Big Data needs Big Relevance

Focus on the results you get after you analyze

It is easy for analysis teams to focus on the wrong things, they can think too much about technology, models or mathematical curiosity. Even the best intentioned teams can slip into a non-relevant operational mode, if they don’t pay constant attention to the business stakeholders at all stages in their daily work. After all, it’s the business stakeholders and decision makers who set the goals and pay the salaries.

The reasons for performing any analysis need to be well understood by the whole team, before they start the analysis. They must have a clear picture of the Why, this allows an understanding of the benefits the results should generate. This understanding needs to be maintained through each part of the analysis process, past the point where the results are delivered to the business.

Relevance from Awareness

As an example; lets imagine a well engineered analysis process for targeted marketing campaigns. This process should be an ongoing engagement, in particular it should be a cycle. Remember that targeted marketing is an opportunity to have a relevant conversation with your customer. Each stage of this conversation needs to take the previous parts into account.

The cycle starts with a supply of data, these data could be from multiple sources, structured or not, it does not matter. These data are worked-on and models applied to them, the result is the target list.

This target list is delivered to a delivery process and content management system that will package each target with a piece of content, and deliver this custom package to each target contact.

Maintain the Awareness

The danger is that this analysis engagement will end here, and the data scientists will sit and wait for the next analysis job. Consider if the business operates its targeted marketing as a cyclical process, but the Big Data team operates on a per-Analysis basis; how massive is the potential loss of feedback? The lost opportunity for improvement?

The way it should continue, is that the results of the campaign are used as the source data for the next cycle of the analysis. The Big Data team should be actively aware of the performance of the analysis, where it failed and where it succeeded, as this intelligence is pure gold. They can modify their models based on what they learned, how their understanding changed.

Continuous Improvement

This way, each time the cycle operates, it has the opportunity to get better, and as the team is solidly aware of the business goals and reasons, they have the opportunity to get better also.

Improving analysis based on observations within your own business operations, is simply a use of your own assets. After all, it’s your data and your business paid for it, they should get the benefit from it.

Analysts and data scientists get better at their jobs if they operate this way, as their operational modes will become more synchronized with the business goals and targets.

Next Steps – Automation

When an analysis operation is built as a cyclical process, you have the opportunity to take it one step further. Here in Fonecta we routinely automate marketing processes that learn from each run of the cycle. Each message delivered has automatically taken previous ones into account. This constant refinement can dramatically improve the results of a marketing dialog.