephone Company (HTC) had a big problem: figuring out how to provide innovative service offerings with fewer resources compared to their urban competitors.
Further complicating things, the telecom industry faces the unique challenge of having to answer to the Federal Communications Commission (FCC) and the National Exchange Carrier Association (NECA). Telecom companies have to accurately anticipate how much access they will require from the FCC’s access charge plan and pay out settlement costs accordingly.
As you can imagine, on top of all the typical financial considerations that go into running a business, having to calculate what could be hefty FCC settlement costs can turn into a balancing act for telecom companies with thin margins.
In fact, it costs more to service a rural area than it does an urban area. As a result, HTC often has to apply for grants just to pay FCC settlement costs. But in order to receive those grants, HTC has to prove that it could profitably extend broadband access to new regions.
Needless to say, this isn’t the type of modeling activity that you want to leave to analysts using Excel spreadsheets.
That’s why HTC chose to go with Synario’s financial modeling and scenario analysis solution. After setting up Synario’s personalized queryable models, HTC saw drastic improvements in financial forecasting (especially when it comes to rolling cash flow statements and financing evaluations), as well as more stakeholder agreement on future capital allocations.
Prior to using Synario, HTC’s CFO, Melissa Green, had to manually pull data from over 1,350 individual general ledger accounts to create her yearly forecast. After implementing Synario, Green sums it up by saying the following:
“I have tried to find another U.S. telecom company that has a tool similar to Synario and I have yet to find one...I personally feel that we are light years ahead of most other companies—we know what the current state of our financials are, as well as where we’re going to be five years from now.”
Before implementing Synario, HTC had wasted countless hours of valuable labor on Excel, which also exposed their financials to human error and potential security breaches. After implementing Synario, HTC’s entire stakeholder team rests easier knowing their financials are accurate and their forecasts reliable.
If you’d like to learn more about what Synario can do for your business, please set up a call with one of our specialists through the link below.