Reading list: consumer search parsimony

Jia Liu & Olivier Toubia, Search Query Formation by Strategic Consumers

More evidence, if more were needed, in the keyword wars: given that consumers prefer shorter queries, and that they may assume that some attributes need more search than others, we should expect that at least some searches using trademarks are from consumers not actually looking for trademark-only results.  This research about strategic search behavior reinforces that conclusion.  Using examples from frequently-searched food-related terms, for example, the authors explain that:

[A] consumer who is looking for information on Kitchenaid and/or Cuisinart food processors[] may be better off using the query “Kitchenaid food processor” rather than a longer query that includes “Cuisinart.” This is because Cuisinart food processors are often compared to Kitchenaid, and the shorter query will retrieve at least one search result that contains all relevant terms. Including “Cuisinart” into the query greatly increases the proportion of results that contain this term, but this comes at the expense of other terms, in particular “Kitchenaid.” The shorter query already performs better under our simple implementation of the “Compensatory-Average” metric that weighs all terms equally, but the difference would be more pronounced if the consumer were primarily interested in Kitchenaid over Cuisinart. In that case, omitting “Cuisinart” from the query would ensure that most results mention “Kitchenaid,” while some results still mention “Cuisinart.”

In sum, our field data suggest that consumers indeed stand to benefit from being strategic in query formation, as shorter queries may be at least as effective at retrieving desired content, compared to queries that contain all the terms the consumer is searching for. The fact that consumers tend to prefer shorter queries makes these shorter queries even more attractive….

But do consumers actually engage in strategic searching?  They also run some experiments:

To sum up, the behavior we observe suggests that participants are able to strategically leverage activation probabilities between words, at least to some extent. Because our study uses a somewhat  artificial lab setting, we do not claim that the extent to which consumers leverage activation probabilities in the real world is the same as in our study. Instead, we view our results as proof of existence that consumers have some ability to strategically formulate queries that contain only a subset of the terms they are interested in, but that are effective at retrieving the other terms….

Our fndings suggest that the content that is of interest to consumers is not simply the content mentioned in their search queries, but also the content retrieved by their search queries.