The option you do not want can still change the decision

Imagine two restaurants: the first serves two dishes you love, the second serves those same dishes as well as one you dislike.

Which restaurant is better?

Surely the second. Nothing has been taken away: you can order exactly what you would have ordered at the first restaurant and leave the extra dish alone.

A 2026 paper in the Journal of Consumer Research suggests otherwise. Adding a less-preferred option to an attractive set can make people less likely to choose the wider alternative containing it.

This may sound like choice overload, but Stephanie M. Smith and Stephen A. Spiller are studying something more specific: multi-option alternatives, where one decision leads to another. You select a restaurant before an entrée, a cruise before an excursion or a theatre before a film.

In theory, a customer can find the best option inside each alternative and disregard the rest; in practice, the rest has a habit of intruding.

A choice inside a choice

Suppose one theatre is showing Forrest Gump, while another lets you choose between The Wizard of Oz and Home Alone. If you prefer The Wizard of Oz to Forrest Gump, the addition of Home Alone should be irrelevant because nobody will make you watch it.

Yet the additional film can reduce the theatre's choice share. Smith and Spiller call this undervaluation of a multi-option alternative: people choose the theatre containing their preferred film less often than they choose that film when it appears alone against the same competitor.

The weak film did not simply make the strong one look worse. In one study, participants rated the preferred film similarly whether it appeared alone or beside a less-preferred film, but their rating of the theatre offering both fell. The set lost value even though its best component did not.

Marketers are used to building propositions by addition, assuming that another feature creates value or sits harmlessly in the list. That logic works in a spreadsheet; a customer looking at the offer may make a messier judgement.

The result survived tougher tests

The paper reports 15 preregistered online studies across films, cruises, short-form videos and risky gambles. Six present the central evidence, while nine replicate the findings, test other explanations and check whether the effect survives changes in design.

The effect appeared in repeated choices and in a one-shot decision between two cruises described as equivalent apart from their excursions. After the preregistered exclusions, 61.5% of people chose the cruise with the more attractive excursion when it appeared alone; when the same cruise also included a less attractive excursion that could be ignored, its choice share fell to 47.1%.

The result also survived larger sets. People became less likely to choose an alternative when an inferior third film was added to a two-film theatre, even though both sides already required another decision. A general reluctance to choose again cannot easily explain that result.

The gap in quality mattered too: as the difference between the preferred and less-preferred options grew, so did the undervaluation of the set. The paper does not claim that every additional option damages an offer. What gets added matters.

Then the researchers followed attention

The paper becomes particularly useful when it turns to attention. People may be averaging the options, but the studies do not establish a conscious calculation; they show that the weak option becomes more consequential when it occupies more of the decision maker's attention.

In the short-form video study, participants moved their cursor over each option to reveal it. Those who spent more time inspecting the less-preferred video were less likely to choose the alternative containing both, although that result alone cannot tell us which way causality runs. An off-putting option might simply attract a longer look.

The next study addressed that limitation with incentivised gambles by controlling how long each component was visible. In one condition, the inferior gamble appeared for roughly six seconds and the better one for three; in the other, the exposure times were reversed. Undervaluation was four percentage points when the inferior option received more attention, compared with one point when the better option received more attention. The one-point effect was not statistically distinguishable from zero, while the three-point difference between conditions was statistically significant.

Those timings cannot be carried straight over to a pricing page. The useful point is narrower: attention to the weaker downstream option did not merely accompany the effect; changing that attention changed the choice.

Two propositions can offer the same thing a customer ultimately wants, but the one that makes them dwell on something they distinctly do not want may be judged differently.

Feature lists have their own politics

Inside a product meeting, the case for one more benefit is hard to argue against. A software plan gains another tool, a membership gains another perk or a travel package gains another activity. Someone might value it; everyone else can ignore it.

The customer still has to evaluate the proposition. A package with ten benefits may contain everything offered by one with five without feeling more valuable. Some additions may be irrelevant or poorly matched, yet giving each one equal prominence invites the customer to take them all into account.

There is an obvious risk of taking this too far. Material features, limits and risks must remain clear, especially in regulated products; the studies offer no licence to conceal inconvenient details or cut anything with low usage. A weak option for one person may be the reason another person buys.

The practical question is one of hierarchy: which parts of the offer must be understood during the initial decision, which become useful later and whether the layout reflects that difference. Product pages often answer these questions accidentally rather than through evidence.

Low use does not mean no influence

Marketing teams often use selection or usage data to decide which parts of a proposition matter, so a reward, feature or activity chosen by few customers is easily labelled inconsequential.

That conclusion may be premature. Selection frequency and influence on acquisition are different measures: a rarely selected option can still affect whether someone chooses the package containing it. Usage data begins after the initial decision, so it cannot reveal which unchosen components helped or hindered that decision.

The studies cannot supply an answer for every commercial offer. Teams would need to test the same underlying proposition with different information hierarchies, then measure initial selection and later usage across relevant segments. The immediate shift is to stop treating an unused feature as an inert one.

The unchosen option stays in the room

Marketers count what an offer includes, with each new option lengthening the benefit list and making the proposition look more complete from inside the business. If nobody has to choose the weak component, it seems harmless.

The customer, however, is deciding what the proposition itself is worth. Smith and Spiller's research shows how an option can enter that judgement without ever entering the basket: ignoring it at the second decision does not erase what it did to the first.

What teams need to know

Is this another example of choice overload?

No. The studies examine initial choices between alternatives containing one or more downstream options, often using sets of only two or three, rather than testing whether very large assortments overwhelm people.

Does adding an option always reduce appeal?

No. The effect concerns adding a less-preferred option to an otherwise attractive alternative. Preferences differ between people, so the benefit of offering more choice depends partly on that variation.

What does the attention evidence show?

Mouse tracking linked greater attention to the weaker component with a lower likelihood of choosing the wider alternative. A later experiment manipulated exposure time, providing causal evidence that attention to the weaker option contributes to the effect.

Should marketers hide less popular features?

No. The research does not justify concealing material information or removing options simply because usage is low. It supports testing how information is prioritised while keeping the proposition clear, accurate and accessible.

What should teams test?

Compare different ways of presenting the same underlying proposition, then measure the initial choice alongside downstream selection, usage and differences between audience segments. This can reveal whether a weak component affects acquisition even when few customers choose it later.