Transforming Complex Data into Meaningful Outcomes
Data collection, analysis and insight dissemination can be difficult for organisations due to several challenges: The amount of data Cross-matching data sources The ability of the user to understand the data Software complexity The management and integration of data can be time-consuming and resource heavy and can cause even the most analytical people a headache. …
Data collection, analysis and insight dissemination can be difficult for organisations due to several challenges:
- The amount of data
- Cross-matching data sources
- The ability of the user to understand the data
- Software complexity
The management and integration of data can be time-consuming and resource heavy and can cause even the most analytical people a headache.
We believe that whilst granularity can be important, particularly in pharma, complex data must be shown in its simplest form to support both efficient and effective decision making.
3 key questions should be considered:
- What decisions (outcomes) need to be addressed?
- What data (inputs) are available?
- How should the data (inputs) be used in the forecast to support decision making?
Here, we outline some of the principles we follow to support clients in translating data into meaningful outputs for efficient decision making.
It’s crucial to understand what factors are most critical to your business decision. A model might include thousands of columns of data, but only two or three might be relevant to those examining the output and later making the decision.
When building a forecast model for a client recently, there was a need to not only identify over and under-performance of drug sales, but also the main attributable drivers. We created a multi-tiered model that offered the opportunity to drill down into the data sets and gain clearer insights. The client could explore ‘potential sales opportunities vs actual sales’ of both regions and accounts, and thereby diagnose performance drivers.
By building the model around the client’s goal, we were able to provide a dynamic solution with the ability for the client to explore the data in more detail as and when required. Key data and insights were easily visualised and with the use of automation, the client could choose what was most relevant for them at a given time.
The objective of the forecast model and the decisions it must facilitate must be clear to all stakeholders from the outset.
A model which is too complex, with extensive granular data requirements can lead to poor assumptions and difficulties in usability leading to errors and a lack of team confidence in the outputs.
The quality of insight obtained from any model to make your decision depends on the quality of the data you have available. If gathering the data is proving to be a challenge, we encourage clients to keep things simple.
A model that requires information which is difficult to obtain will often be underutilised or subject to guesswork, meaning the outcomes and subsequent decisions are of much less value to the business.
To truly take ownership of a model, it must be interactive and engaging. A relatively static model will only offer one interpretation of the dataset, limiting its use to the business. We believe, interactive features such as Artificial Intelligence (AI) enable the user to test and analyse the data.
We look for ways to add interactive elements to models, to help our clients make the most of the tools they have available and make more informed decisions.
In our experience, every team we work with likes to visualise their data in a slightly different way, to which there is no right or wrong approach. This is simply down to preferences. However, we all recognise that looking at thousands of rows of data, is not always that exciting for us or our audience!
We believe representing data in dynamic charts and animated videos can add greater depth and dimension to model outputs. We have seen eyes light up, people rub their hands, some even stand up with excitement as they see the environment, they are forecasting change at the touch of a button.
A simple 2×2 matrix, for instance, can map out high vs low sales and high vs low potential in a way that is very easy to understand, or temporal change added as another dimension without increasing complexity. There are many other features that can be added to communicate your data in a much more meaningful way depending on your stakeholders.
These four principles are pivotal to the model design process, ensuring we capture as much information as possible in a way that’s simple to complete and easy to digest. By prioritising the data and insights that are relevant to the organisation’s decision-making, and clearly defining the model’s interactive possibilities, we can create genuinely useful tools that add value to our clients and help them to communicate their complex data in a meaningful way.