Complexity vs. Simplicity Part 2: The Forecast Model Balancing Act
A successful forecast model design, as our previous article explored, relies on four key ingredients: transparency, tailoring, flexibility and simplicity. However, getting the right amount of simplicity can be challenging when the other ingredients can drive complexity in a forecast model. Toeing the line between complexity and simplicity is a careful balancing act, which must …
A successful forecast model design, as our previous article explored, relies on four key ingredients: transparency, tailoring, flexibility and simplicity. However, getting the right amount of simplicity can be challenging when the other ingredients can drive complexity in a forecast model.
Toeing the line between complexity and simplicity is a careful balancing act, which must take into account the goals and concerns of all stakeholders involved. A senior executive focusing on top-level outputs is going to have a very different outlook compared to a business analyst responsible for inputting data to the model.
Understanding the issue
Pharmaceutical companies are, by their very nature, data-driven. Some more so than others; the sheer volume of variables in oncology, for example, makes a degree of complexity unavoidable when building a robust forecast model.
However, there’s a difference between complexity and unnecessary confusion. A forecast model cannot do everything for a business – it would be unworkable and, ultimately, unsustainable. Make it too complicated and you’ll face resistance from those responsible for inputting the data.
Information that global and regional decision-makers might think is simple to acquire may place a lot of strain on those completing the forecast. A model that’s time-consuming, confusing, or requires data they don’t have will either be filled out poorly or worse, not at all.
Meanwhile, individual countries may push for tailored models to suit their specific needs or market conditions. To prevent unnecessary considerations from creeping into the forecast results, it’s vital to assess whether these requests are valid.
Determining the remit
To be successful, a forecast model must be simple enough for those using and completing the tool, while still giving senior management the outputs they need. To determine where the balance should lie, it’s important to question the remit.
What functions of the business does the forecast model feed into? What outputs do these functions need to make accurate, informed decisions? By defining the remit, you can begin to identify whether features within the model are necessary additions or ‘nice-to-haves.’
This means that stakeholders might have to compromise on the forecast model they would receive in an ‘ideal world’, in return for receiving completed forecasts models with higher quality data.
Offsetting current and future needs
People have a tendency to lump current and future needs into one, but in fact, it’s possible to break down the features of a forecast model into several different phases over the lifecycle of a product. This can boost the simplicity of the model at each stage of the lifecycle, while contributing to the overall longevity of the design.
By gradually adding complexity into the model – either by adding on/off buttons for functionality, or creating multiple versions – you can allow users to familiarise themselves with the design. Rather than overwhelming people with the full set of features, they can be rolled out over several years as users grow accustomed to them.
It’s worth noting that a small investment in training and communication can pay dividends when securing buy-in during this process. It ensures smooth implementation from day one, avoiding gaps in forecast data and addressing any issues early.
Creating a back-up plan
Finally, it’s important to remember that not everyone has the same level of capability where forecasting is concerned. A whole host of people could contribute to the forecast, including business analysts, brand leads, and those from marketing and finance departments. Complexity is a subjective term, and even the most seemingly simple models can cause confusion.
To this end, consider creating a backup plan that users struggling with the forecast model can complete as a last resort. This is often a template that captures key information; it won’t necessarily give the context and insights of a forecast, but it’s a good solution if you’re worried about the model being completed correctly.
Balancing complexity and simplicity is never clear-cut. Every organisation – and the stakeholders within it – is unique, meaning there will be a range of competing factors influencing the shape and depth of the forecast model. However, bear these points in mind when assessing stakeholders’ needs, technical capabilities and implementation, to achieve a clever model design that’s both workable and sustainable for everyone involved.