Recommendation algorithms have vastly altered how consumers choose a vast range of products, from digital entertainment, movies, and television shows, to books and music, to food delivery orders, to tourism options. Their impressive capabilities—based on their ability to integrate vast data about the consumers’ own prior choices but also the preferences of other, similar consumers—often enable people to discover new and appealing content, hidden gems, and options they never would have considered on their own. Yet as more and more service providers rely on algorithms to recommend purchases, and more and more consumers turn to them for ideas, their failures become more and more obvious too.

In a satirical essay, “Why Did the Algorithms Send Me Here?” Joe Queenan detailed a recent consumption experience involving a ticketing site. Having purchased tickets for a talk with a journalist, the site seemingly sprung to life, in his telling, offering well-placed recommendations for other events in his area. But none of those events had anything to do with the ticketed talk, nor any of his other interests.

Queenan chose not to name the platform in question, reflecting his broader point: Even when the algorithms fail, they have become so common and ubiquitous that, for many consumers, it is impossible to function in daily life without them. Thus, a new source of consumer friction and frustration has arisen in the algorithm age. The service failure that occurs when an algorithm is weak, overly general, or based on inaccurate criteria can create deep dissatisfaction.

Consumers have sought some workarounds. On Netflix for example, people have figured out how to reorient or reset their recommendation framework. In the former case, they prune their own search history. In the latter, they create an entirely new profile, which they use to select and watch particular content more intentionally. However, both solutions put the burden on consumers, rather than the algorithm or the service provider.

On other platforms, no such solutions seem available. For example, Goodreads offers book lovers a relevant platform for reviews and recommendations, but its results appear highly questionable. As many users have noted, the site tends to be “clunky and slow,” the available lists are often irrelevant, and the recommendations simply aren’t very good. Thus, the very name of the site seemingly makes a promise that it fails to achieve, reflecting the very essence of a service failure. Promoted as being expressly for bibliophiles, the site is ineffective in its advertised purpose. Yet visitors seem to keep coming back, in search of at least one good reading recommendation.

In a conventional market, we might expect consumers to reject the failed service and find an alternative. But rejecting algorithms altogether likely would represent a significant constraint on consumers’ experiences and enjoyment, which may be why they continue to function, poorly. Is the solution to the problem in the hands of consumers, or should the algorithm providers take the initiative to address the issue?

Discussion Questions

1. In your daily consumption, which recommendation algorithms do you use regularly? How effective is each of them, in your experience?
2. Answer the question that ends this abstract: Who is responsible for resolving the service failure created by poor algorithm recommendations?

Sources: Joe Queenan, “Why Did the Algorithms Send Me Here?,” The Wall Street Journal, April 5, 2025; Eli Becht, “How to Reset Your Netflix Algorithm,” Pocket-lint, October 18, 2024; Jonny Diamond, “The Problem(s) With Goodreads,” Literary Hub, September 5, 2019.