Retail Operations

Data Driven Decision With Big Data

Share this post:

Decision Making With Big Data

Data Driven Decision Making with Big Data

Second part of the series of publications based on the presentation I gave at the event Big Data & Analytics Summit, with the title Marketing strategies for generating data.


Is the new natural resource and its superabundance presents new challenges to our ability to exploit them.

They are like oil or diamonds, in their natural state or crude, have little value, however, when we process it acquires enormous potential for the organization, as a finished product with a defined objective. Because as with oil, according to what you want to solve, it is that the specific product is created like plastic, fuel or lubricants.

With the digital explosion, now we are generating more data than ever, but that does not mean that all will be useful, yet.


They are divided into structured, which we have used for longer and are more used to analyze on a daily basis. Here with the data that is logically organized, you can know what data is and find them easily. As information from a database which has tables and fields with data types defined. Another example are the forms in which specific data is filled in each field.


As for the data that has no structure, they are primarily free text, meaning you can capture anything on them, so further analysis is more complicated. Examples are videos, photographs and texts, which you can get lots of information from.

One limitation to get value from data has been the ability to process them in a timely manner to make decisions with them.


To use them better, consider their qualities.

The first is the volume, the amount of data being generated today is immense and every day new data are created. To get an idea, in the digital world, there are:

  • More than 4,000 million email accounts.
  • 1,440 million Facebook accounts.
  • 500 million Twitter accounts.

That many accounts gives us an idea of the magnitude of big data being created and accumulating daily, only on those 3 digital channels.

Variety is the second in this the different kinds of big data are grouped, with videos and photographs, which can analyze both the content and the name and description, the most representative. The same can be done with free-text comments, opinions and recommendations of users.

Speed is a quality that makes us lose the opportunity to draw conclusions, because with the new data, you have to discriminate to decide how many and to what extent we can use it to make decisions that help us anticipate.

An example is the Youtube metric that every minute up to 100 hours of video are uploaded to the platform.

This is like newspapers, we must have the information ready to be consumed on time today because, although very good, if is from the previous is day no longer useful.

Finally, veracity, you must be able to select the content that really adds value to the business and be able to distinguish it, not only by the confidence we can have in the data but also for its usefulness for our purposes.

Another challenge we face is the ability to identify which big data serves me from all the amount available, and what really enriches my results.

Democratization of Internet

Internet access and ease of writing and publishing online content without restrictions, have made one of its biggest advantages the fact that anyone can post, but at the same time, everyone being able to do this is one of its weaknesses. Therefore, the ability to distinguish and use data that are useful from that that is not, it is essential to develop a marketing strategy based on facts.

How much is Big Data?

For data to be considered as massive, beyond huge amounts, this classification applies when we lose the ability to analyze them and use it, if we only store it, and do not get benefits from it; Even if we can analyze it but takes too long to get results, we have Big Data.


To get results, the data must be analyzed using analytical models or advanced statistics. This allows us to find hidden patterns that if using a manual analysis could not be detected.

From the term analytics, is important to clarify a distinction, because it is used interchangeably to refer to reports, that show only the past as well as the use of predictive models that show us what will happen in the future.

Certainly to make decisions on the results of analytical models we use reports. The distinction lies in that the traditional reports show only what has already happened, allowing us to make reactive decisions, once something happened to correct it.

Using analytics we can be proactive, anticipate what might happen, to take preventive action, which gives us an advantage to take action.


To do any statistical analysis, history is required, the more is generally better, because it helps eliminate biases that affect the predict ability of the models. The past helps us to characterize the future.

With this analysis patterns in the data are detected, that is, facts which are repeated consistently over time and share features that allow us to identify it and point out. Trends are identified.

Trends show me where the market is going, how it is behaving the consumer, what he is are looking for, and how.


Once we have trends, evaluate the odds of that happening, the greater this, the more likely it is to occur. So I can evaluate my actions before executing to run only those marketing actions that can be most successful.

Decision making

The ultimate goal is to reach decisions based on facts or information.

An advantage of using analytics to support business decisions is that they allow us to be proactive, anticipate what will happen. It is as if our business were a car and we will driving with analytical information, always looking forward we would see what comes. This will allow us to take preventive action before negative events occur or risks materialize.


Usually the decisions are based on reports, they are built on past data of what has already happened and with that, reactive actions are taken. If our business was a car and we drive it only using information from the reports, it would be like going looking back, in the rear mirror. If we hit something, we would realize only once has happened and we could not help it.

If sales were declining, it would be detected with analytical, before it becomes a serious problem, if you were to identify it with reports, it may be too late to reverse the negative effects or much more expensive than doing it earlier.

Follow me on Twitter @garabujo77

Learn more

Transform your retail business with engaging brand experiences that strengthen your customer relationships. Check out IBM Retail Industry solutions at

Marketing Technologist

More stories

Combining Internal and External Data for Insights, Forecasting and Visualisation

Retailers and consumer products companies have done a lot of work to apply data science to key challenges around demand forecasting, store replenishment, and optimisation of product assortment. Many have used historical sales and product data to identify patterns and glean insights. Some have used proprietary software products. Others have created data lakes and performed […]

Continue reading

Malls of the future will harness tech to make tenants and shoppers happier

The rise of online shopping has not been kind to malls. Traffic is declining, jobs are dwindling, and many malls are disappearing altogether. By 2022, analysts estimate, one out of every four malls in the U.S. could be out of business. [1] But while more malls will certainly shut down in the years ahead, those […]

Continue reading

Join IBM at the Hermes Retail Week Supply Chain Awards

We are delighted to be a key sponsor of the 2017 Hermes Retail Week Supply Chain Awards on 28th September in London. These awards are all about recognizing and celebrating the retailers, suppliers or individuals that are re-inventing all aspects of the modern retail supply chain. We are hosting a roundtable discussion before the awards celebration […]

Continue reading