Although similar in structure and analysis to choice-based conjoint analysis (CBC), adaptive conjoint allows many more attributes and levels to be tested, reducing what would otherwise be unacceptable respondent burden in a survey. ACBC has much more powerful price testing capabilities than normal CBC, since compounded pricing for specific product configurations can be provided to respondents independent of the price shocking component of the analysis. The deliverable is a simulator that enables the client to manipulate variable values to perform “what-if” analyses.
Apriori association is a powerful data mining tool, mainly used for quantifying which instances (items) occur together. In retail applications, this technique is often used in basket analysis..
CHAID (chi-square automatic interaction detection) is an advanced classification tool that separates individuals into unique groups, based on similarity in the values of a variety of input variables, using a “tree” branching methodology. CHAID provides probabilities of membership in the available classifications for each individual under consideration.
CBC, also known as Discrete Choice Modeling, is analysis of data where respondents are asked to choose a single alternative among a set of options (i.e. make a discrete choice). The deliverable is a simulator that enables you to manipulate variable values to perform “what-if” analyses
An EDA provides frequencies and normality distributions of all available data. This analysis involves extensive use of basic graphics (e.g. box charts, histograms) to quickly and easily describe data and focuses on completeness of data, and whether missing data can/should be imputed.
Gap analysis quantifies the gap between product usage and 1) awareness of the product and/or 2) the market potential of the product. Various basic analytical techniques are used, including multi-level awareness and barrier tournament methods.
Key drivers analysis (KDA) articulates in relative terms precisely which product attributes contribute the most to satisfaction, likelihood to purchase again, or likelihood to recommend. KDA can also be extended through regression to predict the absolute extent of improvement in the dependent variables one might expect through improvements in the drivers themselves.
Latent class analysis is a clustering technique used to separate individuals into homogeneous groups based on similar behaviors or attitudes. It is one of several methods that can be used for consumer segmentation research. Latent class analysis is particularly useful with choice-based conjoint, where consumers can be segmented into groups based on how they make their buying decisions.
Logistic regression is a predictive modeling technique used to estimate a binary outcome (binomial) or predict one of several potential categorical outcomes (multinomial). A scoring algorithm can be developed using the output of the regression to score a customer or prospect database based on, for example, their propensity to buy a certain product or service.
MaxDiff Scaling is a forced-choice exercise that yields the greatest separation (maximum differential) between various product choices. It is also known as “best – worst” analysis. MaxDiff is the most effective way to measure and differentiate the importance of product attributes.
OLS, or linear regression, is a basic analytical technique used to predict the absolute scale of a continuous-level variable (e.g. a satisfaction rating, spending level in dollars, etc.) based on changes to the levels of the various independent (predictor) variables.
Text mining involves the use of quantitative applications to derive context from unstructured text data, such as tweets, open-ended responses, product reviews, etc.
TURF (total unduplicated reach and frequency) provides the total potential market for a finite group of items offered as options to a consumer, as a subset of a much larger array of choices.