Prescriptive Analytics (Optimization) is a very powerful approach that can deliver significant value strategic, tactical and operational use cases in every industry. Unfortunately, until now it has also been very hard to develop optimization models and firms have required the use of PhDs and significant education for management.
Nowadays with more advanced tools like River Logic’s Enterprise Optimizer and several firms like PwC, Grant Thornton, Accenture, etc. increasingly adopting the science, the problem has shifted to one of education. At River Logic we have developed a simple framework to help our partners and customers identify problems (use cases) that are suitable for optimization.
Basically it comes down to whether you can define the problem as a constraint model that can be optimized, and determine the differentiated value/insights that can be generated:
- Decision space – what decisions, trade-offs, interaction between possible decisions are required to solve the problem? Can you define a complex decision tree?
- Definable objective(s) – what variable or combination of variables are you trying to maximize and/or minimize? Examples include maximum profit, minimum cost, maximum throughput or volume, maximum ROIC, best quality, etc.
- Constraints that can be quantified – which are the policies, regulations, physical, financial and market limitations that drive decision trade-offs and that impact the value of the solution and level of insight? (remember constraints are important in defining marginal vs. average profitability). For example, HR policies, resource capacities, marketing budgets, # of marketing messages into an audience within a time period, blending targets for coal, etc.
- Minimum level of complexity – how big is the problem? Consider a minimum of 10 decision variables (e.g. 4 products,3 resources, 3 customers). Other dimensions include time, inventory, channels, marketing events, inputs, working capital policies, etc. A typical River Logic solution has tens or hundreds of thousands of variables
Problems that are suitable to optimization are everywhere, but until one learns to spot them everything looks like an Excel model. Once you look at the problems differently, one starts to see tremendous opportunity to make better decisions. Here are a few places where you can look:
- Policies that guide behavior, where decisions are made out of habit or where people fail to make decisions (e.g. we always source products from this plant to serve this market; we prioritize our customers based on volume/revenue; all our surgeries start at 6am because that’s how we’ve always done it; our oncology clinic opens from 8am-5pm, even though we’ve never really analyzed why)
- On-going complex planning processes that are treated sequentially today or – even worse – that are made in isolation. For example:
- Sales & operation planning, where decisions about sales/marketing, manufacturing, procurement, distribution and finance are made sequentially using different tools and/or spreadsheet models (note S&OP processes exist in almost every industry)
- Decisions driven by silo thinking – tactical and strategic decisions involving resources, product/service mix, marketing, etc. that are made solely within the function
- Highly dynamic situations where input/product prices change constantly, regulations evolve, etc. – these situations require dynamic optimization analyses; examples include commodities industries, the U.S. healthcare industry, chemicals, oil & gas, some finance products, etc.
- High difference in average vs. marginal decision making – when there are multiple constraints, volume contracts and output price differentials, very often there can be up to 100% difference in average vs. marginal profitability for the same products
- Underserved markets/industries – markets where the problems are relatively complex, but where practitioners over-rely on BI and Excel tools to make decisions; examples include some government services, some US healthcare providers, etc.
Look at your department, function, division or even the whole organization. Think about all the possible uses of the people, machines, funds that are around you and the decisions that guide people’s actions today – even something as simple as how tasks are prioritized. You will most certainly begin seeing opportunities for optimization.