Implementing Big Data Analytics In A Telco Marketing Department

Implementing Big Data Analytics In A Telco Marketing Department: 7+1 Steps To Help You Suceed

A Big Data Analytics project in a Telco marketing department is a complex task. It has enormous benefits potential but also a high chance of not obtaining the promised ROI, dedicating a lot of resources to the collection and integration of data and even reaching a freeze situation due to the complexity of analysis tasks.

7+1 Steps That Will Help You Succeed In Your Analysis Project

1. Define the business case Define business cases with business objectives that have an impact on the company’s results. Think big: churn reduction, Arpu improvement… of course there are many smaller objectives such as, encouraging the use of services and improving customer satisfaction. Yet in the end all of them lead to greater objectives: service cancellation reduction and increasing customer Arpu. Big data analytics projects ought to help these objectives so that the investment is worthwhile.

Therefore, it’s the Telco marketing managers who must lead those projects clearly defining the objectives to be achieved. Though underlying technology is important, the ultimate goal of an analytics project for marketing must not be overlooked: improving business objectives.

2. Locate helpful data sources: Collaboration with all areas of the company is a must. Without it success is unlikely. Someone who knows the data that is hidden in other areas can help to accelerate the project and improve results.

For example it is possible to obtain information from devices connected in a home various times a day. From this we can infer the more frequently used devices so we can judge the possible upgrades to offer the household. It is also possible to learn of the service usage experience taken from data of the network elements and collected from customer end devices.

Furthermore, external data can be incorporated. Many geomarketing or sociodemographic information companies can provide data that will enrich the data the operator has of its customers. And of course, you can also incorporate data from social media

3. Data preparation: If there is a critical point in a big data analytics project in Telco, this is it.This is where the greatest part of the success or failure of a project lies, and where 80% of time and resources are consumed.This is owing to the great disparity of data that comes from different systems and areas of an operator which makes it the most time and resource consuming phase. It is also where the biggest problems will arise, due to data quality problems: duplication of documents, wrongly typed customers, disparity of criteria when storing the same information in different systems, customers with erroneously provisioned products…

Have resources at hand for this step, lean on other areas and suppliers with knowledge and experience in data integration in telecommunications.It might even be a good idea to launch a data quality project to address corrections in many systems and make this step less complex.

4. Construction of analytic models: Once the dataset is obtained the stage of the analytical model construction has been reached. This is not the most critical task despite its specificity. No doubt we need to find the best algorithm, but as in any project, cost and benefit must be set on a balance. It’s possible to obtain results in some use cases with analytical models that can be found online, companies which have specialised in certain use cases and even companies that are able to build and evaluate hundreds of models online from a dataset.

Remember it is possible to start a project with algorithms of less precision,yet these begin to provide benefits and improve accuracy in successive and recurring steps.

5. Obtaining and validating insights: Keep in mind that the algorithm with the best precision is not necessarily the best one for the business case being analysed. The output information it generates is not needed.This phase requires great involvement from the marketing manager who needs to review the results and information the model is obtaining before deciding to validate it.

For example an algorithm may be very accurate when calculating the churn score associated with a client and yet , not offer useful information about the reasons that cause such scores. In such case two different algorithms for each output or another algorithm with less accuracy but better output information and greater end benefit, could be used.

6. Definition of actions to be carried out with the obtained results: As we said at the beginning it is important not to lose sight of the purpose of the big data analytics project for marketing. It is necessary to define what actions can be carried out with the results obtained. Some examples of these are ; retention campaigns with valued customers at risk of churn, up selling campaigns with a better offer to be made to a customer via a certain channel. a preventive call to a customer from customer care in case of problems with the use of the service, recommendation of better mobiles rates based on consumption or grouping channels into packages based on channel viewings.

Have resources at hand for this step, lean on other areas and suppliers with knowledge and experience in data integration in telecommunications.It might even be a good idea to launch a data quality project to address corrections in many systems and make this step less complex.

7. Systems and processes integration: Finally the results should not be left as recommendations or in lists which can be manually incorporated in the processes. The automation and integration of these steps is necessary. This is the only way to guarantee that the big data analytics project is generating the expected benefits for the company.Taking us from the exploration of data to an improvement in business results.

Have resources at hand for this step, lean on other areas and suppliers with knowledge and experience in data integration in telecommunications.It might even be a good idea to launch a data quality project to address corrections in many systems and make this step less complex.

8. Bonus: be lean!: Learn from the lean methods used and perfected by startups,allowing them to grow rapidly; build, measure, learn and constantly iterate. Carry out tests modify the input data, the analysis algorithms and even the definition of the business case and the actions to be taken with the results. Do this continuously and be agile. In this way better, faster, fewer risks and lower costs results are obtained.

Have resources at hand for this step, lean on other areas and suppliers with knowledge and experience in data integration in telecommunications.It might even be a good idea to launch a data quality project to address corrections in many systems and make this step less complex.

Conclusion

Prepare the project carefully always with the ultimate goal in mind; to improve business. Technology has matured enough it is time for big data analytics projects to be led by marketing and to focus on solving business problems.

Do not underestimate the work behind it. Above all prepare to focus a lot of effort on some tasks: data preparation is still the most important stage, not the search for the best algorithm. Have sufficient resources at hand for this task, this will highly favour the success of the project.

Finally define the tasks you will do with the results obtained from the process. Incorporating these results into the actions and processes of the operator is what will generate the improvements. Analytics provide insights, you have to convert those insights into €.