scenario forecasting techniques

The author is Head of the Centre for Strategy and Policy at the Open University School of Management. He has also published a number of books, including the MBA textbook, 'Marketing' (Basil Blackwell), and the reader, 'Managing the External Environment' (Sage).

Scenario forecasting has traditionally been seen to be very sophisticated, and is accordingly now used by only a few of the very largest organisations. This article describes a simpler approach, derived from that of Shell but developed on the basis of practical experience with 1,000 students writing more than 4,000 scenarios. It should allow a wider range of organisations to gain the very valuable benefits offered by the technique.

Though the concept was first introduced, as 'La Prospective', by Berger[i] in 1964 and the word 'scenario' itself was reportedly first used by Herman Kahn in 1967[ii], the theoretical foundations of scenario forecasting were mainly developed in the 1970s, especially by Godet (between 1974 and 1979[iii]). By the early 1980s these approaches had developed into a sophisticated forecasting technique which was primarily recommended for the integration of the output from other sophisticated (qualitative) approaches to long-range forecasting. Although it was inevitably based upon judgmental forecasts, its use typically revolved around forecasting techniques which brought together groups of experts in order to reduce the risk involved. The techniques included Delphi and, especially in the context of scenarios, Cross-Impact Matrices, which were popular at that time.

Possibly as a result of these very sophisticated approaches, and of the difficult techniques they employed (which usually demanded the resources of a central planning staff), scenarios earned a reputation for difficulty (and cost) in use. Even so, the theoretical importance of the use of alternative scenarios, to help address the uncertainty implicit in long-range forecasts, was dramatically underlined by the widespread confusion which followed the Oil Shock of 1973. As a result many of the larger organisations started to use the technique in one form or another. Indeed, just ten years later, in 1983 Diffenbach[iv] reported that 'alternate scenarios' were the third most popular technique for long-range forecasting - used by 68% of the large organisations he surveyed.

Practical development of scenario forecasting, to guide strategy rather than for the more limited academic uses which had previously been the case, was initiated by Wack[v] in 1971 at the Royal Dutch Shell group of companies (hereafter, for simplicity, referred to as 'Shell') which has since led the commercial world in the use of such scenarios - and in the development of more practical techniques to support these.

As most of the work reported to date, especially that undertaken by Shell, has mainly focused on improving the effectiveness of the techniques involved we have concentrated our own efforts on making the scenario planning process as a whole easier to use; making it more accessible to a wider range of organisations. More important, perhaps, these developments make the technique available for use by almost all managers - across the organisation - who might want to use it as one part of their overall planning processes; without the need for extended involvement of outside experts, or corporate staff specialists.

Thus, experience in the Open Business School (OBS) - using the simpler approaches we have developed - indicates that it can be implemented with relative ease in most organisations. Its use in dealing with uncertainty - arising from alternative futures, which was the original reason it entered into management theory, still holds true - and usefully complements more traditional forms of planning. In our experience, however, its greatest virtue - especially in the simplified form - is that its use naturally (and painlessly) widens managers viewpoints and helps to extend their planning horizons beyond the short term.

The form of scenarios we chose - as the main vehicle to link the various elements of our distance-taught MBA course, the 'Challenge of the External Environment' which is taken annually by 500-600 of our 6,000 MBA students - was the widest-ranging possible; that of 'environmental scenarios' (sometimes referred to as 'industry scenarios' to distinguish them from 'global scenarios' undertaken by governments). These examine the total external environment - typically for the whole industry within which the organisation operates.

In order to meet the needs of managers who are not experts in long-range planning, it was found necessary, not just to introduce students to other people's material, but to develop our own simpler approach to the subject; albeit that this builds upon the work of others (especially that of Shell, much of it unpublished). In creating this easier-to-use form of scenario forecasting, we have in turn built upon the experience of more than 1,000 students who have taken the course over this time - and who have, between them, written more than 4,000 full length scenarios.

In particular, permission was received from 213 of the students taking the course in 1993 for their work to be used as a basis for statistical analyses. Of these projects, 165 were found to be suitable for further (content) analysis. These were analysed, in the context of this paper, to see how they used the techniques they had been taught.

The most important message is that scenarios CAN be simple; and, in our experience, the simpler they are - and the simpler the process used to derive them - the more powerful they may be; not least because those using them are able to understand how they work. Indeed, our students in general used the technique quite specifically as a simple one; eschewing the more sophisticated elements. For instance, despite having been introduced to the technique on the course, only 3% of the students made any use of the cross-impact matrices which were almost mandatory in the scenarios of earlier times.

In fact, the basic concepts of the process are simple. In terms of the overall approach to forecasting, they can be divided into three progressive groups of activities (which are, generally speaking, common to all long range forecasting processes):

The first of these groups quite simply comprises environmental analyses. The central part represents the specific techniques which differentiate the scenario forecasting process from the others in long-range planning; and it is, not surprisingly, this which will take up most of the rest of the paper. The final group represents all the subsequent processes which go towards producing the corporate strategies and plans.

Scenarios can only be as good as the information they are based upon. For this reason, the analysis must be of as high a quality as possible. The practical advice we give our students is to simply to cultivate a deep curiosity about the external environment; and to maintain maximum exposure to the widest range of media. Beyond this, they are also recommended to develop an informed viewpoint which will improve their chances of recognising early signs of change, no matter from what direction they are coming.

In practice, 72% of students in the survey used general reading as the main source of their analysis, combined with the information they received from the industry and specialist press - which they read as a normal part of their work. A further 18% did not even extend their search beyond general reading! A small proportion, 7%, did undertake a wider search, covering a range of sources as scenario theory would recommend, but there was no obvious major gain in final quality from this. Less than 1% used any form of database; even though these students were required to make extensive use of remote computer services as part of the course. Indeed, it has to be noted that the type of information which is required for environmental scenarios is most probably that which the participants have already assimilated, from their general (and specialist) reading

In addition, there seems to be no special expertise involved in detecting these shifts. Indeed, perhaps the best advice is to analyse the external environment as a team. If nothing else, this extends the coverage of the scanning; but it also seems to go much further; to develop resonances as the team interact with each other - inevitably comparing notes as the process develops - and amplify the early signs of change. Our experience indicates that the most effective team size is between five and seven members.

The subsequent analysis needs more academic rigour. The key here is that the process is one of education for the team - by total immersion - in the facts which define the environment they are studying. When the scenario development finally gets under way it is not the material available on paper that is productive, it is what is in the team's heads. Indeed, the first stage of scenario forecasting - the choice of the assumptions - is embedded in this supposedly earlier process of environmental analysis. It is inevitable that, as the team works together on the analysis, it will start to develop ideas as to what the assumptions might be; and will probably have spent a considerable time - over the weeks, and perhaps months, that the environmental analysis should take - arguing about what these mean.

Even this part of the process is, at its most basic level, relatively simple - requiring just six steps:

The first stage is to examine the results of the environmental analysis, to determine which are the most important factors that will decide the nature of the future environment within which the organisation operates. These factors are sometimes called 'variables' (because they will vary over the time being investigated), but we prefer the term 'drivers' (for change) since this terminology is not laden with quasi-scientific connotations and reinforces the participant's commitment to search for those forces which will act to change the future.

In the ideal approach, the first action should be to carefully decide the broad assumptions on which the scenarios will be based. Only then, as a second stage, should the various drivers be specifically defined. Participants, though, seem to have problems in separating these stages, and indeed only 40% of those in our survey made any attempt to spell out the assumptions separately. Despite this, most still managed to discover the drivers, without any obvious deterioration in quality; so having separate stages is not essential to the process - and we no longer emphasise this separation in our own use of the technique.

Perhaps the most difficult aspect though, is freeing the participants from the preconceptions they take into the process with them. In particular, in our experience, most participants will want to look no further than the medium term, five to ten years ahead (36% of those in our survey did so, and indeed, a further 42% wrote about periods less than five years ahead), rather than the required longer-term, ten or more years ahead (which only 22% addressed). This may not seem a problem, when nine years is a very long time in many areas of commercial activity, but we have found that a time horizon of anything less than ten years often leads participants to extrapolate from present trends, rather than consider the alternatives which might face them. When, however, they are asked to consider timescales in excess of ten years they almost all seem to accept the logic of the scenario planning process, and no longer fall back on that of extrapolation.

There is a similar problem with expanding participants horizons to include the whole external environment. Only 22% of the scenarios in our survey could be considered as being totally externally oriented. The largest category, 42%, took in some of the external environment but mixed it with internal factors. Despite the thrust of our teaching, a relatively high proportion, 36%, wrote what amounted to corporate scenarios (which describe the future of the organisation itself, largely on the basis of internal factors). On the other hand, the results indicated that if participants can be persuaded to address the ten year horizon, perhaps an easier task, this also tends to make them look further out in terms of the external environment.

In any case, the brainstorming which should then take place, to ensure that your list is complete, may extend their viewpoint and unearth more variables - and, in particular, the combination of factors may suggest yet others.

Almost any from of brainstorming seems to work, but our own, OBS, internal usage has revolved around the use of flip-charts, mainly because that has been the available medium and is that often prescribed for creative decision-making sessions.

This step is, though, also one of selection - since only the most important factors will justify a place in the scenarios. The 80:20 Rule here means that, at the end of the process, management's attention must be focused on a limited number of most important issues. Experience has proved that offering a wider range of topics merely allows them to select those few which interest them, and not necessarily those which are most important to the organisation.

In addition, as scenarios are a technique for presenting alternative futures, the factors to be included must be genuinely 'variable'. They should be subject to significant alternative outcomes. Factors whose outcome is predictable, but important, should be spelled out in the introduction to the scenarios (since they cannot be ignored).

At this point it is also worth pointing out that a great virtue of scenarios is that they can accommodate the input from any other form of forecasting. They may use figures, diagrams or words in any combination.

The next step is to link these drivers together to provide a meaningful framework. The theoretical approach adopted by the course encouraged students to first of all build 'event strings', by linking together the various drivers and developing them to see the links between them, as well as their onward progressions over time. The great majority, 74%, of scenarios in the survey claimed to include this step, but inspection indicated that there was actually relatively little development - they merely restated the drivers in different combinations. It might seem, therefore, that the basic requirement at this stage is to simply group the drivers into combinations that are meaningful to the participants. Once again, we now use this simpler approach for own planning processes.

This is probably the most (conceptually) difficult step. It is where managers' 'intuition' - their ability to make sense of complex patterns of 'soft' data which more rigorous analysis would be unable to handle - plays an important role.

Again, in our own internal (OBS) planning we tend to use flip-charts to organise the drivers into meaningful patterns. The whole process appears to flow very naturally, once the participants accept the simple principles behind it.

The outcome of the previous step is usually between seven and nine logical groupings of drivers. In our experience this is usually surprisingly easy to achieve.

Having placed the factors in these groups, the next action is to work out, very approximately at this stage, what is the connection between them. What does each group of factors represent?

The main action, at this next stage, is to reduce the seven to nine mini-scenarios/groupings detected at the previous stage to two or three larger scenarios. The challenge in practice seems to come down to finding just two or three 'containers' into which all the topics can be sensibly fitted. This usually requires a considerable amount of debate - but this typically produces fundamental insights into what are the really important (perhaps life and death) issues affecting the organisation.

There is no theoretical reason for reducing to just two or three scenarios, only a practical one. It has been found that the managers who will be asked to use the final scenarios can only cope effectively with a maximum of three versions! Shell started, more than two decades ago, by building half a dozen or more scenarios - but found that the outcome was that their managers selected just one of these to concentrate on. As a result the planners there progressively reduced the number to two, which is the number we also now recommend.

As used by Shell, these two scenarios are complementary; since this helps avoid managers 'choosing' just one, 'preferred', scenario - and lapsing once more into single-track forecasting (negating the benefits of using 'alternative' scenarios to allow for alternative, uncertain futures). This is, however, a potentially difficult concept to grasp, where managers are used to looking for opposites; a good and a bad scenario, say, or an optimistic one versus a pessimistic one. In practice, we found that this requirement, to our surprise, posed few problems for the great majority, 84%, of those in the survey; who easily produced 'balanced' scenarios. The remaining 16% mainly fell into the expected trap of 'good versus bad'.

Having grouped the factors into these two scenarios, the next step is to test them, again, for viability. Do they make sense to the participants? If the scenarios do not intuitively 'hang together', why not? The usual problem is that one or more of the assumptions turns out to be unrealistic in terms of how the participants see their world. If this is the case then you need to return to the first step - the whole scenario planning process is above all an iterative one.

The scenarios are then 'written up' in the most suitable form. The flexibility of this step often confuses participants, for they are used to forecasting processes which have a fixed format. The rule, though, is that you should produce the scenarios in the form most suitable for use by the managers who are going to base their strategy on them. This is essentially a 'marketing' decision, since it will be very necessary to 'sell' the final results to the users. On the other hand, a not inconsiderable consideration may be to use the form the author also finds most comfortable.

Most scenarios will, perhaps, be written in word form (almost as a series of alternative essays about the future); especially where they will almost inevitably be qualitative. Nearly half (47%) of those in our survey chose to use the normal business report format - hardly surprising where they, and their audience, would probably use this in their day to day communications. A further 13% reduced the material to something closer to an expanded series of lists. Just over a quarter (28%) enlivened their reports by adding some fictional 'character' to the material - perhaps taking literally the idea that they are stories about the future - though it was still clearly intended to be factual. On the other hand, they may include numeric data and/or diagrams - as those of Shell do (and in the process gain by the acid test of more measurable 'predictions'); though less than 1% of our students did so. Finally, 9% of the students took some delight in using a fictional form, as they were entitled to; assuming the character of a leader writer of the Financial Times in the year 2010, for instance.

The final stage of the process is to examine these scenarios to determine what are the most critical outcomes; the 'branching points' relating to the 'issues' which will have the greatest impact (potentially generating 'crises') on the future of the organisation.

It is important to note that these - final - scenarios may be used in a number of ways:

Most basically, they are a logical device, an artificial framework, for presenting the individual factors/topics (or coherent groups of these) so that these are made easily available for managers' use - as useful ideas about future developments in their own right - without reference to the rest of the scenario. It should be stressed that no factors should be dropped, or even given lower priority, as a result of producing the scenarios. In this context, which scenario contains which topic (driver), or issue about the future, is irrelevant.

At every stage it is necessary to iterate, to check that the contents are viable and make any necessary changes to ensure that they are; here the main test is to see if the scenarios seem to be internally consistent - if they are not then the writer must loop back to earlier stages to correct the problem.

Perhaps the main benefit deriving from scenarios, however, comes from the alternative 'flavours' of the future their different perspectives offer. It is a common experience, when the scenarios finally emerge, for the participants to be startled by the insight they offer - as to what the general shape of the future might be - at this stage it no longer is a theoretical exercise but becomes a genuine framework (or rather set of alternative frameworks) for dealing with that future.

Scenarios are, of course, only a means to an end. They identify the long-term forces, and consequent events, which the organisation's conventional long-range planning must address. This next step, therefore, starts by matching the organisation's limited internal resources to the essentially unlimited external challenges which may face it. The special contribution of scenario planning, in this context, is to allow, and indeed encourage, the development of a robust set of strategies. These will not necessarily result in an optimal outcome for a specific situation, but should offer the possibility of achieving the best overall outcome. In particular, they should best protect, as far as possible, against all the major threats potentially facing the organisation, and then exploit the most important opportunities open to it.

The use of scenarios, therefore, should help ensure that as many as possible of the long-term threats and opportunities facing the organisation are identified and addressed. Shell[vi],[vii],[viii] has demonstrated a number of times how such an awareness - of the alternatives facing it - has enabled it, not just to handle changed market conditions, but to capitalise on them. At the beginning of the 1970s, when it first adopted its scenario planning approach, it was probably the weakest of the major oil multinationals. Two decades later it has become the strongest of them. These facts are not necessarily connected, but Shell's senior management is convinced that - at the very least - its dedication to scenario planning has made a significant contribution to this dramatic improvement in performance.

Scenarios do, however, still demand some considerable investment of time, often of scarce senior management time, so if they are to be justified they must earn their keep. In essence this means that they must be genuinely useful, and used, as the (external) basis for corporate strategy. Thus, it is recognised, by all those we have met who are actively involved in the scenario planning process, that persuading line managers to make use of the scenarios produced is by far the most difficult part of the whole process. It is no accident that the large corporate planning department at Shell say that they spend at least half their time promoting their scenarios.

Indeed, the whole process of scenario forecasting must be imbued, from the very start, with the objective of positively influencing the strategy of the organisation. This means that, on the one hand, the scenarios must be carefully balanced between stretching the imagination of the management and being believable. On the other, it requires a significant investment in the education of the senior management. The marketing of the scenarios needs to be every bit as sophisticated as the writing of them. The introduction of the process should be seen as a long-term project. It can take a number of years before senior managers really trust scenarios sufficiently to put their faith in the strategies which are developed from them.

The cultural problems facing those who wish their organisations to take scenarios seriously should not, therefore, be underestimated. As the final part of the course, our own students subsequently went on to produce a corporate strategy; based on the scenarios they had developed for their industry - or at least that was the brief. Based on the anecdotal reports of their tutors, backed by the 32 cases tracked, we found that in practice, despite all the teaching and very explicit instructions we had given, almost all of the students lapsed into producing 'traditional' strategies. These were of the sort which are the staple diet of our other strategy course, and which almost entirely addressed short-term internal factors, ignoring the external long term trends with which this course dealt! Only 13% delivered strategies which effectively addressed the outcomes of their scenarios!

Failure, in this way, to have an immediate impact on published strategy should not, however, discourage those considering use of the technique. Our own experience, and that of Shell, was that the first scenarios produced are relatively neglected in the subsequent planning. It may take literally years for the process to deliver all its benefits.

Even so, our observation is that the participants do obtain major benefits from the process - even in the short term. On the other hand, the main benefit is much less direct than that usually claimed, and indeed is often not even obvious to the participants themselves. It is the enduring change in viewpoint of all those participating; extending their perspectives to include the wider environment and the longer term. This has been our own experience, not least in terms of the 'surprising' elements which have emerged during the process; to be accepted as key determinants of strategy. Ours students have also reported that this widening of their perspectives represents the main benefit they obtain from the whole course, not just from the scenario planning part of it. In view of the 'short-termism' exhibited by so many managers, this shift in attitude must - by itself - be invaluable. Above all, therefore, scenario planning should be seen as a process of learning.

We are grateful to the many OBS academics, and the even greater number of OBS students, who have contributed to this work. In addition, it should be obvious, from the many comments in this article, that our overall inspiration, and many of the detailed ideas, were drawn from the experience of the corporate planning group at Shell. We thank them, and in particular Graham Galer, for their invaluable help, and hope that, as a result of this article, others will emulate their success in future.

[i] G. Berger, Phénoménologies du Temps et Prospectives, Presse Universitaires de France (1964).

[ii] H. Kahn, The Year 2000, Calman-Levy (1967).

[iii] M. Godet, Scenarios and Strategic Management, Butterworths (1987).

[iv] J. Diffenbach, Corporate Environmental Analysis in Large US Corporations, Long Range Planning 16 (3), (1983).

[v] P. Wack, Scenarios: Uncharted Waters Ahead, Harvard Business Review September-October (1985).

[vi] P. Wack, Scenarios: Uncharted Waters Ahead, Harvard Business Review September-October (1985).

[vii] P. Wack, Scenarios: Shooting the Rapids, Harvard Business Review November-December (1985)

[viii] A. Kahane, Scenarios for Energy: Sustainable World vs Global Mercantilism, Long Range Planning 25 (4), (1992).