The topic of prescriptive analytics or optimization (i.e. linear programming) has received increasing attention over the last few years, including new coverage by analyst firms such as Gartner, Advisory Board, Forrester and several industry thought leaders. Why the attention?

Prescriptive analytics is analytics science that guides decision-making processes for businesses. It adds tremendous value to such processes by providing significant clarity the other types of analytical approaches and data that are required to support decision-making.

Below are 3 use cases that illustrate the value that prescriptive analytics has brought to organizations. To highlight its wide range of applicability, we have chosen cases with different industries (profit-oriented vs. not-for-profit) and different levels of planning (tactical, strategic, etc).

Whether it’s a government organization, a process manufacturing company or even a retail business, prescriptive analytics can be leveraged to make the most informed, sound decisions for in any industry.

Three Use-Cases for Prescriptive Analytics:

Use Case #1: Sales & Operations Planning

Typically used by manufacturers, S&OP used to be largely a process supported by spreadsheets. Eventually, as organizations began to grow in complexity, systems of record came along to support the process. Their goal has been primarily to establish “one version of the truth” by helping users plan and collaborate against one set of data. Next generation S&OP – also called systems of differentiation and systems of innovation, are incorporating prescriptive analytics to help users find not only one version of the truth, but the best version of the truth. In this context, prescriptive analytics helps organizations uncover hidden sources of value. It also helps them identify risk and react to unplanned events.
It is useful to examine how prescriptive analytics supports S&OP:

• Minimally and as a system of differentiation, prescriptive analytics helps companies optimize their supply plans given a certain demand plan. In this context, demand is defined as a number of units of each product by month, location and with an average selling price. The supply plan attempts to meet as much of the demand as possible while minimizing variable cost of supply. The key metrics are how much demand is met, and the total variable cost incurred. Typical value is around 1-3% reduction in supply variable cost
• At the other end of the spectrum, systems of innovation extend the use of prescriptive analytics to full enterprise optimization. Prescriptive analytics are used not only in supply planning but also to optimize demand and finance plans. Demand plans are expanded to include volume/price optionality. Supply planning decisions are expanded to include inventory/outsourcing and capacity planning trade offs. Fixed and variable costs are considered, as well as financial goals and constraints. The key metrics are expanded to net income, growth, cash flow and strategic objectives such as new product introductions, risk mitigation, etc. Typical value results in 1-5% of revenue translated into additional profit and significant increases in business agility and financial forecast accuracy.

Regardless of the extent, the use of prescriptive analytics in S&OP helps business managers identify previously unforeseen insights, increase business agility and establish a better grasp on performance predictability.

Use Case #2: Integrated Delivery Network (Healthcare Provider) Optimization

Most healthcare provider organizations run separate processes to make their key decisions. Clinical guidelines establish their patient care rules and guidelines. The strategy process determines the business model, which markets to compete in and the level and type of service to be offered. The budgeting process is largely concerned with their financial objectives and results in budget and employee (headcount) constraints for each department and location. Asset planning (usually done on spreadsheets by department as an input into an HIS) helps managers determine the use of space, beds and equipment. Finally, workforce planning tools help managers create schedules to ensure key personnel will be available when/where needed.

Now imagine there is a prescriptive analytics system that sits on top to capture the essence of each process. The clinical rules and guidelines establish the baseline requirements for patient care, including treatments, staff qualifications, equipment, supplies and metrics such as length of stay. Strategy guidelines are reflected in the form of objectives and targets. Financials are represented in the form of costs and constraints, such as the firm’s budget, cash flow or borrowing capacity. Assets are reflected including ED capacity (PODs, rapid assessment zones, etc.), floors/wards (number of beds), operating rooms, etc. Each asset is linked to the clinical guidelines and includes activity definitions such as throughput for different types of patients as well as yields such as complication rates. Finally all key staff are represented including physician types, nurses, technicians, etc. All activity, staff and assets are tied to costs. Finally, revenue is represented explicitly either as government funding, tied to activity (i.e. fee for service) or on a pm/pm basis.

The prescriptive analytics model then provides an end to end view of system performance to support evaluation and optimization holistically. Now questions that were evaluated in siloes can be optimized to uncover new efficiencies and opportunities with the confidence that each set of assumptions results in a feasible plan. Example optimizations include:

• Strategic questions –
  • How to best balance different goals? (i.e. financial performance, quality of care, access to care, research)
  • Which programs should we implement? (i.e. patient diversion, outreach, faster bed turnover)
  • Under what conditions should we acquire a physician group?
  • What services should be offered in each location?
  • Should we evolve from fee for service to at-risk models? Should we contract with ACOs?

• Tactical planning –
  • How many beds should we allocate to different floors and types of patients?
  • What should the overflow rules be?
  • How many ORs do we assign to different procedure types?
  • How much weekend and overtime should we plan on for different types of staff?
  • Should we hire another doctor in the ED or in another department?
  • What is the right policy for staffing an OR or consuming supplies?

• Financial planning –
  • What is the optimal budget allocation across the system?
  • Which are the right profit/contribution targets by department and location?
  • What happens if the government were to reduce their payments to us by 1%, 5%, 10%?

Since healthcare provider organizations have not yet implemented unified planning in a significant away, prescriptive analytics in the form of IDN optimization brings an even higher level of value relative to manufacturers adopting S&OP. Thus a $350m organization is able to find $20m per year in additional profitability and a government run organization in Canada is able to uncover significant improvements in cost, access to care and quality of care by better managing their program investments and corresponding trade-offs.

Use Case #3: Strategy & Enterprise Optimization in Oil & Gas

It should be clear to readers by now that the best use of prescriptive analytics is by applying it across an enterprise. In Oil & Gas, this means modeling the key inputs, asset base, supply chain, demand and financials to improve design and on-going operations.

• Design is improved by understanding which assets to own, in which locations and what type of equipment. For example, how much capacity will be optimal? And how much flexibility should be there for the equipment to process different types of input (i.e. different types of crude)? The use of stochastic optimization enables the model to optimize decisions under uncertainty, and by enabling monte-carlo simulation, users can evaluate the range of events under which the decisions still meet corporate objectives
• On-going operational planning is optimized by making holistic decisions. Based on existing contracts, forecasted prices and inventory on hand… which products should we make? Which raw materials should we acquire? Where should we make each product? How much capacity should we plan to run? What if exchange rates fluctuate? What if prices change?

Prescriptive analytics has helped resource-based organizations unlock significant amounts of value. In one case, a large company was able to identify 4% of revenue in additional profit while shortening decision making from weeks to a few hours. In another case, a company was able to optimize the design of a new plant by selecting the most efficient combination of capacity, financial performance and risk.

Next Generation Prescriptive Analytics

We have explored strategic and tactical use cases that apply prescriptive analytics to achieve enterprise optimization. These are not the only use cases, as thousands of applications exist in support of siloed decision-making, for example route planning, juice blending, and so on. While these more operational use cases add significant value, the use of prescriptive analytics across the Enterprise is truly innovative and brings transformational value to government and commercial organizations.

Pursuing enterprise optimization is nowadays much easier with the advent of next generation prescriptive analytics software. Enhancements such as visual, code-free programming, embedded expert knowledge, financials incorporated out of the box, stochastic/what-if representations and the ability to deploy them in the cloud have brought down the time and effort required by as much as 80% while also reducing the risk.

Carlos Centurion is the President of River Logic, a SaaS company that has made its name by being the only optimization software company leveraging prescriptive analytics for true enterprise optimization. Carlos oversees design, development, marketing and delivery of all River Logic solutions. He works closely with prospects, customers, partners and industry thought-leaders to continually improve value delivered to global organizations.

(image credit: Charles Williams)

The travel industry is no stranger to Data Science disruption – especially air travel. Online flight finders have been snapping up data scientists from other industries in order to make better use of the available data, and airlines have been building and opening APIs.

Many services are using this opportunity to offer the cheapest or most convenient flights, but FLYR Labs have a slightly different perspective: Rather than have customers watching flight prices fluctuate, trying to find the perfect time to buy, FLYR offers a predictive service that allows customers to see how cheap a flight is likely to get and lock in that price.

We spoke to Alex Mans, Co-Founder and CTO of FLYR Labs, to find out more about his story and the vision of FLYR Labs.

You’ve been involved in successful technology startups for over a decade now. Can you share any observations or remarkable milestones from that period, from a personal perspective?

The major changes that have driven my decision making over the past decade relate to the fact that with every year that passes, formerly impossible tasks become within reach of anyone who dares to take them on.

When it comes to utilizing data for bettering people’s lives or simplifying the complex, I learned not to be afraid of thinking big as there’ll pretty much always a solution within reach.

How about in terms of technology shifts and the development of data driven technology?

Over time, we moved from descriptive analytics (“What happened?”) to diagnostic analytics (“Why did it happen?)”.

Early on for example, I was involved with a company that analyzed (internet) network traffic to determine whether security threats were present on a company’s local network. This was diagnostic analytics while most people were still focussed on general error reporting or building dashboards to manually sift through data.

Right now, every industry or vertical is trying to move towards the next major category in data analysis. Predictive analytics (“What will happen?”).

The most important and life-changing shift that we’re starting to see today is a move into the realm of prescriptive analytics (“How can we make it happen?”). Intelligent, prescriptive technologies are the ultimate force in forcing changing and simplicity in people’s lives. Done right, it gives people peace of mind.

FLYR’s technology falls into that category as we predict what airfares will look like in the near future and structure that information in ways that the use can easily consume and act upon (e.g. “Don’t buy this flight yet, we are confident that we can find you a better deal”).

Fluctuating airfares is one of the most interesting predictive analytics applications we’ve come across. How did you land on this as your problem to solve with FLYR?

As many of us, I noticed the huge volatility in airline ticket pricing and figured there had to be some kind of recognizable logic to it.

I did a lot of research into how airlines price their seats. I figured that with the right data, understanding airlines based on how they change their ticket prices would be feasible and value could be extracted from that knowledge.

One of your offerings is the Foresight API. What kind of use-cases can we expect to see for this? What does your target market look like?

Foresight is the API that sits right on top of our prediction technology. Foresight can return a day-by-day prediction of how airfares will move for specific flights, complex combinations of flights or simply at a aggregate (route) level.

This technology and API is made available to any company that wants to integrate it into their products, including online travel agencies, meta-search companies, etc.

Looking at the consumer facing side with GetFLYR.com – can you break down how the services works?

Getflyr.com is dedicated to the 98% of people that search for flights, but have no intention of booking today. For those people, we provide the tools to deal with highly volatile airfares while they finalize their plans.

A graph (powered by Foresight) will tell you how much money is at stake if you wait while fares are expected to increase. If we believe fares are going down, you can quickly see in what savings holding off on your booking would result.

To track a flight and it’s predictions for longer periods of time, you can set a FareBeacon alert that tells you when the time to buy has arrived.

If complete peace of mind is what you want, then FareKeep is the thing for you. Instead of purchasing a flight at say 300 USD today, you can lock-in that fare for a small fee (usually anywhere in between 3 USD and 30 USD) for a week. If fares go up over those 7 days, we will pay for the increase. If fares go down or we find you a better deal, you have the option to pocked the savings.

How much data goes into powering your product, and what does your technology stack look like?

Every hour, FLYR receives millions of airfares from all over the world that feed into our data pipeline. This data pipeline normalizes the data before storing it in a massive database. We use a combination of NoSQL and SQL databases, depending on the storage purpose (e.g. pricing data vs. transactional data).

Whenever we get new airfare data, our prediction algorithms are automatically retrained and distributed to the server cluster that sits right underneath our Foresight API.

As an entrepreneurially minded person, what technological advances excite you most for the future?

As mentioned previously, I see us moving in a world full of prescriptive technology. Technologies that understand what we are looking for and help us to achieve our underlying goals based on predicting will happen next.

The major considerations and in some cases also challenges will evolve around balancing relevance and trust between the user and technology.

Alex Mans, CTO of FLYR Labs, is an entrepreneur with 8+ years of experience in launching and growing online and offline tech products. Focussed on assembling great teams with diverse complementary backgrounds and expert skill levels. Ability to connect and work with entrepreneurs, engineers, designers, marketeers, investors and academics by continuously trying to achieve a deeper understanding of technology, design, research and finance.

(image credit: curimedia)