What Big Data could bring to your organisation or your community?

Among the most cited keywords in recent IT related articles, the Big Data concept is not always used as appropriate, hence many people don’t really understand what it is and how their organisation could benefit from it. Let’s clarify what the real challenges are.

Cave exploringCollect data for a clear purpose to avoid blind exploration

All businesses collect data about their customers, members, sales, web traffic and e-campaigns. If you combine all this information to structure it in order to analyse it as a whole, the volume may become a challenge and the exploration could take a lot of your time. The methods and tools supplied with Big Data can help organisations to set up efficient data collection and analysis systems.

However, this might not be the most important thing for your business, you should focus on the purpose of this endeavour first. Identifying the most discriminant and pertinent data to answer the questions which are important for your business is key.

Knowledge brings power and leads to efficiency

The more you know about your customers, members and competitors, the better you will be able to anticipate changes. Big Data frameworks allow organisations to learn about what’s happening from the real data. For big companies, this will most likely be implemented with ongoing monitoring, for smaller organisations it could be done with repeated yearly analysis or one shot analysis according to the budget allocated for such investigations.

Smaller organisations and especially non profit communities will have more difficulties to assign budget dedicated to Big Data analysis than big companies, but they will be much prompter to react once the results have been released.

Big Data analysis is based on complex algorithms

The data analysis complexity is very dependent on the data type and the objectives of the analysis.

In the case of e-commerce, a common purpose can be identified for most organisations: Increase sales. Moreover, the content (products) is easily categorised using simple description items (product family, type, colour, price, sales volume…)  Models have been developed to address this question and many easy to plug components may be found on the market – providing even small organisations with tools that deliver reliable results for a fair budget.

The case of text data is very different and will require much more time and attention from the entire team in charge of the analysis because the content structure is not based on simple characters. News and publishing organisations and scientific or technical communities are facing a much more complex situation when they want to build their content data model.

From analysis to recommendation

Big Data analysis is very often used for business intelligence purposes and increasingly to feed recommendation systems. Content suggestion has become very popular in the field of e-commerce (Amazon) and video repository websites (Netflix) and is used frequently by News (BBC News) and publishing houses to suggest content to their customers.

The computing of these recommendations may be based on any or all of the following data:

  • Customers’ or members’ profile for a non profit community, e.g. country, age, gender, history, engagement…
  • Web traffic logs collecting anonymous and authenticated visitors behaviour data.
  • Website content data model
  • Other individual data such as engagement in external groups (LinkedIn, Google+, Facebook…)

To learn more about recommendation systems

Before you start

  • Start with a clearly defined business question you very much need to answer so that the return on investment will be easier to measure.
  • Make sure your organisation’s management supports the initiative and allocates sufficient budget to run it. This could be a long and costly project.
  • Take your time to find the right partner who can bring with them the experience to facilitate the system build and the expertise to implement the appropriate tools.

 

 

 

 

 

 

 

 

Leave a Reply