# Assessing Your Profit Risk Due to Volatility

Oftentimes in business, we are forced to live with uncertainty.  Really, nothing is certain, but even moreso certain volatile costs like gasoline, produce and currency exchange rates.  For example, a quick look at the following price of regular gas (credit: GasBuddy.com) shows some pretty heavy fluctuations between \$1.70 and \$2.80 per gallon for 2015 and 2016 YTD.

While this causes mixed feelings at the pump, imagine the headache that major businesses face when an upward or downward trend can make the difference between a more or less profitable year.  Clearly, volatility is not something that can be reasonably dealt with on intuition alone.  As an executive, it is easy to say that gas will float below \$2.00 per gallon once again and profits will soar.  But what are the odds that \$2.00 will actually happen?

Most of us don’t enjoy math, but we all do love food, so I will attempt to explain volatility risk with pizza.

Imagine you are an executive at a large pizza chain.  Your best selling pizza (for simplicity purposes) is a medium cheese pizza.  Also for simplicity, your material costs to make the pizza consist of dough, cheese, and sauce.

Let’s assume that dough and cheese costs rarely fluctuate up or down.  In fact, dough and cheese costs remained at \$4.00 and \$2.00 respectively for all of 2015 and are expected to remain the same for all of 2016.

Pizza sauce, however, is as volatile as gasoline, and fluctuated between \$4.00 and \$5.00 per pizza in 2015, with a definite upward trend heading into 2016.  At a sales price of \$25 per pizza, this amounts to a total material profit of \$15 (60%) at the beginning of 2015 dropping to \$14 (56%) by the end of 2016 for a decrease of 4%.  At about 160,000 pizzas sold, this amounted to profits of almost \$100,000 less than expected.

Your colleagues are very optimistic that pizza sauce prices will level out to begin 2016, following the unprecedented increase of 2016.  Therefore, they are forecasting your material margin to remain at 56% for January 2016.  However, given the upward trend, this does not settle well with you and you want to assess the risk associated with this assumption.  What can you do?

One approach would be to perform a Monte Carlo simulation to determine the likelihood that your margins exceed a given percent based on the likely fluctuations in pizza sauce.  Simply put, we can run a simulation of hundreds or thousands of possible outcomes of margin percentage given the likely fluctuations in pizza sauce.

In 2015, we saw our pizza sauce costs increase monthly by an average of 2.6%, with a standard deviation of 0.05.  Assuming a normal distribution, we generated 1,000 random possible outcomes for pizza sauce increases (or decreases) in January 2016.  As you can see from the below distribution, there were in fact numerous outcomes where margin did, in fact, reach your colleagues’ estimate of 56.0%.  To be exact, 476 of the 1000 outcomes reached 56.0% or higher margin.

However, let’s look at this another way. In the below chart, we have rearranged the data to show the percent likelihood of each threshold.  In this view, we see that we only have a 48% chance of reaching our forecasted 56.0% margin.  However, we can say with 84% certainty that we will reach a 55.0% margin, and with 94% certainty that we will reach 54.5% margin.

Using this method, you are now armed with the ammunition you need to contest the 56.0% margin forecast.  As an executive, do you feel comfortable betting your company on 48% odds?  Or would you rather work with 90% or better odds while focusing on other pieces of your business that are more under your control?

The answer is different for all types and sizes businesses, but the key takeaway is that you should not move forward with decisions without understanding the risk.

# How To Quit Non-Value Added Reports

As we mentioned previously, data is being collected at a record and ever increasing pace.  While the mounds of data continue to grow, businesses continue to invest in IT infrastructure, data warehouses, and software solutions.  They also spent a lot of time cleansing, manipulating, and scrubbing data to get to the information that they need.

A LOT OF TIME.

So much time, that it is easy to get stuck in the rut of going through the motions of creating a report instead of focusing on what is important: the end result.

When it comes to recurring reports in particular, the process of prepping data and creating the report can often become so time consuming that the process eventually becomes the goal in and of itself.  The reason the report was created in the first place is forgotten and the analyst (or whoever is creating the report) views the report creation process as the most important part of her job.

Here are four ways to focus on the end result of your data — valuable and actionable information — rather than the report creation process itself:

This should go without saying, but if a report does not add value, then it shouldn’t be created in the first place.  There are much better places to spend your time and labor than on reports and processes that mean absolutely nothing.  To ensure that you are only reporting on information that adds value, start with the desired outcome and work backwards.

If the goal is to decrease shipping costs, then the desired outcome may be, “We need to size up the amount we would save by bringing all carriers up to a weighted average.”  You can then work backwards to determine what shipment data and other data you need to reach that outcome.

Likewise, if you need to increase material margin by 5%, then you can set your goal and think about what information may get you there.  In this case, Pareto may be your friend, and you may also gather other metrics that will help you determine whether price increases are warranted or if you can obtain discounts with your suppliers.

Whether the goal is monetary or measured in some other way, every report should have a specific ROI measure in mind or it should not be created at all.

Never manually create a report more than once.

Every executive has experienced that monumental hunch that warrants pumping the brakes on everything else and requesting supporting information immediately.  Sometimes that hunch leads to real results and the assumption is that it will continue to lead to the same results month after month.

However, report creation is subject to the same laws of diminishing returns as everything else.  What took five hours to create and saved the company \$1 million a year ago may continue to take five hours to create with a much lower rate of return.

Assuming the report is still relevant in the first place, there is no reason to be investing the same amount of time month after month.  If you are going to create a report more than once, every step of the process should be automated and (easier said than done) documented.

Automate early, automate often.

Continually review reports and cut out the ones that add the least value.

Perhaps the single more worthwhile exercise than automating your data cleansing and reporting processes is eliminating the reports that do not add value in the first place.  When the ROI on a report is non-existent, then you need to free up time and resources for processes that add value.

100% improvement on zero is still zero.

Hire a second set of eyes.

Sometimes it is too difficult to buck the trend of reporting for the sake of reporting.  When things become stagnant, it can often be helpful to bring in a second set of eyes to re-energize the business and provide insights that would not otherwise be obvious.  This is particularly a win-win when it frees up your resources to do other things, like completing operational tasks or tackling other issues.

Rather than viewing help as a cost, see it as an investment in greater returns.

If there is one major takeaway from this article, it is this:  Analytics should not be viewed as a cost and a necessary evil.  Rather, analytics is an investment that, when applied correctly, can yield significant ROI.

# What Is Your Logistics Data Worth?

Most areas of your organization are generating data faster than ever, and your supply chain and logistics functions are no exception.  However, we often find that this data is used to make more simple daily decisions, such as which trucks to send via existing lanes and providers, rather than more complex changes that result in significant ROI for the organization.  Here are a few analytics techniques that we find useful, and can greatly reduce your shipping and logistics costs.

Use Case 1: Company A decided to engage in a project to better understand what it is paying for its shipping lanes, which are a mix of truckload and less-than-truckload lanes across the United States.

The logistics group has built long-standing relationships with certain providers and, although they are not the lowest cost providers available, the combination of service and value makes sense to the group.  Additionally, being the loyal group that they are, they feel that they owe it to these providers to maintain the long-term relationships that they worked so hard to build.  Besides, changing the network is a grueling task and it is difficult to get the information to prove a change should be made.

This is common, and often works just fine for the business.  However, just how much are you leaving on the table when you take this approach?  The answer typically lies in the combination of your historical shipment data and other third-party data sources.

For one, you can check load boards, such as DAT, to check the current rates for lanes that you want to question.  In addition to load boards, there are myriad companies offering third-party benchmark data where you can enter lane and product information to determine what others in the industry are paying.

You should use your historical shipment data in conjunction with these third-party data sources to 1) benchmark shipments against industry standards, and 2) benchmark your shipment lanes against one another.

Using both of these methods provides you with a couple of data points to determine which 3PL providers you may be overpaying.  Particularly, when both internal and external benchmarks point to lanes with high costs relative to other lanes (after factoring in product dimensions and density, lane frequency, etc), there is a significant opportunity to decrease costs in your network.

Many organizations that take this approach realize savings of 10% or more in their annualized shipment costs.

Use Case 2: Company B makes pricing decisions based on either material margin or gross margin excluding shipping costs, due to the complexity of determining shipping cost at the product level.

How can you determine the optimal price for your products if you do not have a full view of the cost to deliver them to the customer?  Businesses often ignore or make general assumptions about shipping costs for one reason: they are not available at the product or transaction level.  Due to the nature of shipments, these costs are typically available at the shipment waybill level, meaning dozens of products may fall under the same cost umbrella.

The good news is, with some careful planning, there is a way to more accurately view shipment costs at the product and/or transaction level.

First, you need to bridge the gap between your sales transaction data and your shipment data.  Typically, this involves merging multiple data sets together based on nothing but date, waybill/shipment #, and little else.

Second, you need to determine the best way to allocate shipments across products for your organization.  Depending on the types of products that you sell, you may chose to allocate based on product weight, dimensions, quantity, or some other measure.  It is recommended to keep your allocation logic flexible so that you can test out each of these methods until you find the measure or measures that work best.

Finally, you need to write the logic to perform the allocation so that you can add the allocated cost to your transaction data.  This is usually best completed using an automated SQL script or similar method so that it can be reproduced on a daily, weekly, or monthly basis.

Once you have solidified the allocation of shipment costs into your transaction level data, you can now roll it up to the product level for a better understanding of total costs.  Although you may not find surprises across the board, there will likely be some significant opportunities to increase prices based on your deeper understanding of total costs.

Do you have additional suggestions for how to turn logistics data into higher margins?  Please let us know in the comments, or visit us at http://www.clevity.com to learn more!

According to IBM and other sources, over 90% of the world’s data has been created in just the last two years.  At first that sounds crazy, but if you stop to think about it, we now have sensors counting visitors to stores, POS systems that track customer information, and just about any detail you can think of around each part that is manufactured and each item that is sold.

Companies everywhere are investing heavily in IT.  Some are building an analytics function as well.  The rest, hopefully, will be following suit.  Here are five “must haves” as you look to build your analytics function:

The right technical tools differ from business to business.  In most cases, it will suffice for the analytics team to pull the necessary data into a SQL data warehouse to analyze the data, and tools like Excel and Tableau to present findings.  However, some companies, particularly those with terabytes of data, may require “big data” tools such as Hadoop and NoSQL.  If you are not sure, seek opinions from a third party to make this determination.

3) Executives and upper level management must be bought in

Any investment you make in the analytics function will be all for naught if management refuses to listen to the results.  Despite the upward trend in data, many businesses still prefer to run on intuition and personal knowledge rather than listening to what the data is telling them.  Both intuition and data are valuable and are best used hand-in-hand, so long as your are willing to let the data tell you when you are wrong in addition to when you are right.

4) There should be dollars tied to findings

Of course, management cannot be expected to listen to the data without good reason.  The analytics function is one that should be able to pay itself off.  Most, if not all, analytics projects should be tied to a dollar figure that shows management how much the bottom line will be increased should they follow the given recommendations.  For example, bidding out existing freight will save X% of shipment costs or optimizing labor schedules will result in \$Y million revenue uplift.

5) You should build the anaytics function now

If you have not yet built an internal analytics function or brought in an external team to assess your business, chances are that cost is the main culprit.  However, what many organizations do not realize is that the ROI on such projects typically far outweighs the costs.  A small investment in the right resources now will pay dividends in the future to your bottom line.

Many sales organizations are driven by either revenue or margin targets.  The more the sales staff sells, the more commission is paid out, and the better they perceive both their own and their company’s performance.  After all, it is comforting to know that your pockets are deeper at the end of the day, right?  Well, maybe that is not the case.

Consider the following sales snapshot for a given fiscal year, where the organization as a whole brought in an average of over \$100,000 revenue per sales transaction and kept almost \$50,000 per transaction in margin.  From the perspective of the sales team, they take home a nice commission, the business reports a nice return to shareholders, and everyone goes home happy.

However, what happens when you dig beneath the surface?  Is every transaction a good transaction, every product a good product, and every customer a good customer?  Chances are, if you look at your sales on a transaction by transaction basis, you will see some clear winners as well as some less than desirable scenarios.  Consider the following graphic drawn from the exact same data as the above, but showing transactions spanning \$2k to \$250k with margins between 20% and 60%.  That is quite a range.

There are many questions that this brings up.  Here are a few of the key things you may want to ask:

1) How much time and resources are we spending on the smaller transactions compared to the larger ones?

The Pareto Principle has taught us that roughly 80% of effects come from 20% of causes.  Let’s consider this in the context of our sales data with the following two statements:

80% of revenue and margin come from only 20% of customers.

80% of revenue and margin come from only 20% of products.

This is a concept that is simple enough to prove or disprove for your organization.  It is pretty easy to determine who are your biggest customers and products and who/what is in the long tail that represents the bottom 20% of your margin.  This isn’t to say that the bottom 20% is not important; however, this may be a good place to start when considering how to move forward with your business.  Ask yourself these questions:

1. Is there anything your sales staff can do to move the “bottom” customers to the top?
2. Are there any products where you can decrease cost or boost sales?
3. Can you use automation to minimize the time you spend selling products that don’t add much to the bottom line?

Keep in mind that you will likely need additional information to answer these questions.  We will discuss some of this additional information below.

Smaller transactions may look great for your gross margin, but by the time you factor in the overhead of sales and administrative costs, as well as time spent by your operations staff, you may be selling some of your product at break-even or at a loss.

2) Are the lower margin transactions known and intentional, or are they flying under the radar?

As we saw in our “Revenue vs Margin” chart, even high revenue transactions range from “GREAT!” down to “What’s going on?”  In order to gain a better perspective on where you are seeing high margin vs. low, let’s consider the margin percentage relative to sales velocity (in this case, revenue dollars).

As you can see, the transactions represent four distinct scenarios:

1. Low Velocity, Low Margin (lower left)
2. Low Velocity, High Margin (upper left)
3. High Velocity, Low Margin (lower right)
4. High Velocity, High Margin (upper right)

Why are we wasting time on low velocity / low margin products and customers?  Maybe there is something that can be done to increase both the margin and the velocity.  However, maybe it is time to drop some of these products and/or fire customers that do not contribute to the bottom line.

Low margin / low velocity products and customers are not providing adequate return on your investment dollars.  Use your data to find the best place to put your money.

3) What are we doing right with those high revenue / high margin transactions?

As we see in the green box in the quadrant chart above, there are some products and customers that represent both high velocity and high margin.  Challenge your organization to determine what sets these apart so that this can be replicated.

If even a small portion of the lagging products and customers were dropped so that your organization can invest more deeply in high velocity / high margin transactions, then you may be able to see huge gains in your bottom line.  This would lead to higher commissions for your sales staff as well.

Clearly, we cannot just create a few charts and declare that we should drop x% of our products and fire y% of customers.  Rather, we need to look department by department, branch by branch, customer segment by customer segment, etc. to seek out potential opportunities.  Many organizations complete in the hundreds of thousands to millions of transactions in a given year.  In order to make the best use of this data, you must employ the proper tools and skill sets.  Visualization tools such as Tableau and QlikView, used in conjunction with a data warehouse, can help us to tie together separate charts and sources of information so that your organization can easily make a determination on where to continue to invest its resources.

For information on how your organization can utilize data analytics on its sales transaction data to reduce complexity and boost profits, please visit Clevity at http://www.clevity.com.

# IT Is Not Analytics. Here’s Why.

A CEO that I worked for once gave me the following feedback when I proudly sent him an Excel file containing the data that he had asked me for less than an hour prior:

Never give me data.  Only provide me with information.

This was one of the most valuable lessons that I have ever learned, and I have since taken it to heart each and every time I am tasked with turning raw enterprise data into digestible information.  That afternoon, I was sent back to the drawing board to create summaries and metrics from the raw data, which helped us to better understand transit times and claims rates for a client we had been providing with logistics services.

However, prior to my arrival at the company, it was clear that all of these types of requests went through the IT group, who in turn provided a very similar data dump to what I had provided the first time around.  This was when I truly came to realize the difference between what IT and analytics bring to the table.

Analytics, big data, and data science are relatively new terms.  Rarely do you walk into an organization and find an analytics or data science team, or even one person solely responsible for turning data into actionable information to improve the business.  In fact, most have not even developed robust reporting solutions that enable the business to progress with relatively low human interaction.  Rather, they have bolted reporting and analytics responsibilities onto the already busy schedules of their IT professionals.  This is as mistake.

Information technology is not analytics, and analytics is not information technology.

Almost every client I have walked into gets its information as follows:

1. Stakeholders think they have stumbled upon a eureka moment that is going to progress the business
2. Said stakeholders contact the IT group to request a “report” with the information that is needed to back up the eureka moment.  Rather than a request for valuable insights to get to the final outcome, this tends to be a very specific request with little room for creativity, such as “Give me the number of accessorial charges that we paid to each carrier by week over the past three months.”
3. Three weeks pass.
4. IT provides the “report”, which typically is a data dump into Excel that comes out of a SQL query the IT professional wrote in between his actual IT tasks.
5. Several iterations occur until the stakeholders gets exactly what they asked for (e.g. a variation of the original data dump, but with additional columns).

This is counterproductive for a variety of reasons.  One, the productivity in IT has suffered and, rather than focusing on database maintenance and other IT tasks, the group has been struggling to keep up with ad hoc requests from the rest of the business.  Two, the stakeholder has likely chased down a self-fulfilled prophecy in that, by asking for a very specific data dump, he has gotten exactly what he needed to prove his original point.  Three, the question that gets answered it likely not the question that is going to improve the business.

Analytics professionals are the liaison between the data that is stored on the company’s servers and the insights that are needed to propel the business forward.

Every business that is looking to grow should have one of these analytics professionals in the trenches.  If you are unable to develop an internal analytics team at this time, then bring in a consultant or team of consultants to provide this service.  The ROI on the insights that you will gain will far outweigh the up front costs, and the time to completion will be a fraction of the “just have IT do it” route.

At a minimum, your analytics team should have technical skills, such as the ability to create a data warehouse out of various sources of data, experience on writing SQL code and stored procedures, and skills in other analytical tools, such as Excel, R, and visualization tools.  Additionally, the analytics team should have experience on the business side.  They should understand finance, sales and operations, and should be comfortable dealing with the executives in every branch of the organization.

Often, it may make sense to start small by taking on just one project, such as finding ways to reduce your logistics costs, or looking at which of your products are contributing to the majority of your margin (and which are just over complicating your business).  Then, once you have gained confidence in the payoff of analytics, you can develop internal analytics resources and expand the function to a more holistic view of the organization.

If your goal is to obtain data dumps containing exactly what you asked for, then by all means, bury your IT group in report requests.  If you would rather gain valuable insights to propel your business forward, regardless of whether they have been discovered or not, then build up your analytics function.

# The Big Data Myth

Big Data has become the craze of the business world.  Companies are spending millions of dollars on the latest technologies and are hiring data scientists in droves, seemingly in a rush to stay technologically relevant.  But what is “Big Data” and is it right for your business?  For many organizations, the answer may come as a surprise.

The definition of Big Data is as varied as the companies that claim to have it.  Consider this definition from a quick Google search:

noun: big data

1. extremely large data sets that may be analyzed computationally to reveal patterns, trends, and associations, especially relating to human behavior and interactions.

“much IT investment is going towards managing and maintaining big data”

There is no question that Big Data does, in fact, offer myriad opportunities to reveal invaluable patterns and trends that can lead organizations to both uncover unknown problems and discover completely new products, services, and ideas.

Take, for example, the popular book (turned movie) Moneyball.  Major League Baseball is a game of hundreds of players throwing thousands of pitches during scores of at bats.  Entire companies exist to collect the millions of records of statistics each year to detail who threw a pitch, who was at bat, details of weather and wind speed, how fast was the pitch, whether it was a fast ball or curve ball, if the batter swung or not, whether he made contact or not, where the ball went… the list goes on and on.  Clearly, the vast amount of data available across multiple seasons of baseball are not manageable with everyday tools by the normal computer user.  That is why teams like the Athletics and Red Sox have become notorious for hiring data scientists to navigate the data and create information to better their organization.

The need exists in other industries as well.  The airlines have countless streams of data around flight schedules and the flights and passengers themselves.  Twitter and Facebook have millions of posts on their sites a day.  In these cases, hiring a team of computer science and statistics PhD’s is an understood necessity to utilize the data.

So, why are we calling “Big Data” a myth?  Let’s go back to the definition and consider the term “extremely large data sets”.  What does that mean?  In the case of Major League Baseball, airlines, and social networks, that can mean terabytes of data representing billions of data records.

However, chances are that most people reading this do not work in one of these places.  The typical organization is going to have thousands (most manufacturers and B2B organizations) to millions (larger retailers) of transactions.

This does not represent Big Data.  This is just Data.

Big Data requires huge investments in infrastructure, rapidly evolving tools such as Hadoop and NoSQL, and scores of data scientists, who tend to have mostly academic backgrounds.

Data requires more modest investments in infrastructure, common (and often free!) tools such as Microsoft SQL Server, PostgreSQL, Excel, and Tableau, and experienced analytics professionals with real world business experience.  Most organizations need analysts who are both technically skilled and have the ability to translate data into actionable business insights.

# Data Detox: The Importance of Cleaning Your Data

Imagine enjoying years of everything going your way.  You land a great job, you have an amazing family and friends, and you live in your dream home in your dream city.  In fact, the only real problem that you have is finding the time to fit everything and everyone into your busy lifestyle.  But, rather than sacrifice your dedication to your job, or time with your spouse and children, you instead forget to focus on yourself.  Rather than engaging in regular fitness and eating healthy, you have evolved a sedentary life as a desk jockey, and your idea of a healthy meal has been to forego the supersize option.

Then, one day, you look at yourself in the mirror and realize that while everything around you seemed to be going perfectly, your lack of health has caught up to you.  Your family continues to grow, but you can’t keep up with them to the point that you decide you need to take action.  So finally, you enter detox mode:  you start to eat healthy and you join a gym and begin to work hard on yourself.  Three months later, you can’t even imagine what made you get to that point in the first place.  You’re more fit, you feel better, and you have the energy to keep up with the rest of your life.

For many businesses, this describes exactly the state their data storage and analytics functions are in.  When the business was small, a few Excel spreadsheets or even a notebook and pen were though to track the business and take it to the next level.  However, the business continued to grow and eventually you could no longer keep up.   More and more data began to pour in and your one page spreadsheet became a database (or, even worse, many Excel spreadsheets that you refer to as a database!).  And your database became a data warehouse.  But you don’t know where anything is or how it all ties together.

Imagine I were to ask you the following question right now:  What was your revenue by product/service for each month this year, and how many of each product/service did you sell?  You would be amazed how many businesses I come across that cannot answer this simple and important question.

How can you run your business without even knowing if you are making money?  Luckily, there are some steps you can take to put yourself on the path to greater knowledge of your business.

1) Assess your data and take stock of what information you have available

As your business grows, data silos emerge.  Sales has its own database with transaction data, and they are told to “sell sell sell” with only a view of revenue and sales targets, and not margin or EBIDTA.  Supply Chain has a list of shipments by waybill that it cannot tie to sales so they manage themselves solely by the truckload.  HR manages payroll, but has no idea what each employee produces.  The manufacturing floor keeps making more and more low margin product, but has no idea that they are under-producing the highest margin products (hint, you should be utilizing the 80/20 Principal).

Every department holds key information that would be invaluable to everyone else around them.  Interview stakeholders in each area to find out what they need, as well as what they can offer.  Find out what software they use, what reports they create, and where the data comes from to create those reports.  Once you know what information your organization has, you can determine how you can use it to improve your business.

2) Clean your data!

What good is your data if it doesn’t make sense, is difficult to navigate, or is inconsistent to the point that you can’t trust it?  You need a resource who can learn the data inside and out and whip it into shape.  Are you consistently seeing sales transactions with negative revenue?  You need to know exactly when that represents a product return, and when it represents a more glaring issue.  Do you have a store that is sometimes referred to as “Times Square”, while other times it is “TSQ”?  You have a data consistency issue that needs to be addressed.

This may also be a good time to hire a data governance expert.  Think of this as your data project manager whose sole purpose is to install guard rails around your information to enforce data integrity.

3) Create a system going forward.

A massive data cleanse is all for naught if you spend months completing it, only to go back to your old ways.  Document the steps that you took to clean the data and put systems in place (preferably as automated as possible) to continue keeping the data clean going forward.

If you complete these three steps, then I promise you will reap the rewards by giving your team the tools to discover vast opportunities for margin improvement.

Once the data has been cleansed, you can provide your team with dashboards, tools, and insights that can be used to find that next “eureka” moment to launch your business forward.