Bringing a new product or service to market (or adding a new feature to an existing product or service) can be a bit nerve wracking.
With so many variables in play, it can be hard to pinpoint the exact combination of features that will resonate best with your customers. And, whether or not your offer proves to be a hit, it can be even more difficult to figure out the real reason behind its success (or failure).
Now, you could adopt a trial-and-error strategy to determine these reasons, making small changes to your offer over time and analyzing how they affect sales and customer satisfaction. That would not only cost a ton of time and money, but it would also put your company at risk of losing a good amount of your customers too.
Instead of just releasing a new product or feature to the public and gambling on whether or not your audience will be receptive to it, you can gauge their reception to your new offering early, through the use of conjoint analysis.
In this article I’ll explain what conjoint analysis is, how it’s used in marketing research and the different forms it can take. I’ll also discuss some of the pitfalls inherent in using conjoint analysis, and show you how you can avoid stumbling into them throughout your analysis process.
Sound good? Let’s get started.
What Is Conjoint Analysis?
As alluded to in the introduction, conjoint analysis is a method of gauging the value of a specific offering in the eyes of the target consumer.
As CEO of Price Intelligently Patrick Campbell explains:
“When trying to hunt down value, both from an attribute and a magnitude perspective, conjoint analysis is an extremely powerful tool to understand your market, especially if you have enough survey respondents.”
Conjoint analysis is conducted by asking respondents to choose the most favorable option of a group of feature sets. By asking respondents to compare multiple permutations of various product features, surveyors hope to better understand what their customers actually want from them in terms of a product offering.
Attributes And Levels
Conjoint analysis first requires surveyors to determine the specific attributes of a product or service to focus on, then define specific “levels” for each of these attributes.
The attributes focused on are typically the most important features of a product or service. For example, if the product in question is a new smartphone model, the attributes assessed would likely be features such as storage space, battery life and screen size.
The levels of these attributes would then be as follows:
- Hard Drive Size: 32GB, 64GB, 128GB
- Battery Life: 8 hours, 10 hours, 12 hours
- Screen Size: 3.5 in., 4.5 in., 5.5 in.
Note: Price is also a major component of conjoint analysis, as it can help gauge a customer’s propensity to spend more in exchange for higher functionality [for example, whether customers would pay more for a larger hard drive]. However, conjoint analysis shouldn’t be used to determine a fixed price point for your offering. We’ll dive more into this later on when we discuss the pitfalls of conjoint analysis.
As mentioned above, conjoint analysis works by presenting various permutations of a product’s features, and asking respondents to choose the preferred option. Using the above example, two feature sets might look like this:
- Set A: 32GB hard drive, 10 hour battery life, 3.5 in.
- Set B: 64GB hard drive, 8 hour battery life, 4.5 in.
This example essentially asks respondents to decide whether they value hard drive size over battery life length (or vice-versa). This, of course, is a relatively simple example; we’ll get into more nuanced situations in a moment.
Using Conjoint Analysis Results
Along with providing an opportunity to better understand what your customers actually want from you, conjoint analysis can help improve your offering (and your overall organization) in terms of:
- Product and feature development
- Market segmentation
- Brand positioning and messaging
In other words, it’s not so much that conjoint analysis helps you figure out what your customers want – it’s that it helps you figure out what they want that you’re capable of giving to them. In short: conjoint analysis helps define – and strengthen – your customer value proposition.
CEO of Optimization Group Jeff Ewald says:
“If you ask a rational person a rational question, you’ll get a rational answer.”
Conjoint analysis allows you to present your potential future offerings to your target consumer in a way that allows them to provide the most rational response possible.
In the next section, we’ll take a look at everything that goes into creating surveys to be used for conjoint analysis, as well as how to analyze the data you collect through these surveys.
Designing Surveys And Collecting Data For Conjoint Analysis
This probably goes without saying, but in order to get the most out of conjoint analysis, you need to optimize the design of the accompanying surveys and also understand the nuances of analyzing the information the surveys provide you.
Designing Conjoint Analysis Surveys
When designing a survey to be used for conjoint analysis, your focus should be on defining the features to be assessed, as well as resources needed to complete the entire process.
In terms of defining the features of your product to be focused on, they should be both relevant and accurate (or realistic).
Regarding relevancy, the features you focus your survey on should be those that, if altered, would drastically change your product as a whole.
Using the example from above, a smartphone’s hard drive size, battery life, and screen size are all relevant features: a change in one of these aspects would change the overall experience for the typical user. On the other hand, the length of the phone’s charger cable probably wouldn’t make too much of a difference to the average customer.
Regarding accuracy, you need to ensure that the features presented are actually realistic in terms of what your company is capable of offering. Again using the example above, you wouldn’t want to include a feature set that includes a 512GB hard drive or a 48-hour battery life, as these specs are way out of reach of currently-existing technology.
On this same token, you also want to ensure the features you present are in-line with industry standards, as well as the expectations of your customers. Presenting features outside of these expectations can either result in respondents providing invalid responses (due to excitement over brand new, never-before-seen features), or providing responses that don’t tell you anything you didn’t know (e.g., most smartphone customers wouldn’t choose a phone with an 8GB hard drive even if you were practically giving it away).
Finally, it’s important that you allocate an appropriate amount of resources to your campaign throughout every step of the process:
- Before Conjoint Analysis: Before even undertaking the process of conjoint analysis, you should have a firm understanding of your industry’s standards, the climate and expectations of the current market, and your customers. This may require that you conduct surveys regarding your customers’ attitudes as consumers, and also analyze the current trends within your industry.
- Survey Distribution: As you distribute your conjoint analysis survey, make sure you’re taking into consideration the characteristics of your respondents, and that you know what data you’re looking to collect. If your customer base is made up of a variety of consumer personas, you might want to only reach out to a specific segment at a time so the data you end up collecting is more focused and reliable.
- Collecting and Analyzing Data: Although you can certainly utilize automation to collect and organize your survey data, the entire point of conducting such surveys is to understand your customers’ choices as they apply in real-world scenarios. Knowing this, it’s clear that manual analysis of the data you collect is of utmost importance.
Analyzing Collected Data
As mentioned, the data you’ll collect through conjoint analysis surveys is largely quantitative.
Sawtooth Software reports there are four types of quantitative data with regard to conjoint analysis:
- Nominal Data refers to code numbers used to represent categories or feature sets. This simply makes it easier to organize the data being collected.
- Ordinal Data refers to the ranking of a specific feature set within the entire collection of data. If, for example, five feature sets are presented, the most-preferred set would be ranked 1, and the least-preferred 5.
- Interval Data refers to the “distance” between two ordinal data points with relation to standard deviation and mean. In other words, it explains the strength of a consumer’s preference for feature set 1 versus their preference for feature set 2.
- Ratio Data compares the above-mentioned “distance” with reference to each other. For example, a customer’s preference for feature set 1 may actually be twice as strong as their preference for feature set 2.
Another aspect of conjoint analysis is the concept of utilities, or part-worths.
Essentially, utilities determine the “weight” a specific feature has in terms of how it affects a consumer’s overall preference for an overall feature set. Using our smartphone example, the collected data may end up showing that consumers care more about hard drive space than they do about screen size, in which case hard drive space would be weighted more heavily.
Types Of Conjoint Analysis
A large number of conjoint analysis methods have been used for a variety of purposes throughout the years. Today’s marketers, however, tend to favor the following three types of conjoint analysis:
Choice-Based (or Discrete-Choice) Conjoint Analysis
Choice-based conjoint analysis is the most popular method used by marketers. In this method, feature sets are presented in full, and respondents are asked to choose the one they prefer most.
Rather than mixing-and-matching features, choice-based conjoint analysis presents predetermined feature sets to gauge customers’ interest in a (hypothetically) fully-developed product. After a choice is made, the process is repeated with another selection of products.
This, in essence, is what consumers do when actually browsing products in a real-world setting: they peruse a finite group of products and analyze each in terms of their given features in order to determine which one they want to buy.
Though the consumer-facing process simply asks respondents to state their preference for a specific feature set, their responses are compared on a much deeper level in order to determine which features matter most to the customer base as a whole.
For example, if consumers consistently show a preference for more hard drive space (even as other feature attributes vary), you’ll know for certain that a larger hard drive should be the focal point of your offering.
Adaptive Conjoint Analysis
As the name implies, adaptive conjoint analysis adapts to the preferences of individual respondents as they go about completing the survey.
Adaptive conjoint analysis surveys begin by determining a baseline for the respondent’s preferences (which can be solicited beforehand). Feature attributes and levels are then presented, with respondents asked to select their preferred option.
As the survey continues, and the respondent’s preferences become more and more clear, certain options will begin to disappear – while others will become more prevalent. For example, if a respondent consistently shies away from choosing the smartphone with the smallest hard drive, feature sets including small hard drives will cease to be presented. The process will repeat with all attributes until a specific product profile is determined.
While it may seem like such a process might take a while to complete, the personalized aspect of adaptive conjoint analysis surveys help engage respondents
Because adaptive conjoint analysis surveys, by definition, adapt to individual respondents, they are much more engaging than “standard” surveys distributed to the masses. Furthermore, as certain attributes are eliminated as the survey progresses, it eliminates boredom and fatigue (as respondents won’t ever be provided with the same choices twice).
MaxDiff Conjoint Analysis
MaxDiff conjoint analysis requires respondents to define their most and least preferred feature (or feature set) among a given list. Four or five feature sets are presented at a time from a much larger pool of feature sets – but respondents will only choose their first and last choices (the other three or four choices don’t matter for the time being).
As the survey continues, respondents will end up unconsciously “ranking” the presented feature sets (as determined internally by the survey tool). For example, a respondent might report that they prefer Audi in the above screenshot and all subsequent questions involving Audi. However, when Audi is not one of the choices provided, they choose BMW over all others. Though not directly stated, it would then be clear that BMW is their second-favorite brand of automobile.
This, in fact, is the main benefit of MaxDiff surveys: while consumers are pretty adept at at defining their favorite (and least-favorite) of a given set of choices, they have a tougher time deciding which is their second- or third-favorite option. Instead of asking them to rank their choices point-blank, MaxDiff conjoint analysis does all the “grunt work” while still leading to accurate and reliable data.
Pitfalls And Misconceptions Of Conjoint Analysis
While conjoint analysis can certainly provide valuable insight into the preferences of your target consumers, it’s definitely possible to be misled by the data you glean from the process if you’re not careful.
Firstly, although we’ve said that (at least with regard to the choice-based method) conjoint analysis is meant to simulate a real-world shopping experience, the truth is that doing so is, for all intents and purposes, impossible.
No matter how immersive the experience may be, respondents will always understand that they aren’t really choosing to make a purchase while completing a conjoint analysis survey. In other words, the choice they’re making is hypothetical (i.e., “If I were to make a purchase, it would be this one”); it’s in no way an indication that they truly would make such a purchase.
Along with this point, respondents completing the survey essentially have to make a decision that ends in making a purchase. Obviously, this is not the case in the real world; they can always walk away empty-handed if they so choose.
Another thing to consider is the fact that, while completing such surveys, customers are responding to features as defined by the provider rather than providing their own suggestions for features. This goes along with the above sidebar, in that respondents may simply be choosing the best available option – not the one they actually prefer. Again, this doesn’t translate to a propensity to buy; it’s simply a statement of hypothetical preference.
Lastly, while conjoint analysis can be used to determine how much the price of an offer factors into a customer’s decision to purchase the product in question, it won’t tell you exactly how much they’re willing to spend. For this, Patrick Campbell suggests using Price Sensitivity Meter, which will allow you to “do more with less and get some really good elasticity data.”
Implemented and utilized correctly, conjoint analysis can uncover valuable data regarding your target consumer’s preferences for specific products and individual product features.
As with all customer-facing surveys, though, it’s important to approach conjoint analysis from a real-world perspective. In other words, collecting data through conjoint analysis survey methods is merely the first step in understanding your customers; the next step is to figure out why their preferences are what they are – and how your offerings fit into the whole picture.