Build Data from your Knowledge
You are likely recording a lot of information with your CRM software. Are you using these data to their full potential? Do you have partial but meaningful information about several of your target HCPs?
Being able to formally integrate this knowledge into your segmentation process can dramatically improve the accuracy of segmentation and targeting.
This is one of our key field of expertise. Over the last decade, data engineering has become a crucial step to maximise the value of data sources. Broadly speaking, it consists of manipulating complex or unstructured data in order to extract meaningful indicators that simplify the original data while maintaining its core information, so that they can then be integrated into a machine learning algorithm (MLA).
This step is needed primarily because most data sources do not come in a ready-to-use format. Think for instance about all the historical data contained in your CRM, all text notes that you have collected about HCPs, geographical data available at a very detailed level. It is very likely these data sources do contain very valuable information for targeting purposes, but you cannot simply use them as is as an input for a MLA.
Data Engineering offers a variety of tools to deal with this situation and to create routines to extract relevant features. These routines used to process the raw data can be automated, but deciding which routine to use and finding the best set of indicators to extract requires testing and the comparison of several potential solutions before a decision can be made. Therefore, data engineering is not a process that can be completely automated if one wants to make the most out of the original data.
Your Best Data Source: Your Field Force
Still unsure whether you actually do have access to such information? Think about the following: Your sales reps and medical liaisons know a lot about your target HCPs. They likely take notes from their visits, and they have information about HCPs that could benefit the segmentation if they were structured the right way.
Text mining is a powerful tool to extract meaningful summaries that can be incorporated into the segmentation process and even help formalize how the HCP visits are reported.
Furthermore, your field force can allow you to gather original information with a very limited increase in their burden. We can help design and test questionnaires with objective questions that literally take a few seconds to fill and can become a strong and reliable data source.
The In-House Intelligence Building Tree
Machine learning is a field where quality has a high priority over quantity. A large set of messy data will never replace a much smaller set of validated and reliable data. Therefore, during all our interventions, focus will primarily be put on preparing quality data. Beyond the advanced techniques described above to achieve this goal, there is a wider set of tools that can be applied on most data sources to prepare them adequately.
This tree is our summary of the various facets of data engineering. All branches in the tree help maximising the quality of your data. Some of them might seem more straightforward while other require more advanced data analytics tools. Still, each contribute to create a unique global data source and to put in place procedures to easily update existing data, and include new sources in the future.