Please login to the form below

Not currently logged in
Email:
Password:

X-ray of customer choice

Conjoint analysis need not be complicated or expensive to garner accurate insights

bright_fish The idea behind conjoint analysis is not difficult to understand. Imagine that you are a doctor and I offer you a product with high efficacy and high price or a product with low efficacy and low price. If you tell me that you prefer the product with high efficacy and high price, I can infer that efficacy is more important to you than price. If you say that you prefer the product with low efficacy and low price, I can infer that price is more important to you than efficacy. Conjoint takes this one step further by telling us how much a doctor will like any combination of product features and how much each feature contributes to the overall likeability of the product.

The contribution that any given product feature makes to the overall likeability of a product is called - in conjoint-speak - its 'part-worth'. A conjoint study is designed to allow the part-worth for every product feature to be estimated.

So, for a product with high efficacy, high safety and low price, the conjoint might tell us that the corresponding part-worths are: high efficacy = 5, high safety = 3 and low price = 2. Adding these part-worths together indicates that doctors would give this product a score of 5 + 3 + 2 = 10.

Another product with the features of low efficacy, medium safety and low price might have part-worths of low efficacy = 1, medium safety = 2 and low price = 2. Again, adding these part-worths together shows that doctors would give the product a score of 1 + 2 + 2 = 5.

Specific terms
In a real survey, vague terms like 'high efficacy' and 'low price', are not used. Specific terms, like '85 per cent response rate' and '£4 per day's therapy' are employed.

These likeability scores are known as 'preference scores' in conjoint-speak. Of course, you, as a marketer, will want to know how these preference scores will translate into market shares. In the early days of conjoint, it was thought that market shares would be proportional to preference scores. So, for the two example products with preference scores of 10 and 5, market shares would be in the ratio of 10:5 = 2:1, ie 67 per cent and 33 per cent.

However, this is not accurate. The market tends to reward better products more than their preference shares suggest and to punish inferior products more than their preference scores suggest. So, in the example given, the market shares might be more like 80 per cent and 20 per cent. There are simple mathematical tricks to produce this 'exaggeration' in the market shares. There are also simple ways to boost the accuracy of conjoint by combining the impact of promotional share, launch order effects, co-prescribing effects and so on, but unfortunately there is not space to go into those here.

The ability to estimate how a product's features determine how much doctors like it, and hence its market share, is phenomenally useful. It is worth considering some of the uses for this X-ray view of the mind of the prescriber.

Marketing
Your product is different from that of your competitors. You have to decide which advantages to stress in your marketing messages. When you have the results of a conjoint study, that decision is relatively simple.

It is simply a matter of looking at the part-worth scores and emphasising those features where there is a big difference between your product's part-worth score and that of your competitors.

For example, one client's promotional strategy was focused on subtle differences in efficacy for its 'me-too' brand. A conjoint revealed that doctors were actually far more excited about how easy is was to squeeze the dropper bottle that these products were administered in, thereby opening up a promising new marketing avenue.

Research and development
R&D development project teams usually include someone from strategic marketing whose job it is to bring the 'voice of the market' to the meeting. This is to ensure that the product is developed for maximum appeal to future customers, including prescribers and payers.

In one case, a project team was developing a four-times-daily dosing regimen, because, given the pharmacokinetics of the drug, this was the dose frequency required for maximum efficacy.

However, a conjoint study revealed that doctors were prepared to trade off a significant level of efficacy in order to achieve twice-daily dosing. The development plan was changed and the product category went on to be the leading therapy for this respiratory disease.

Pricing
You could give a new product a high price feature or a low price feature, according to where you decide to set the price. A conjoint will tell you the part-worth for different prices. You can then estimate your product's market share at different prices.

For example, the market access team for a major new product was developing a reimbursement support package that focused on the product's more favourable means of administration. They wanted to argue that better administration led to better compliance, which in turn led to better efficacy. A conjoint study revealed that payers assigned very little importance to administration-centred arguments, as is often found to be the case. As a result, the reimbursement case was totally refocused, leading to a successful launch.

Forecasting
As has been demonstrated, conjoint preference scores can be turned easily into market share forecasts. This allows you to estimate the market share for your new product, or if you have an established product, to estimate the market share that a new competitor may steal.

When a marketing team was facing competition from an impending product launch by one of the world's most successful and aggressive pharma companies, it was ready to throw in the towel, remove promotion and abandon the market to the bigger, more powerful rival. A small conjoint study, however, clearly showed that the competitor's profile was considerably less attractive than that of its own product. It also highlighted why the new threat was less attractive. The small incumbent company was able to refocus its promotion on the attributes that mattered and, despite being massively out-spent by the rival, retained its market-leading position.

Reasons for under-use
You might expect that such a potent market research tool would be in widespread use in marketing and market research departments. This, however, is not the case. Conjoint analysis tends to be used mainly by corporate head offices. Indeed, it has become something of a rite of passage for a new product to have a multi-country conjoint performed while it is in phase III of development. The reasons for this appear to be derived from the idea that conjoint is expensive, so it can only be justified and funded by a corporate head office.

Conjoint was 'invented' by American academic, Professor Paul Green, in 1972. I say 'invented' because, like most market research ideas, it was, in fact, stolen from another discipline - in this case, psychology.

Over the last few decades, academics have tinkered with conjoint analysis, resulting in greater complication. This has made it increasingly difficult to understand and ever-more expensive. However, head-to-head studies comparing the accuracy of complex conjoint techniques and simple and easy-to-understand conjoint techniques have delivered some good news: the simplest, cheapest and most easily understood type of conjoint is at least as accurate as the most expensive and hardest to understand.

As long ago as 1990, writing in the Journal of Marketing, Professor Green acknowledged that the very simple – or self-explicated – form of conjoint was as accurate as the more complex varieties, saying, "The empirical results to date indicate that the self-explicated approach is likely to yield predictive validities roughly comparable to those of traditional conjoint analysis."

Then, in 2000, Professor Henrik Sattler, Unilever professor of marketing at Jenna University, reviewed all the head-to-head studies comparing simple with complex conjoint techniques and found that the simplest types of conjoint were at least as good as the more complex approaches.

Regarding this simple, self-explicated approach, Prof Sattler said, "Our comprehensive analysis of empirical studies comparing these approaches fails to confirm the superiority of [complex] conjoint measurement." He also concluded that "given the clear majority of empirical findings not in favour of [traditional] conjoint measurement, future applications of [traditional] conjoint measurement...seem to be at least questionable because of the advantages of self-explicated approaches in terms of ease, time and costs."

In light of the now established empirical fact that complex, hard-to-understand and costly forms of conjoint are no more accurate than simple, easy-to-understand and far less expensive forms, it is reasonable to expect that pharma would have happily switched wholesale to the simpler techniques, but this has not been the case.

Why would market research agencies want to continue to sell a more expensive solution when a much less expensive one is as effective? The answer is in the question, of course. Another causal factor is illustrated by an amusing scientific study conducted in the 1970s by American academic, Donald Naftulin. He set out to show that even the most sophisticated audience is subject to the rule that 'bullshit baffles brains'. He chose psychiatrists and psychologists as his victims and arranged for a British academic 'Dr Fox' (who was actually an actor) to give a series of talks on the implications of game theory for physician education.

Needless to say, game theory has nothing to do with physician education. The entire talk was nonsense and full of self-contradiction, but it was sophisticated nonsense delivered by an imposing and dignified looking 'doctor'.

After talking for an hour, Dr Fox answered questions for half an hour. As the actor knew nothing about the subject, he answered with scripted nonsense – but again 'sophisticated' nonsense.

Surely, such a worldly, experienced and educated audience was not taken in? Indeed it was. Feedback was unreservedly positive, with accolades such as 'clear and stimulating'.

We are all prone to be intimidated by 'sophistication' and to use 'complexity' as a security blanket in the face of uncertainty. Indeed, in my experience, the less people understand conjoint, the more they are 'wowed' by its more complex and arcane versions.

Perhaps it is time to stop viewing conjoint as a complex, hugely expensive and mystical technique reserved for big-spending corporate head offices and see it instead as a simple technique that can be widely used – alongside other techniques – by marketers everywhere.

The Author
Gary Johnson is managing director of Inpharmation Ltd
To comment this article, email pm@pmlive.com

28th May 2009

Share

Featured jobs

Subscribe to our email news alerts

PMHub

Add my company
Genetic Digital

We help healthcare, pharma & medical device organisations use web and mobile technologies to effectively connect patients and healthcare professionals....

Latest intelligence

Big data, privacy and the rise of genomic testing
Blue Latitude Health speaks to Johan Christiaanse, Marketing Director at BGI, to find out how the medical profession can overcome one of the major barriers to precision medicine – big...
Deal Watch January 2018
...
Emotional Intelligence and Blended Learning in Healthcare
...

Infographics