3 principles of integrated forecasting
Nikhil Gokhale, business planning manager at Actelion, and Robert Sigmund, Actelion’s director of commercial analytics, on the three principles of integrated forecasting modelsBy Mar 20, 2012 on
When Actelion launched its pulmonary arterial hypertension drug Tracleer in 2001, the most optimistic internal forecasts projected that 8,000 patients would benefit from the product globally. Today, more than 40,000 patients use Tracleer, and that number is growing by the year.
Which is great news for Actelion but bad news for their forecasting model.
In retrospect, a number of flaws became apparent. For instance, Actelion’s original forecasting model extended only five years, so it shed little light on a long-range plan. In addition, the model had no scenario building capabilities, and since an external company owned the tool, making changes cost Actelion money. Those changes then had to be repeated in every country where Actelion had a presence, and each year new personnel needed training to utilize the mode.
“If you look at the model, there were so many unknowns going in,” Nikhil Gokhale, business planning manager at Actelion, said at eyeforpharma’s Pharma Forecasting Excellence conference in Berlin. “No one was really aware of the prevalence or the incidence. Mortality rates were unknown … No one was aware of market shares, penetrations, and lastly, the combination rates [were] a complete guess.”
Of course, the most thorough research and rational plans often look misguided five or ten years down the road. The art of forecasting is to respond to the past and build its lessons into future models. This is how Actelion did just that.
1. Minimize the unknown
The company highlighted three principles to serve as bedrocks of a new, integrated forecasting model. The first was to minimize the unknown. As Galileo, the Italian polymath, once said, “Measure the measurable and make the immeasurable measurable.” For Actelion, this meant using measurable parameters obtained by commercial analytics and primary competitive intelligence.
2. Connect the dots
The second principle was to connect the dots. Of those parameters that remained hard to quantify—prevalence, sales, patients—it was important to connect them to historical data so that the team could develop market understanding and again minimize unknowns. Likewise, forward-looking outputs should be linked to current market parameters to enable market understanding.
3. Keep it simple
As William of Ockham, a Scottish philosopher, phrased it, “Pluralitas non est ponenda sine necessitate.” That translates roughly as ‘Entities are not to be multiplied without necessity.’
“If it’s simple, it’s probably right,” said Robert Sigmund, director of commercial analyticsat Actelion and Gokhale’s co-presenter at eyeforpharma’s conference. “So don’t have five variables, 20 variables, when you can have two or one variable.” Keeping it simple also makes it easier for senior management and local operating companies to understand and act upon forecasts.
In the spirit of simplicity, Actelion based its new model on prevalence rather than incidence. The old model attempted to estimate incidence by following patients across different lines of therapy, even though forecasters lacked market research data that signified which specific lines patients tracked. In the new model, forecasters take prevalence and apply it to the percentage of patients treated with advanced therapy.
Actelion’s product is only applicable to patients in the first line of therapy, so forecasters then look at what percentage of patients treated with advance therapy fall into that line. From there, they winnow down the field based on competitor profiles and launches and current market shares, plus revenue factors and price to arrive at sales in their specific niche.
“Again, the ‘keep it simple’ principle came in,” said Sigmund. “It would be nice to forecast the whole market size … but really we don’t want to make a forecast for Bayer or for Pfizer or Glaxo; we want to make a forecast for us.”
Another way Actelion keeps it simple is by forecasting for the key markets and the key markets alone. Sigmund: “We don’t forecast for Kazakhstan or Ukraine or Portugal. We just forecast for the key markets, the EU-5, Japan and the US, and then we project upwards.” The team distinguishes its projection factor among different groups of countries mainly based on the company’s loss of exclusivity in those countries.
Integration and outcomes
One success of the model is that it brings far more departments into a contributing role and thus makes forecasting a more integrated function in the company.
For example, the prevalence statistics at the beginning of the model come from the epidemiology group and local operating companies. The advanced therapy data, the various lines of treatment information, and the competitor profiles and launch data come from global commercial analytics. Finally, the revenue factors like compliance and dose and price and exchange rates come from the business finance group.
The results, said Sigmund and Gokhale, have been positive. The new model has led to fewer unknowns festering in the forecasting process, and the involvement of key stakeholders throughout the process has enabled an easier flow. “The flow is quite simple,” said Sigmund, “so you can put that in one excel spreadsheet per market and that’s understandable for the affiliates, that’s understandable for the local operating companies, which is very important because this is our key business.”
It’s also easy to conduct planning with different patent expiry scenarios, as inputs are simple to modify and are linearly linked to outputs. Gokhale and Sigmund report that most affiliates have requested to be able use the model for their own planning, and senior management like the increased transparency, noting that before it really felt like “a black box,” said Sigmund.
For more on forecasting, join the sector’s other leaders atForecasting USA in October in Boston.
For an overview of eyeforpharma’s forecasting coverage, seeHighlights from eyeforpharma’s Forecasting coverage.