Life is really
simple, but we insist on making it complicated.
(Confucius)
Why is analytics
so important? Making
analytics part of the competitive strategy of a company nowadays offers an
extremely powerful competitive advantage,[1] since a good analysis allows
companies to make better decisions, which mean reducing costs, increasing
margins, making customers more loyal, or gaining more customers. Moreover,
there seems to be a shortage of human resources in data analysis (from more
technical to more business-oriented roles) that will last for several years.
What is Business
Analytics? The BABOK®
Guide[2]
defines business analysis as “the set of tasks and techniques used to work as
liaison among stakeholders in order to understand the structure, policies, and
operations of an organization, and to recommend solutions that enable the
organization to achieve its goals.” According to the guide, business analysts
should be competent in several knowledge areas, such as planning, monitoring,
elicitation, management, requirement analysis, and so on. Besides, they must
have robust underlying competencies to find solutions and facilitate the
decision-making processes necessary to attain the company’s goals. To find
these solutions, business analysts perform a series of tasks using several “techniques”:
here, at least for me, is where the most exciting part of business analytics begins.
What the BABOK®
Guide calls “techniques” is a set of models, techniques, templates, and best
practices that business analysts use more or less creatively to solve a
problem. After reading several books, articles, and posts, taking years of
university and business courses, and applying several techniques as a business
analyst, what I was lacking was an organized and simple toolbox containing all
the models, techniques, and templates that I had used and learned so far. However,
what I found in my search was either too limited or too extensive and technical;
in other words, I could not find a complete and balanced toolkit that is not too
complex but has enough details to implement the models. An addition problem was
that information was usually spread across several sources, and the more complete
ones only contained models concerning a specific topic (e.g. statistical
models, business models, or pricing models).
Checking
for references among books and websites, I found several options; for example, Business
Analysis Techniques: 99 Essential Tools for Success[3]
is a good reference book that organizes and explains several techniques. In
addition, www.real-statistics.com and www.excelmasterseries.com are two notable websites that I have used for many of the statistical models
proposed in this book. I suggest visiting them for more in-depth information on
statistical analysis in Excel.
However,
what I was looking for was a reference with practical models for fast decision
making, so this book focuses on those models and techniques that helped me to
make the predictions, forecasts, and estimates that I needed to make business
decisions. Since I decided to concentrate on analytical models, this book omits
those models that are more focused on other aspects, for example project management.
This is why
a year ago I started this project that you can read now and that aims to create
a toolkit that is as complete as possible, with ready-to-use templates for descriptive,
predictive, and prescriptive analysis. I tried to
make the toolkit as simple as possible, so almost all the tools that I propose
do not need specific technical skills or the use of specialized software. In
fact, for each tool presented in this book, it is possible to download a
ready-to-use template in Excel. I have to admit that I also included some more
complex models that require the use of specific software, but for the rest of
the models the template includes all the formulas and instructions required to
execute them.
I greatly
appreciate simplicity, and I think that experts (and non-experts!) often tend
to complicate simple things, maybe to show their knowledge or maybe just
because we think that complex problems need complex solutions. However, I strongly
believe that, to solve a problem efficiently and effectively, it is usually better
to simplify it as much as possible, starting by formulating a simple question
that reflects the objective of the research or analysis: Do we have to increase
or decrease prices? How much? For which product? For which customers? When?
The second
step is to identify the scope of the analysis:
- - What happened? (Descriptive analysis)
- - Why did it happen? (Diagnostic analysis)
- - What will happen? (Predictive analysis)
- - What are we going to do? (Prescriptive analysis)
The third
step is to select the model or technique that is most suitable for the
situation. The fourth step is to check for available data and/or determine how
to gather additional data from both internal and external sources. It may
happen that it is not possible to collect all the data needed for a specific
model; in this case we can either 1) try to use proxy data as a substitute for
the unavailable data or 2) change to the second-best model. The fifth step
concerns the proper analysis and validation of the results, and the final step
is to communicate the results properly.
An
excellent book, which I highly recommend to all analysts, is Superforecasters,[4] which summarizes an extensive study
concerning the patterns behind the most accurate forecasts and the best
practices to achieve a good prediction of future events. Among several best
practices, one that I admire is the “Fermi estimate.” This method, attributed
to the physicist Enrico Fermi, consists of breaking down a complex estimate into
several simpler estimates (see 2. SUPERFORECASTING).
Before
leaving you to the tools of this book, I would just like to suggest that,
whenever you perform an analysis, try to be flexible, be creative, and enjoy
the process. Happy analyzing!
