The Context
One of the challenges regularly faced in market research and forecasting is the alignment of data from market research with the creation of a robust model.
When building a forecast model, you gather primary and secondary data and develop assumptions that senior decision-makers can use to build future forecasts. This involves market research, allowing you to explore the market qualitatively and then quantify the gaps identified from both primary data and secondary data to guide the forecast outputs.
The challenge lies in alignment: market research and forecasting ‘fit’ often isn’t as seamless as it should be. Market Research isn’t always designed with a forecasting model in mind, resulting in a poor fit when you come to integrate the data.
This is a scenario that J + D Forecasting hear about time and time again from our clients. In this article, we will explain why it’s so important to tailor market research to forecast models, and how to overcome this challenge within your organisation.
The Problem
Often, pharmaceutical companies outsource the execution of market research and/or forecasting model development. These tend to be carried out by two separate providers, with little communication or transparency between the two. This is a big problem; unless the market research team know and understand the model requirements, it will be virtually impossible to ensure the market research and model are seamlessly integrated.
The assumptions within a forecast model can be measured in two ways: sensitivity and confidence. Sensitivity relates to how much the assumptions matter to the forecast – is the information critical to the forecast? Do changes make a big difference to the outcome?
Confidence, meanwhile, concerns how robust and well-structured the data itself is, which will determine how appropriately it can slot into the model. Unless the market research delivers robust data for the key sensitive assumptions, such as the source of business or uptake curve, there might be flaws or discrepancies in the resulting forecast.
The Consequences
Increasingly, new products are developed for very niche patient types and diseases, such as severe, uncontrolled patients with a specific biomarker. In this scenario, the forecast model is therefore designed for this specific context; if the market research questions are asked in a general context – for example, of all severe patients, what proportion would be prescribed this new product? – the results will not be compatible with the model.
At worst, this would mean the data cannot be used, as it would lead to inaccurate forecasts. At best, it will generate additional work for converting the data into a usable state. This ultimately results in wasted resources, delaying the forecast process and adding to the costs involved and therefore, general inefficiencies.
The assumptions within a forecast model can be measured in two ways: sensitivity and confidence. Sensitivity relates to how much the assumptions matter to the forecast – is the information critical to the forecast? Do changes make a big difference to the outcome?
Confidence, meanwhile, concerns how robust and well-structured the data itself is, which will determine how appropriately it can slot into the model. Unless the market research delivers robust data for the key sensitive assumptions, such as the source of business or uptake curve, there might be flaws or discrepancies in the resulting forecast.
The Solution
So how do you assess your assumptions and data, to ensure they are suitable for your forecast model? By conducting an assumption audit for sensitivity and confidence, you can identify assumptions that score ‘low’ or ‘high’ and plot them in a matrix. Any assumptions that have a high sensitivity and low confidence should then be addressed through secondary or primary market research.
To give you near-certainty that the two will align, consider outsourcing your market research and forecast model to a company that can manage the whole process. J+D take a strategic approach to forecasting, ensuring that the product’s outlook is both detailed and accurate. A holistic data management perspective optimises assumptions and promotes efficiency in resources, thus leading to a more robust forecasting tool that all stakeholders can have confidence in when making decisions.
Designing market research to better fit forecast models and company processes takes time and thought, but it’s a critical step to developing a robust forecasting model, lending the necessary visibility to launch and successfully managing a product.