What is optimization good for anyway? If your business is running data science models and utilizing machine learning without realizing ROI, it’s time to look at optimization. Many leading firms have saved millions of dollars using optimization. Others have sped up delivery times, made better use of inventory or warehouse space, and met “stretch goals” without increasing employee counts.
Optimization is the process of finding better ways to allocate resources than you (or anyone) can figure out manually. For line-of-business leaders, it’s a means of maximizing ROI and driving real-world results. When demand slows down or financial conditions become tight, optimization is more valuable than ever.
Differing Viewpoints: Analysts vs. Line-of-Business Leaders
Analysts working in the line of business see the potential of decision models, while line-of- business leaders are focused on cost reduction and efficiency. Their perspectives differ, but both play key roles in achieving business goals.
Too often, “data scientists” who work for a consultancy or remote central group aren’t close enough to the line of business to see the local opportunities. The best way to achieve results is to empower your analysts in the line of business with simple, powerful analytic resources they can use, taking advantage of their knowledge of how the business works.
It’s really the perfect partnership. You, as the line-of-business leader, don’t need to get into all of the math in the decision models your analysts will build, but you can understand what’s possible, and focus their efforts on the operational “pain points” that matter most.
Data from the Ground Meets Solver Decision Models
Here’s a simple example of bridging the gap between math and real-world conditions . . .
A consulting firm or software vendor might offer a solution for scheduling deliveries using a fleet of trucks. They use public information on distances between addresses, EPA listed fuel economy for trucks, and cubic foot space allowances listed on the truck spec sheets.
On the ground however, your drivers know they are losing two miles to the gallon over a mountain pass, especially in winter months. They also know about lost space because of custom cargo racks that are necessary for safe transportation of your products.
A line-of-business analyst who has this real-world information can tighten constraints and return optimized results that better match the reality on the ground. Now, managers can see the value in changing a route and/or using cargo racks that hold more inventory, as a means of better serving customers at a lower cost. New options are on the table because your analyst truly understands the factors at play on the road.
Where Data Science and Machine Learning Fall Short
Why bother with optimization when we invest in data science and machine learning? Isn’t optimization just an extra, more esoteric part of what data scientists do?
The question warrants a response. As many line-of-business leaders like you have seen, it can be difficult to demonstrate tangible ROI from data science and machine learning. These tools return predictions, but that simply is not enough to change outcomes.
With predictions alone, your staff is still making decisions based on “best guesses” or past experience. Humans can’t be expected to find the “best” among thousands or even millions of alternatives. For many problems, data science and machine learning just aren’t enough. Optimization is a tool to actually find the “best” choice among many alternatives.
Taking Optimization to the Decision Level
In the previous section, we discussed how data science and machine learning fall short. The inability to make the best decisions, given only predictions, is exactly what we meant. Optimization is the ticket to actually realize the promise of decision intelligence.
Think of it like this: Decision + Action = ROI.
Without real decision intelligence, your staff is taking action based on their best guesses.
So, how can YOU take advantage of optimization as a line-of-business leader? You have several options:
Buy industry-specific packages with optimization included.
They exist for many industries but there are a few big downsides. The packaged approach is the least flexible and may not fit your specific situation well. They are also the most expensive option.
Hire a consulting firm.
Some great firms are available, and they can build excellent optimization into decision models. They are great for one-off models but are not cost-effective for ongoing work. Your investment also doesn’t build internal capabilities in the company to impact ROI continually.
Build and grow optimization capabilities within your company to maximize impact.
Most MBA grads in the past couple of decades have exposure to Analytic Solver®. So, you likely have multiple internal employees with the capability to build optimization models already. Moreover, they will have the domain experience to really optimize effectively – and ensure the model is maintained for long-term usage and success.
Before going down an expensive rabbit hole of outsourced options, put out an internal request and leverage the talent already lurking within the company. If you don’t already have the best-in-class solution, have your analyst try it free now.
Optimization: What NOT to Do
We’ve seen three common mistakes made when trying to use optimization – and decision intelligence generally – to impact ROI:
Mistake 1 – Simply not using these methods – especially when competitors are using them. Many are still flying by the seat of their pants when it comes to making final decisions. While gut instinct is important, it’s not a replacement for having real decision options returned from a model that captures all the important decision variables and constraints.
Mistake 2 – Optimization sounds like a mathematical method, so business leaders leave it to the experts. Applying optimization effectively however, requires domain knowledge to understand the problem and build and solve the right model. Relying on a central group or external short-term consultants is often a mistake. Make sure that people with the right domain expertise are included!
Mistake 3 – Businesses stop when the model is initially built and solved. It reaches the boardroom for presentation but is never embedded into actual operations. Optimization needs to update, evolve and continue working within the framework of operations to make an impact. Make sure your team plans for deployment and ongoing use!
When combining internal talent with Analytic Solver, you have a cost-effective means of optimizing decision intelligence that can have major implications for ROI year over year.
Optimization in the Real World
The impacts of optimization, especially when applied through Analytic Solver, are backed by many success stories.
Downstream Advisors is an interesting case study from the petroleum industry. They developed a Smart Refinery software package that uses Analytic Solver optimization models. The virtual refinery can simulate real-world conditions and where and how crude products are routed and delivered.
Note how optimization reaches the decision level by showing the best routes and delivery mechanisms to maximize efficiency and profit. It’s applicable, not predictive.
In the world of finance, the concept of “Efficient Portfolios” is also an optimization model. Allocating money into stock portfolios without accounting for correlations against each stock is a dangerous proposition.
With optimization models however, accounting for positive and negative correlations reduces risk exposure. One savvy financial advisor removed the guesswork from his process as a means of better serving clients.
The model he built for his independent advisor role put him on equal footing with large corporate financial advisors with access to a suite of optimization tools.
With Analytic Solver, better decision intelligence realized through optimization models can be leveraged by your existing team, in the near term. Empower your teams to be analytic heroes. Start your free trial now.