Predicting the Future The Business Case for Forecasting

Contributed by:

Kim Ribbink

NOTE: The content below contains the first few paragraphs of the printed article and the titles of the sidebars and boxes, if applicable.

By Kim Ribbink

Risk is an inherent part of the pharmaceutical business — risk of failure in clinical trials, risk that a product won’t achieve regulatory approval, risk that a product underperforms in the market, and risk of withdrawal after adverse reactions and the costly lawsuits that can follow. “Pressure from payers and competitors has reduced profit margins, leaving less room for error when making decisions about capital investments, promotional resources, and licensing deals,” says Todd Johnson, director of forecasting at MattsonJack. It is critical, therefore, that companies find ways to mitigate those risks, and one way to do this is through well-conceived forecasting models. Consultants and specialists in the field of pharmaceutical forecasting recommend that companies dedicate resources to developing an in-depth understanding of market drivers and influencers and they embed forecasting throughout the organization so that the function incorporates all of the fundamental aspects of the business. “Every quarter, CEOs go to the Street to announce how their companies are doing, but the forecasters who are focused on products often don’t have a grasp of the big picture, which is that everything leads to one question: how is an organization going to do in its overall profits?” says Sanjiv Sharma, VP of commercial affairs at NicOx. Experts say integration is key to the success of the forecasting function. “Having a uniform set of assumptions and consistent information across functions helps to make sure that everyone ‘rows in the same direction’,” says John Brown, senior manager, forecasting, at Daiichi Sankyo. “Being able to decipher the leading drivers from a variety of functional areas and consolidating the data into a forecast are what make this such a challenging job.” The good news for the function, and ultimately the industry, is that forecasting is receiving greater recognition from management at most pharma companies. “When I first started, we were, metaphorically speaking, relegated to the basement offices providing short-term forecasts,” says Jay Glasscott, senior manager, commercial assessment, global business intelligence and commercial operations, at Shire Pharmaceuticals. “Today, forecasters are being given a seat at the table; we are supporting and participating with marketing, strategic planning, managed care, and business development teams.” Forecasting Stages In today’s market conditions, forecasting around strategic planning has become all the more critical. “Now more than ever, we need to factor in external forces that are driving the market,” Mr. Brown says. “Ex-trend events such as economic downturns, patent challenges, competitive strategies, new entrants, and salesforce downsizing all combine to make for a very dynamic forecasting scheme.” The rising importance of customers such as payers means forecasters need to consider the future from multiple perspectives, says Nick Guthrie, forecasting capability lead, global marketing, at AstraZeneca. As a strategic planning tool, experts say forecasting needs to reflect the business, be part of the input, and communicate the right information with the right level of confidence. “In early-stage research, for example, there’s a lot of uncertainty surrounding the clinical properties of drugs in development and also the target patient population, so the forecast tools need to reflect this uncertainty, provide sensitivity to a range of clinical inputs, identify probabilities of commercial success, and so forth,” says Doug Willson, senior VP, advanced methods, at GfK Healthcare. With limited resources, forecasting enables companies to make critical evaluations about potential products. “New product teams are working closely with their clinical counterparts to explore compounds’ marketing and sales potential and using forecasts to optimally strategize clinical development,” says John Yi, Ph.D., forecasting modeling specialist at TVG Marketing Research & Consulting. Ken Sobel, senior VP, TNSfyi, says a forecast must do more than just accurately estimate the size of a new opportunity. “A good forecast also will help the pharma brand team actively manage the profitability of the new product launch,” he says. The relative accuracy of the forecast, however, usually depends upon the complexity of the situation, so in the case of pharmaceuticals, this depends on whether a product is already on the market, a new product, or a new concept. “Forecasting the revenue stream of a stable product is relatively accurate because a trajectory has been set up and the only things to understand are the events that can change the direction of the trajectory,” Mr. Sharma says. “With a new product — whether developed internally or in-licensed — developing a forecasting model to determine whether the product will recover the cost of the investment money is much more complex. And when it’s a new class of drugs or new mechanism of action, the process is even more complex.” Business development is another key area for forecasting; analytical tools and methods can be used to structure licensing deals in a way that limits risk, Mr. Johnson says. Because more companies are chasing fewer late-stage opportunities, many organizations are looking to in-license early-stage products. “There’s a lot of uncertainty when evaluating products that early in development; the predictability error is high, as are the financial stakes,” Mr. Glasscott says. “At Shire, these forecasts receive a lot of visibility. Not only are they used for valuation purposes, but they are very useful to identify and quantify risks.” Mr. Glasscott says central to these types of forecasts is good integration of the data from a strong evaluation team, including experts from clinical, regulatory, scientific evaluation, marketing, managed care, competitive intelligence, and finance. “In one situation, the forecasts, based on input from the evaluation team and market research were 65% lower than the licensor’s expectations,” he says. “The opportunity was not pursued and later on, when the clinical data were released, the team’s insights were validated and the company’s value fell about 70% to the level our forecasts had projected.” Defining A Clear Process According to Princeton Brand Econometrics, the pharmaceutical industry has a poor history when it comes to forecasting. For example, the company has observed, nine out of 10 forecasts for newly launched brands differ from the in-market results by at least plus or minus 30%. Companies also consistently expend 40% of their details for in-line brands against situations that good forecasting would have shown had no chance of paying off. However, Mr. Willson says because there are a number of things that can happen in the development process, being off by a certain amount is the nature of the game; the goal is to minimize the large, costly mistakes as much as possible. The role of forecasting, experts say, is not to drive sales or development but to help companies make better decisions. “Forecasting is basically a numerical expression of a plan, whether it be a short-term forecast reflecting the operational plan or a long-term forecast reflecting the strategic plan,” Mr. Guthrie says. Mr. Johnson says while most companies have structured planning cycles with forecasting central to decision making, many of the details pertaining to tools and processes are frequently overlooked. Where the breakdown is most apparent is linking pipeline and in-line products. “Ideally, forecast models should be built with the future in mind so that there is no need to change methodologies when a new product launches,” he says. “And to populate and update the models there has to be a commitment to providing the requisite data.” For short-term situations, a combination of a strong statistical base and local market knowledge yields the best results, Mr. Guthrie says, while long-term forecasts require much more strategic thinking, with knowledge of market trends and events that could disrupt those trends. “For example, when AstraZeneca was looking at life-cycle management opportunities for Seroquel it was important to understand the scientific profile of the compound and how it might differ from its atypical antipsychotic competitors, but at the same time understand how the market view of atypical compounds was developing,” he says. “By bringing this knowledge together, AstraZeneca has developed a strong life-cycle management program for Seroquel that maximizes the benefits to patients and seeks to open up new opportunities for the company.” Companies that take a centralized approach to the forecasting function are in the strongest position, experts say. “A best practice is to standardize the information that comes out of the forecast to make it easily understood by management by using standard language,” Mr. Willson says. “This is important because the users of the forecasts aren’t forecasters.” Another best practice is to first assess the current market environment and determine what is driving it, Mr. Brown says. “Knowing how the forecast is used by your customers is crucial to integrating the process,” he says. “The second step is to gather data from various sources and determine how the information will help to produce meaningful forecasts.” For brand teams, this means looking at existing data about a compound and thinking about future needs, not just existing needs, Mr. Guthrie says. Before the forecasting process begins, experts recommend identifying other measures, marketing tools, and data sets that are central to the process. “In the forecasting process, it’s important to identify a common set of terms, definitions, and parameters that will have the same meaning across functions,” Dr. Yi says. “This will increase the usefulness of the forecast, and its applicability to other important processes — clinical development, marketing strategy, portfolio management, sales operations strategy, and risk management — is enhanced.” The forecast should be developed independently and without the bias of financial objectives, Mr. Glasscott warns. An independent forecast can establish a base case around how marketing and sales, supply chain, and finance develop their respective plans. “Several companies are now separating the forecasts from the financial plans and salesforce objectives,” he says. “For example, from a salesforce incentives perspective, the base case forecast is an extrapolation of past performance to predict future performance. Achievement of the base case means that expectations were met, but not necessarily an improvement over past performance. However, exceeding the base case forecast yields a higher bonus; a recognition of the creativity and drive that takes a sales team beyond the historical to achieve a new level of accomplishment.” Mr. Sobel says the best pharmaceutical companies think strategically about new product launches and aren’t focused on the narrow tasks of setting up the right processes, tools, and data. “The most progressive companies are embracing the concept that a strong forecasting process will empower them to develop profit-optimized plans focused on their new products and not similar products that other companies have recently launched,” he says. For Mr. Sharma, best practices can never be a cookie-cutter approach but rather must focus on understanding customer insights. “When it comes to developing customer insights there are two components: one is identifying who the customers are, and second, a clarity and focus on what we want to learn,” he says. “If these two data points are on target, then we can work to execute the required tactical pieces to get insightful answers.” F PharmaVOICE welcomes comments about this article. E-mail us at feedback@pharmavoice.com. Pharmaceutical companies, whether large or small, engage in a constant juggling act to ensure balance within the pipeline and profitability of products entering the market. Forecasting can be the foundation to ensuring all those needs are met. John Brown Daiichi Sankyo Today’s forecaster needs to be a good listener, able to discern the few really important influences, and understand the mechanism of cause and effect. Dr. John Yi TVG Ultimately, forecasting products early in development guides decisions that will impact the company’s future portfolio and shareholder value. Todd Johnson MattsonJack While forecasting needs can’t always be answered with the same approach, having a forecasting platform does serve as the foundation for most needs and helps to instill a sense of standardization that can be incorporated into the planning document. Ken Sobel TNSfyi In pharmaceutical forecasting, success comes from projecting both the likely future business outcome and the promotional support plan needed to achieve that outcome. Sanjiv Sharma NicOx The current forecasting best practices are implemented with a cookie-cutter approach, which doesn’t work; the problem is most people keep doing it without questioning its relevance or adapting to their specific situations. Doug Willson GfK Healthcare Alot of forecasting work is done to support valuations for licensing and these deals are essentially risk-sharing arrangements. The forecast should support the valuation and provide information on how variable the returns are, the drivers for success and failure, and so on. Jay Glasscott Shire Pharmaceuticals The forecasting field has taken great strides to incorporate newer methods and approaches to reduce error. Gone are the days of using just one forecast model. Nick Guthrie AstraZeneca The biggest challenge for brand teams as they develop their strategic thinking is keeping the future range of possibilities in mind while developing a single strategy and executing one plan. Experts on this topic John Brown. Senior Manager, Forecasting, Daiichi Sankyo Inc., part of Tokyo-based Daiichi Sankyo Co. Ltd., is a global pharmaceutical company with a central focus in thrombotic disorders, malignant neoplasm, diabetes mellitus, and autoimmune disorders. For more information, visit daiichisankyo-us.com. Jay Glasscott. Senior Manager, Commercial Assessment, Global Business Intelligence and Commercial Operations, Shire Pharmaceuticals, a global specialty biopharmaceutical company focused on meeting the needs of specialist physicians. For more information, visit shire.com. Nick Guthrie. Forecasting Capability Lead, Global Marketing, AstraZeneca, which is one of the world’s leading pharmaceutical companies focused on gastrointestinal, cardiovascular, neuroscience, respiratory, oncology, and infectious disease medicines. For more information, visit astrazeneca.com. Todd Johnson. Director of Forecasting, The MattsonJack Group, a Kantar Healthcare Company, which provides business analytics and strategic decision support for the pharmaceutical and healthcare industries. For more information, visit mattsonjack.com. Sanjiv Sharma. VP, Commercial Affairs, NicOx, a product-driven biopharma company dedicated to the development and future commercialization of investigational nitric oxide-donating drugs. For more information, visit nicox.com. Ken Sobel. Senior VP, TNSfyi, working with TNS Healthcare, a Kantar Healthcare Company, provides globally consistent solutions and custom advisory services to support product introductions; brand, treatment, and sales-performance optimization; and professional and DTC promotional tracking. For more information, visit tnsglobal.com/healthcare. Doug Willson. Senior VP, Advanced Methods, GfK Healthcare, a provider of integrated healthcare marketing research. For more information, visit gfk.com. John Yi, Ph.D. Forecasting Modeling Specialist, TVG Marketing Research & Consulting, and Assistant Professor, St. Joseph’s University; TVG provides an integrated, global view of the healthcare marketplace through marketing research, training, and consulting services. For more information, visit tvg-inc.com. a forecasting evolution In the past five to 10 years, forecasting for pharmaceutical products has gone through a second revolution, following an initial revitalization that occurred in the late 1980s and early 1990s. Initially, pharmaceutical forecasting relied on models that focused on market share. “In particular, in those dark ages of forecasting, the objective was to forecast the peak-year share of prescriptions, based on analyses of analog launches, or perhaps analog launches along with measurements or assumptions about efficacy and safety profiles of the new product relative to existing therapies,” says Ken Sobel, senior VP, TNSfyi. The first important improvement in forecasting was a move from prescription-based models to models that reconciled prescriptions and epidemiology data describing the size of the patient universe. “This was indeed a major advance, as far as it went, but forecasting models still depended on an evaluation and analysis of analog product launches,” he says. “If a marketer’s new product was similar enough to the analogs selected, then the forecast might work out accurately. If not, the results weren’t so helpful.” The more recent major revolution has seen forecasting move beyond dependence on just epidemiology and syndicated data and toward a more comprehensive view of a launch, including the ability to explicitly account for promotional and marketing expenditures, and how they can influence peak share and uptake curves. “These advances are critical because promotional spending is the largest element of cost for a new product’s launch,” Mr. Sobel says. “To manage the profitability of the launch, it is essential to know how much additional revenue will be generated by additional spending — and, more to the point of most launches these days, how much revenue would be foregone if promotional spending is reduced.” Mr. Sobel adds that when forecasts provide this sort of information on the relationships between promotional spending and eventual revenue, then forecasting becomes an invaluable tool for managing the profitability of a new product’s launch. Forecasting Practices and Data Integration The forecasting field has taken great strides in incorporating newer methods and approaches to reduce error. Gone are the days of using just one forecast model. At Shire, Jay Glasscott, senior manager, commercial assessment, global business intelligence and commercial operations, uses a combination of different forecasting models, including decomposition, structured and quantitative analogs, expert systems, and extrapolation. He says combining forecast models, sometimes referred to as triangulating, reduces error. In fact, recent studies at the University of Pennsylvania show that combining models reduces errors by more than 6%. “Integrating market-research data into a forecast is becoming more effective,” Mr. Glasscott says. “Decomposition models are designed to break the forecast down into its constituent parts. As a result, these models are good for incorporating market research into respective constituents.” (For more information, see the Forecasting Framework table on this page, which describes the major constituents of a decomposition forecast, according to Mr. Glasscott.) “Newer tools borrowed from econometrics include order of entry, based on the principles of Zipf’s law, which essentially says all product attributes being equal, first to market captures the lion’s share,” Mr. Glasscott says. “Again, using different approaches to derive peak market share, then combining or triangulating reduces error.” Mr. Glasscott says another useful forecasting tool is game theory. “One example is war games, which may be thought of as a qualitative exercise,” he says. “Game theory goes to the next level; the calculation of adoption curves is based on product messages and competitive counter-messages.” Mr. Glasscott adds that new econometric approaches have sharpened the development of algorithms that translate “intent to prescribe” from market research into market share. Forecasting Challenges and Market Research While pharmaceutical companies depend on a combination of data and market research for their decision making, projections can at times go awry when marrying the two. One such example lies in integrating primary market-research results with other secondary data sources, such as prescription or unit data. “The first thing that the forecaster will notice is that the market shares expressed in the forecast model might be very different from the shares reported in the preallocation phase of the market research,” says Todd Johnson, director of forecasting at MattsonJack. “This could merely be a random variation around the sample size. It could also result from targeting only the high prescribers in the market-research study, or there could be other systematic issues, such as overreporting of brand use by physicians who do not know how many of their prescriptions are converted to generics at the pharmacy.” John Brown, senior manager, forecasting, at Daiichi Sankyo, says a challenge with integrating market research into a forecasting strategy is determining the overstatement factors needed to adjust both physician and patient responses. “I find an enormous amount of effort is placed into measuring responses, while not nearly enough is known about quantifying the overstatement,” he says. “The adjustment factors appear to be the weak link in the chain, which are just as critical as the research itself. Attempts need to be made to ground the research to current results.” To alleviate this issue, experts recommend forecasters be involved in the development of the surveys and questionnaires. Sanjiv Sharma, VP, commercial affairs, at NicOx, believes that while it’s important to get a broad picture of what the data — internal and external — present, without drilling down to the details the forecast will be fundamentally flawed. He describes Pfizer’s success of Lipitor as proof of the value of uncovering the details. “Lipitor was the seventh statin on the market, and Pfizer brand executives understood that in addition to showing that their product was slightly more effective than existing products, they needed to change the yardstick of effectiveness, which made all other products even less effective” Mr. Sharma says. “All of a sudden every other product was not able to ‘target lipid goals’, so Lipitor became the de facto gold standard.” F Forecasting Support Model When forecasts provide information on the relationships between promotional spending and eventual revenue, then forecasting becomes an invaluable tool for managing the profitability of a new product’s launch. Ken Sobel TNSfyi Jay Glasscott Shire Pharmaceuticals New econometric approaches have sharpened the development of algorithms that translate “intent to prescribe” from market research into product market share within a therapeutic class. forecasting framework Market Analysis Targeted Market(s) Peak Market Share Adoption Curve Compliance Duration Pricing Future Events Jay Glasscott, senior manager, commercial assessment, global business intelligence and commercial operations, at Shire, uses the decomposition model, which he calls a forecasting framework, to incorporate market research studies, such as longitudinal data, physician and patient segmentation, discrete choice, message testing, pricing, and promotional analytics into the respective constituents. For example, he says under market analysis, longitudinal studies and patient flow models are good at achieving clearer insights on concurrent treatments and treatment algorithms. Source: Jay Glasscott, Shire Pharmaceuticals. For more information, visit shire.com. Sound Bites From The Field PharmaVOICE asked experts to outline the key criteria for developing accurate forecasting models to ensure that products in the pipeline fairly closely meet projected patient uptake. Patrick Brundage, Practice Leader for Life Sciences Analytics, marketRx, a Cognizant company, provides analytics and related software services to global life-sciences companies. For more information, visit cognizant.com. “Good forecasting encompasses the perspectives, influences, financial motivations, and anticipated uptake by what I call the ‘4Ps’ of the bio-pharma success equation: patients, physicians, payers, and pharmacies. There is a tendency to put too much emphasis on one of these components — usually physicians, sometimes patients — at the expense of developing a robust model that appropriately weights all four according to their relative impact on the ultimate prescribing and dispensing decision. In addition, there is sometimes a tendency to forecast in a vacuum without due consideration of the impact current and future competitors will have on a company’s brand.” Art Cook, Managing Principle, ZS Associates, a global management consulting company specializing in sales and marketing consulting, capability building, and outsourcing. For more information, visit zsassociates.com. “Patient uptake — the adoption curve — has been one of the most notorious variables to estimate in new product forecasting. Along the curve there are three categories of users: the innovators, the followers, and the laggards. Innovators will use a product first, then the followers, and finally the laggards. The key is to understand how many physicians fall into each category for the therapy area of interest. This enables the forecaster to more accurately determine the shape and speed of the new product’s adoption, thereby maximizing return on marketing investment and allowing the pipeline product to meet the projected uptake.” John Doyle, Dr. PH, Practice Leader, Market Access — U.S., Quintiles, a provider of professional expertise, market intelligence, and partnering solutions to meet the dynamic needs of the pharmaceutical, biotechnology, and healthcare industries. For more information, visit quintiles.com. “Forecasting models may be designed from a top-down or bottom-up approach. Generally, top-down models are anchored by comparable products’ historical sales data. Bottom-up, or patient-based, models are multifaceted, incorporating epidemiological, treatment, economic, and market factors. Ideally, the evidence substantiating each model variable is sourced and quality scored. For lifecycle planning, a forecast model should include the range of target indications and patient subpopulation-specific treatment algorithms, taking into account risk factors, and evaluating multiple future market scenarios that can serve as a tool to prioritize clinical development options. One goal of patient-based forecasting is to optimize the risk-benefit of a product by targeting accessible patient populations with the most pressing unmet medical need.” Jeff Hewitt, VP, Product and Portfolio Strategy, IMS Health, a provider of management consulting and market intelligence to the pharmaceutical and healthcare industries For more information, visit imshealth.com. “It’s important to define the forecast well, with regard to patient populations, geographies, launch timing, etc.; have clarity on the potential profile — efficacy, safety, etc. — of the pipeline product; factor in competition from both in-market products and products that may be launching in a similar timeframe; factor in uncertainty represented by multiple scenarios of how the future might unfold; and consider pricing and reimbursement issues for the product. Forecasters also need to use a solid basis of information including qualitative or quantitative market research, analogs of comparable products, and the judgment of experts.” Rob MacLennan, President, PharmAnalytics, a consulting firm specializing in designing patient-based and prescription-based sales forecasting models for clients in the pharmaceutical and biotechnology industries. For more information, visit pharmanalyticslive.com. “The most important start to developing a good forecasting model is to take some time to carefully plan out the model’s design. We begin by understanding what the client’s key model deliverables are or questions that will need to be addressed. From there we recommend focusing on defining the market in question — brands, patients to be included in the market, etc. — and then sketch out the patient flow of the market, usually starting with diagnosis and ending with drug treatment. It may be necessary to develop several draft versions before finalizing the model.” Kim Morneau, VP, Forecasting, Synovate Healthcare, part of Aegis Group Plc., helps the global pharmaceutical and medical-device industries achieve business objectives. For more information, visit synovate.com/healthcare. “An accurate model to project patient uptake must take into account the flow of patients into the physicians’ office for a new prescribing decision, the physicians’ desire to prescribe the new product relative to all other treatment options that will be available, and any market barriers that may prevent the physicians from using the product they prefer, such as restricted formulary access or insurance reimbursement. We use primary market research in conjunction with a proprietary database of new product launches to account for these and a variety of other factors, including salesforce size and market entry position.” Christian Schuler, Partner, Simon Kucher & Partners, a global provider of strategy and marketing consulting. For more information, visit simon-kucher.com or e-mail christian.schuler@simon-kucher.com. “Developing accurate forecasting models for pharmaceuticals largely depends on the quality of the data and assumptions that one puts in such a model. Robust forecasting models require sophisticated primary research, not just secondary data. In this research it is crucial to analyze the impact that the envisioned price for a product has on its overall market potential. Ideally, companies investigate the acceptance of a new product by key stakeholders, such as payers, physicians, and patients. This research should make use of state-of-the-art methodologies, such as discrete choice modeling, to derive accurate estimates on key variables for the forecasting model such as achievable peak market shares, speed of market uptake, and price sensitivity in the target market.” Craig Scott, CEO, TargetRx, which provides pharmaceutical and biotech companies with insight into the drivers of physician prescribing and brand adoption. For more information, visit targetrx.com. “Forecasting models that rely on physicians’ stated share, using a one-size-fits-all adjustment for overstatement are prone to error, because what physicians say they will prescribe versus what they actually prescribe varies dramatically by brand. In fact, for launch brands we’ve tracked in more than 25 therapeutic categories, actual share at one year post-launch ranged from 19% more than stated to 99% less than stated. To accurately forecast market share, companies need an in-depth understanding of how physicians are positioning the new drug in their minds, how they’d incorporate the brand into their treatment patterns, and how to translate prelaunch attitudes into likely postlaunch prescribing.” John Tapper, Ph.D., CEO, Ziment, a provider of strategic marketing research to pharmaceutical, biotech, and medical-device companies worldwide. For more information, visit ziment.com. “Many factors are essential in building accurate forecasts — market size, epidemiology, market access, competition, regulations, order of entry, etc. But none of that matters if market share estimates are wrong. Marketing research is a start, but there is so much overstatement in those data that it’s important to have a rigorous calibration approach. Too many companies rely upon rules of thumb or what ‘feels right.’ To get this right, companies must rigorously interview physicians and adjust responses using a well-reasoned, historically based calibration factor. This is one of the hardest things to do, but one of the most important.” Todd Johnson MattsonJack The first thing that the forecaster will notice is that the market shares expressed in the forecast model might be very different from the shares reported in the preallocation phase of the market research. Industry events • Pharmaceutical Forecasting Excellence Summit 2009 April 20-22, 2009, Mirasierra Suites Hotel, Madrid. For more information, visit eyeforpharma.com. • Operational & Strategic Forecasting in Pharma April 21-22, 2009, Woodbridge Hotel & Conference Center, Woodbridge, N.J. For more information, visit forecastpharma.com. • Marketing Research for Forecasting Seminar April 23- 24, 2009, Chicago. For more information, visit mattsonjack.com. Forecasting Best Practices: Two Perspectives Experts in the field of pharmaceutical forecasting describe their best practices approach to managing and implementing a forecasting strategy. Doug Willson, Senior VP, Advanced Methods, GfK Healthcare, says: • Centralize the forecasting function; use a standardized approach to presenting data, and ensure the right tools are used in the right environment. • Start early and update often; it’s easier to refine a forecast in response to new information than to start from scratch. • Learn from experience. With early-stage forecasting there are large differences between a product still in research and the one that enters the market later on. So there’s a need to continuously look at assumptions, forecasting methods, and results. • Use all available data — primary and secondary sources — and recognize the strength and weaknesses of each. • Recognize the fundamental role of uncertainty. In pharmaceutical research, it’s not possible to know the clinical properties of every drug in question or the behavior of competitors, so there’s a need to measure the impact of this uncertainty. John Yi, Ph.D., Forecasting Modeling Specialist, TVG Marketing Research & Consulting, says: • Seek comprehensive buy-in. The forecasting process needs to involve all internal stakeholders with data about the product — marketing, market research, clinical, etc. — to ensure all relevant information is incorporated, preconceptions are minimized, and output is trusted. • Cross-validate forecasting approaches and data sources. Multiple approaches, especially for market sizing and share prediction, and sources of data are needed to confirm results, minimize biases, and increase overall confidence in the forecast. • Listen to unbiased experts. Using unbiased experts — internal or external — to validate the approach and usage of data not only minimizes bias but gives an opportunity for the forecasting team to look at the process from a different and fresh perspective. • Standardize the process. Consistency and transparency across all product forecasting will produce valuable efficiencies. • Practice knowledge management. Key learnings from analyses can be grouped into categories and stored into a database for fast retrieval building institutional memory. • Gain top management support. Senior management has to be involved and support the approach taken, data used, and most importantly, shelter the forecasting process from any political pressures. Doug Willson, GfK Healthcare Dr. John Yi TVG Marketing Research & Consulting

Posted in:

Post a Comment

You must be logged in to post a Comment.

FEEDBACK