Would any company spend $75 million making and launching a new product without looking into whether the public is likely to buy it? The entertainment industry does precisely that with alarming ease. Hollywood’s major movie studios and their marketing executives seem to believe that show business is different from other businesses, and that each motion picture is unique.

In some ways, their view is correct. Unlike other packaged consumer goods, for example, movies represent a deep personal experience for their viewers. Films appeal to a customer’s emotions, and this subjectivity makes movies difficult to evaluate using traditional marketing tools which tend to rely on the objective attributes of a product. In other words, research into previous movies is of little use in planning the release and marketing of future motion pictures – even if the movies are made by the same director. Steven Spielberg’s “Jaws” may be a box-office sensation, while his “Hook” fails to grip audiences.

All this means that the multi-billion dollar U.S. entertainment industry has been chary of decision-support systems and forecasting tools for their marketing exercises. That could change, based on research by Wharton’s Jehoshua Eliashberg along with colleagues from a Dutch university and the Kellogg School of Northwestern University. The authors recently published their research in a paper entitled, “Moviemod: An Implementable Decision-Support System for Prerelease Market Evaluation of Motion Pictures.”

Eliashberg and his co-researchers have created Moviemod, a pre-release market evaluation tool for motion pictures which is designed to generate box office forecasts and to support marketing, advertising and distribution-related decisions for a new movie after it has been produced but before its public release. In the U.S., the average amount of money spent on these activities is approximately $25 million. Because Moviemod does not need historical sales data for calibration, unlike other forecast models, film producers and distributors can also use the tool to fine-tune marketing strategies to maximize the tickets sold at the box office.

In two different trials in the U.S. (in 1993) and in Rotterdam (in 1997), claim Eliashberg and his colleagues, Moviemod has shown accurate results, predicting total sales and revenues to within 5% of the actual figures in two different markets. Based on previously proven marketing science models, Moviemod can be inexpensive to implement because the data is gathered from a consumer clinic. And the fact that it can be used before a movie is released and in a limited three-hour time frame is well suited to the small life cycle of a product like a movie.

Eliashberg believes that Moviemod can be extended to other types of entertainment goods like the theatre, games, TV programming and music events. “It can also work in a variety of markets, as the two successful experiments at USA and Holland have shown,” he adds. Cinema buffs tend to share experiences with friends and colleagues immediately and movie forecasting models have to pay close attention to its nature, the spread of positive or negative information and its intensity, as well as its duration. Moviemod tries to capture explicitly this “word of mouth” impact, Eliashberg explains.

Unlike other consumer goods, distribution strategies have an impact on total ticket sales as well not only because of the availability of the movie but also because intensity of distribution – the number of screens it is playing on – affects future demand. Moviemod incorporates dynamics and effects of distribution intensity on the word-of-mouth diffusion process. By measuring the levels of impact of various mass media influences such as film advertising or critics’ reviews, it provides diagnostic information on consumer adoption intentions to managers to help them make the appropriate advertising and distribution decisions.

The model’s mechanism works essentially as follows: A representative sample of movie goers, before being exposed to the movie, is considered to be in the undecided stage. Then it is exposed to the movie theme and promotional materials such as previews, posters and advertisements. The sample is then classified into so-called “considerers” and “rejectors” based on its responses before being exposed to the movie. After experiencing the movie, the sample is further divided into “positive” and “negative” spreaders.

Through this process the impact of marketing variables such as the movie theme, promotion, distribution strategy and movie quality are measured during the various states of transition. For example, the movie theme could influence the number of undecided consumers who reject the idea of seeing the movie once exposed to its theme. Or the distribution strategy may have a negative effect on the number of considerers who can be persuaded to see the movie if it is not readily available. Moviemod may even help cut down marketing budgets. “If the movie is likely to generate very positive word of mouth there is no need to over-advertise it,” says Eliashberg.

In the first experiment at a U.S. university, the predictive results were encouraging. Based on a sample of 140, Moviemod projected that “Groundhog Day,” starring Bill Murray, would gross $69.4 million at the U.S. box office. The actual box office figure was marginally higher at $70.8 million. The diagnosis provided by Moviemod indicated that the movie was quite sensitive to its advertising. Hence, doubling the advertising support for this movie could have improved its ticket sales significantly.

But Moviemod’s real road test was in Rotterdam, where it was put to work in co-operation with a Dutch distributor and exhibitor. The title was “Shadow Conspiracy”. Based on a consumer clinic and the base media plan provided by the distributor, Moviemod projected a Dutch national attendance of 13,170 visitors. Disappointed with this projection, the distributor wanted to examine the possibility of increasing the penetration rate of 6.17% by changing the media plans.

Combining consumer responses about the awareness levels of different media vehicles and the levels of media vehicle intensity, Moviemod tested different media plans for higher national attendance and incremental profitability. Again, the results were striking. After the modified media plan was put in place, the projected number of national attendees was 19,352 against an actual attendance of 18,612, while total revenues projected — 222,500 guilders — marginally exceeded the actual box office gross receipts of 214,000 guilders.

Compared with other marketing benchmarks used in Holland such as U.S. box office figures, which Dutch distributors commonly use, Moviemod is like a generational leap. Eliashberg and his colleagues believe that while Moviemod predictions were within 4% of the actual figures, U.S. box office receipts as an indicator would have yielded a minus 60% variation from the actual figures.

The managers of Pathe, the Dutch exhibition firm involved in the experiment, found Moviemod useful and felt models like it could contribute significantly to the quality of decision making in the entertainment industry. Eliashberg says that Pathe has started using the forecasting tool, although Moviemod has not yet found any takers in Hollywood.

With movie-going tastes changing annually, forecasting models like Moviemod could help movies like “Titanic” increase their box office grosses while helping duds limit their losses. Some day, science may even come to mean more than just science fiction in Hollywood.