Article

The power of proof-of-concept

A proof-of-concept (POC) phase is specifically designed to ensure that all the online/offline promotional channel and financial data available from sales, marketing, medical, market access, etc., is good enough to leverage A/IML in an ROI model. The phase has three main objectives:

1. To establish whether the data is robust, complete and relevant. Is the dataset reliable and comprehensive? Are there a sufficient number of data points? Has the data been collected in the right way? Can it be linked to its sources? And are the individual promotional channels sufficiently significant to add value when entered into the model?
2. To determine whether costs are available for every event. Is all the relevant financial information (i. e., individual cost of a meeting, advertisement, website, etc., and financial overview) available and linked to every promotional activity?
3. To reach an informed view. Make a confident go/no go decision to start the ROI modeling process, based on a thorough assessment of the data.

Proof-of-concept challenges and lessons learned

Completing a good proof-of-concept phase can be a challenge for many companies. Experience from local pharmaceutical affiliate projects offers some valuable pointers. Among key lessons learned are:

1. Pulling promotional and financial data takes more time than anticipated. Realistically, the fastest time to gather the data is about three months. In some cases, it can take six months or longer to gather available data for a given period of time. If the data has been collected correctly, it should be easy to pull it from the systems for analysis.
2. Unexpected gaps in the data collection process are identified. There is often uncertainty within companies as to which department should deliver the data, whether it has been stored or whether systems have been designed to even collect it. When collected, it is many times housed on different drives in the organization and in individual files, making it difficult to access and use. This is particularly the case where costs related to promotional channels are concerned.
3. Lack of clarity on channel definitions within the launch/brand team can cause confusion. In the absence of explicit definitions, channels are frequently subject to very different interpretations. For example, while some teams or individuals would understand a face-to-face conference to be a personal meeting with a healthcare professional (HCP), others would define it as a contact following an online address by a doctor. It is important for all involved parties to understand what is meant by a specific promotional channel.
4. The importance of consistent, continuous data collection is underestimated. Despite their increasing reliance on the data inputs, teams are not always fully aware of why collecting and storing detailed data on an ongoing basis is vital for ensuring it is truly valuable for analysis. Companies with a strong focus on communicating the essential need for data collection have seen marked improvements in their data quality and more importantly the insights to act upon.

Best practice ROI modeling using AI/ML

To maximize the benefits of using AI/ML modeling and achieve the desired insights, structured, consistent, accurate collection of promotional and financial data in cross-departmental alignment is essential. Before starting any ROI optimization project, following the learnings of the many projects we executed, we recommend the following actions:

1. Create a cross-functional team to ensure speed of data retrieval and alignment on input. Ensure the project team comprises members from different departments (finance, sales, marketing, market access, medical). Implementing a cross-departmental project team allows a clear, aligned overview of what promotional and financial data is available and how it is structured and defined. This ensures clarity, completeness and speed and avoids internal frustrations.
2. Design the campaign with the goal in mind. Before every campaign, consider exactly what it is intended to achieve. What variables are to be measured? How are these going to be collected? In other words, design the campaign with the end goal in mind so that everything is set up to optimally collect, analyze and learn from that event.
3. Ensure third-party agreement on data sharing. For any campaign that involves a third party, it is important to understand and/or agree whether the promotional channel data can and will be shared. This avoids the potential loss of valuable information that can and should be entered into the model.
4. Collect, analyze, experiment and learn – continuously! Executing promotional channels for events in a fixed process of set cycles every year, no longer holds good. Today, success is all about continuously collecting, processing, analyzing, experimenting and learning.Every single day an HCP engages with a company, providing an important opportunity to gain valuable information about what they want and what they need – information that requires immediate and continuous processing and monitoring. This is the true value of multi-channel management as well as AI/ML, to gain direct current insights to act upon.

Having the most advanced promotional optimization model and the best AI/ML abilities available will only be of benefit if the data is robust, complete and fit-for-purpose. A proof-of-concept phase unmistakably aids a vital understanding of readiness to utilize AI/ML modeling and derive the powerful insights it can deliver into optimal ROI.