Pharma forecasting: To offshore or not to offshore?
*Peter Mansell examines the business arguments for centralized versus departmental or regional forecasting*
By on Apr 5, 2011Peter Mansell examines the business arguments for centralized versus departmental or regional forecasting
With pharmaceutical companies under pressure to achieve cost-efficiencies across the board, it seems counterintuitive to put the squeeze on forecasting, a function that ideally should be intrinsic to optimal use of corporate resources and capabilities.
How companies arrange their forecasting, though, inevitably reflects to some degree the broader context in which they operate.
That environment is one of increasing complexity, from the products industry develops through to the ever-widening span of markets they are sold in and the numerous stakeholders who determine conditions for pricing, reimbursement, and access.
For Alec Finney, principal at Rivershill Consultancy, the issue cannot be boiled down to a simple choice between centralized or departmental-level forecasting.
First, he stresses, there must be a distinction between agreeing the qualitative assumptions on which forecasts are based and making the forecast per se.
Crucially, Finney adds, these underlying assumptions need to reflect consensus across all business areasincluding, for example, marketing, clinical, finance, and supply chainbefore the forecast moves to the modelling stage.
Creating a forecast, on the other hand, can be an independent function, one that is supra-departmental and perhaps one step further removed from the kind of internal politics that tends to muddy both assumptions and outputs.
I know lots of organizations, even outside pharma, are now thinking of putting that group outside the management structure and reporting to the CEO, Finney says.
There are risks attached, though. One is that they become a bit detached and a bit elitist, says Finney.
What youve got to do to fix that is make sure these people have a commercial background; theyre not just skilled modellers. They can talk to people who are there already and have some credibility.
Independent forecasting groups
When this arrangement works, though, it works very well, Finney believes.
The independent forecasting group can produce a most-likely forecastin other words, their best 50/50 bet on how things may turn out.
This is then delivered to the various parts of the business, which, rather than coming up with their own forecast, will shape the most-likely model to their needs.
At that level, then, it is less about forecasting than drawing up departmental plans and budgets, Finney suggests.
The supply chain, for example, may decide it cannot live with a 50/50 forecast as the manufacturing plan will not withstand that level of risk, and will reflect this; financial plans may have some element of target-setting, and so on.
Plans generally manage risk and manage expectations by understanding that no forecast is going to be completely right.
In all of this, though, forecasters need to make absolutely certain the forecast speaks clearly to senior management, Finney stresses; hence, the importance of maintaining independence while avoiding the ivory-tower mentality.
If youre presenting, say, to a portfolio approval committee, you should present the forecast with the assumptions, the model, and the associated uncertainties, according to Finney.
And if you paint that picture, and give them the numbers last, they will follow the story.
But if you show them the numbers first, Finney warns, some will be happy with the numbers and some will not. What can follow is a sterile debate on what the right number should be.
Once again, this underlines how critical it is first to agree the underlying assumptions across all departments, as well as the mechanisms by which those assumptions can be changed.
If you tell [senior management] the story of the assumptions and how you got there, Finney says, then they have to challenge your logic rather than just challenging the numbers.
(For more from Alex Finney, see Unlocking the hidden value of forecasting.)
Agreeing assumptions
Robert Siegmund, director of global commercial analytics at Actelion Pharmaceuticals, also sees agreeing assumptions as the critical step in the forecasting process.
Growing environmental complexity is all the more reason to ensure those assumptions are informed and reviewed by a whole range of stakeholders, including key affiliates in the US, the EU5 countries, and Japan.
There are always tensions around assumptions, such as pressure from the financial markets to put out an optimal figure or the temptation for affiliates to rein in near-term forecasts if their bonuses depend on achieving a medium-term target, he observes.
Once that bridge is crossed, the forecasting itself is a relatively straightforward.
The mathematics of forecasting are actually very simple, although people sometimes want to make them appear complicated to justify their own agenda, Siegmund argues.
It doesnt make sense if lots of different people are responsible for the actual technical steps to put the forecast together.
Actelion generates its short-term forecasts through the finance department and its long-term figures at company headquarters, where the responsible person attached to Siegmunds team works closely with marketing in particular.
But then thats not really the critical thing, because the technical steps are just Excel or basically multiplying sets of numbers, he adds.
Web-based forecasting platforms
While Actelion is still reliant on Excel, some companies have switched to a Web-based forecasting platform that enables stakeholders to log in from different locations and contribute their own perspectives.
You dont need any software, Siegmund notes. So you can just do it in Internet Explorer. And you cannot change any annex of the model. So its all about the assumptions.
This approach can be especially useful for in-line products marketed across various geographies, Siegmund suggests.
With drugs that are a long way from the market, where there may be only three or four people working on the forecast, there is less need for Web-based input.
It is still necessary, though, to have some form of central forecasting competence or department, someone responsible for facilitating the generation of the numbers, Siegmund emphasizes.
Ultimately, he observes, the responsibility for the numbers themselves lies with whoever is in charge of marketing the product.
But someone still has to coordinate the forecasting process and, crucially, smooth the path to agreement on assumptions.
That involves taking a range of perspectives into account.
Manufacturing has to guess the cost of goods, says Siegmund. Then marketing has to come up with assumptions on peak market share and peak class penetration. Clinical has to think about the treatment duration and the dosage. And health economics has to decide what is an achievable price range.
All of these people have to work together, Siegmund notes, but, in the end, there has to be one person who facilitates the agreement on the assumptions and actually handles the technical steps, whether through a Web solution or otherwise.
One forecast for all
According to Arnaud Grunwald, director of global market analytics and forecasting for Novartis, online forecasting has been a buzzword in the industry for a number of years.
Like Siegmund, Grunwald sees this approach being of most value not to the forecast itself but the processes that inform it.
Its more a communication and knowledge-management tool, he says.
In Grunwalds view, industry could now move quite easily to wholesale online forecasting.
The problem in the past tended to be inadequate technology; the Web was still in its 1.0 phase and online platforms were not up to scratch.
In fact, Novartis is rolling out a central portal for forecasting at global level.
But, as Grunwald points out, we cannot force the countries to participate. We know that some of them will be very willing to participate and that some others will not at all.
As he acknowledges, the ultimate goal of forecasting would be one forecast for all.
However, politics, strategy, and the budgetary cycle tend to get in the way.
At Novartis, for example, Indonesia does its own forecast, but at headquarters level we have absolutely no interest in looking at Indonesia, Grunwald comments.
Strategically, it would be difficult to cover everything.
We would still have individuals doing the high-level business cases for senior management, but at the same time wed need to have a manufacturing forecast for Indonesia, he says.
One fundamental stumbling block, Grunwald explains, is that forecasts are essentially a translation of business units strategies, objectives, targets, and tactics.
I do not see how a region as powerful as the US or Europe would want to have their forecast done by global, he says. Because then they dont have a grip on their budgets, theyre not able to allocate their resources the way they want.
In theory, centralized forecasting should be possible in an organization of Novartis size, Grunwald adds.
But it would require a completely independent forecasting resource that understood all the intricacies of the business.
For all that, Novartis has responded to the industrys productivity challenges by moving slowly towards a central hub model for forecasting.
The company has a forecasting team at Novartis Global Operations, its center of excellence in Hyderabad, India, that handles modeling, scenario creation, and updating slides for the global pharmaceutical business.
Novartis has retained a small team in Basel to be the face of forecasting, as Grunwald puts it.
You always need someone to have these one-to-one conversations with the brand teams, with pricing, with all of the stakeholders, he says.
(For more from Arnaud Grunwald, see Forecasting: "No such thing as 'rest of the world' anymore" and Forecasting for pharmerging markets.)
Offshoring forecasts
It did take a number of years to understand what the best working model was for offshoring.
Initially, all of the analysts were in India, barring one forecasting director at the Basel headquarters.
But this did not provide sufficient balance to ensure all the necessary information was collected, expectations were managed, and deliverables came through on time.
Teething problems aside, the success of the Hyderabad operation meant there was a strong case for extending it to support regional and country-level forecasting.
This is an ongoing process, with forecasts for whole continents and new product lines shifting out to the Indian hub.
I can see that in the future we will have a big team in Hyderabad, people who are really well trained and have really good communications skills, interacting directly with the forecast owners that sit onshore, Grunwald predicts.
The more countries are supported from this central hub, the more Novartis will reap the rewards in terms of global alignment.
These people are sitting in the same room, he says. So the alignment will not be a process any more, it will be something that occurs naturally on a daily basis.
Siegmund is more sceptical about the benefits of offshoring or outsourcing: I think this was all the rage about three years ago but I see that its already going quickly out of fashion.
For the technical steps, it may have some validity but Siegmund also believes the focus should be more on simplifying the forecasting model: because in the end, its just multiplication.
In Siegmunds view, its maybe because some of these models were made so complicated that you had to then outsource the actual process of generating the forecast. If you have a model that is fairly straightforward and transparent, you can actually do all the stats yourself.
He cites pharma forecasting guru Gary Johnson, who observed that there is no return on complexity: You can make a model more and more complex, but in the end you dont gain in precision.
For all the latest trends in forecasting, join the industrys key players at Pharma Forecasting Excellence Europe on June 14 and 15 in Berlin.