Adding a third option can make one of the first two look better, even when nobody plans to buy the newcomer. That is the decoy effect, also called asymmetric dominance or the attraction effect. The original evidence is real. The popular version claims more than the replications support.
What the bias is
A decoy is an option that is worse than a target on every attribute and worse than a rival on only some of them. The decoy is dominated by the target, so the target looks clearly better. The rival is not dominated, so it is the option that loses share.
Standard choice models predict that adding an option cannot raise the share of an existing one. The principle is called regularity. The decoy effect appears to break it, which is why the result drew attention in consumer research.
The mechanism is comparison, not preference. The target and the decoy sit close together on the attributes that matter, and the target wins on all of them. Once the target looks clearly better than its near twin, the chooser feels more confident about it. The rival, which is a different kind of option, stops being the obvious reference point. People who would have chosen the rival alone can end up choosing the target instead.
This also explains why the effect is fragile. It depends on the decoy being close enough to the target to invite comparison, and on the attributes being visible at the moment of choice. Remove either condition and the pull weakens.
The original evidence
Huber, Payne and Puto (1982) gave participants choices between options described on several attributes. In one condition they added an option that was dominated by one alternative but not by the other. The target’s share rose when the decoy was present. The authors framed this as a violation of regularity and proposed the similarity hypothesis: the decoy looks like its dominating twin, so the twin gains.
Simonson and Tversky (1992) offered a competing account. Their tradeoff contrast explanation says buyers avoid options that are extreme on one attribute. A decoy can make the target look like the sensible middle, which pulls choices toward it. The two accounts make overlapping predictions, so one experiment rarely separates them.
Ariely (2008) popularised a version with a subscription page: web-only, print-only and a print plus web bundle priced the same as print-only. The bundle looks like a bargain next to the print-only option. I could not confirm an independent replication of that exact example, so treat it as an illustration, not as evidence.
Replication status and effect size
The effect is conditional. Frederick, Lee and Baskin (2014) argued that the attraction effect depends on how attributes are represented and is often small or absent in the designs they examined. I could check only the paper’s title and bibliographic metadata, not its full text, so the description of its findings is hedged accordingly. The original authors replied in the same journal the same year, defending the effect and disputing the critique’s design choices (Huber, Payne and Puto, 2014).
The fair reading is this. The effect exists under some conditions. Whether it shows up in your market depends on the product, the attributes shown and the choosers. Trueblood, Brown, Heathcote and Busemeyer (2013) extended context effects to perceptual judgements, which supports the claim that these effects are not only a consumer quirk.
Many of the early experiments used student or panel samples, hypothetical products and a handful of attributes. That is a narrow base for predicting what a real customer does on a live checkout page. A replication that matters for business would use real purchases, realistic prices, several weeks of exposure and a pre-registered outcome. Few published studies meet all four conditions, so the current evidence should be read as a set of plausible patterns rather than a dependable lever.
I could not verify a pooled effect size from a meta-analysis. For that reason I give no percentage here. Any single number you see quoted is one study’s result, under one design.
| Source | Claim | Status |
|---|---|---|
| Huber, Payne and Puto (1982) | Adding a dominated option raises its dominating twin’s share | Original study, widely cited |
| Simonson and Tversky (1992) | Tradeoff contrast and extremeness aversion explain part of it | Competing account |
| Frederick, Lee and Baskin (2014) | Effect depends on attribute representation, often small or absent | Independent critique |
| Huber, Payne and Puto (2014) | Effect is real under proper design | Rebuttal by original authors |
| Trueblood et al. (2013) | Context effects also appear in perceptual choice | Extension to new domain |
| Ariely (2008) | Subscription bundle example | Popular illustration, no replication confirmed |
Everyday and business examples
Pricing tiers are the most common place to see the decoy. A “Business” plan with more seats for a small premium sits beside a “Team” plan that looks like a poor deal. The mid-tier then appears to be the sensible choice. This pattern is common on software pricing pages. I found no independent, measured lift for a specific tier design, so do not quote one.
Here is how a three-tier page can look. The numbers are invented for illustration, not data. A Starter plan at 9 per month includes 5 projects. A Pro plan at 19 includes 20 projects. A Business plan at 22 includes unlimited projects and a second user seat. The hope is that buyers who would have picked Starter now pick Pro, because Business makes Pro look like the reasonable middle. Whether that pattern holds for your customers is an empirical question, and the answer will depend on your own traffic and your own prices.
Shortlists work the same way. Adding a weaker third candidate to a hiring round can move the decision toward the candidate who now looks like the compromise. The same happens with vendor quotes, where one clearly inferior offer makes the preferred supplier look reasonable on price.
Personal decisions follow the pattern too. A flat that is clearly worse than the one you want can make the target look like the sensible middle, even when you would have rejected the target alone.
How to reduce it
- Write your criteria before you see the options. Score each option on the attributes that matter for you, not on the set.
- Remove dominated options before comparing. An option that is worse on every attribute does not belong on the shortlist.
- Compare the top two options directly, one pair at a time, and ignore the rest. On a pricing page, convert each plan to the price per unit of the feature you will really use, such as per project, per seat or per thousand records.
- Ask whether you would still pick the chosen option if the third option did not exist. If the answer changes, the decoy did the work.
- If you design the choice, do not use this to mislead. A decoy that is placed to steer people against their own criteria belongs with sludge and dark patterns, not with good choice architecture.
In my view
The decoy effect is real enough to design around, but too conditional to plan revenue on. The original experiments are clean, and the later critique does not show that the effect is an illusion. It shows that the effect depends on details that most pricing pages never test. Research should run preregistered field tests on live pricing pages, with a minimum effect size stated in advance, before anyone builds a tier structure around a predicted lift.
If you run a pricing page or a choice flow and want to know whether a third tier changes what people buy, our applied behavioural science work for businesses starts with exactly that measurement.
Frequently Asked Questions
What is the decoy effect in one sentence?
A third option that is worse than one existing option on every attribute can increase that option’s share of choices, even though nobody is expected to pick the third option itself.
Has the decoy effect replicated?
Partly. The original result has been reproduced in several settings, but Frederick, Lee and Baskin (2014) found it conditional and often small or absent. The original authors disputed that critique. Treat the effect as real under some conditions, not as a universal rule.
Can companies use the decoy effect on purpose?
Companies can place a third tier with the aim of steering choices, and this is common on pricing pages. I found no independent measured effect size for specific tier designs. Treat any claimed lift as a hypothesis to test on your own customers before you rely on it.
References
- Frederick, S., Lee, L., and Baskin, E. (2014). The limits of attraction. Journal of Marketing Research, 51(4), 487-507. https://doi.org/10.1509/jmr.12.0061
- Huber, J., Payne, J. W., and Puto, C. (1982). Adding asymmetrically dominated alternatives: Violations of regularity and the similarity hypothesis. Journal of Consumer Research, 9(1), 90. https://doi.org/10.1086/208899
- Huber, J., Payne, J. W., and Puto, C. P. (2014). Let’s be honest about the attraction effect. Journal of Marketing Research, 51(4), 520-525. https://doi.org/10.1509/jmr.14.0208
- Simonson, I., and Tversky, A. (1992). Choice in context: Tradeoff contrast and extremeness aversion. Journal of Marketing Research, 29(3), 281. https://doi.org/10.2307/3172740
- Trueblood, J. S., Brown, S. D., Heathcote, A., and Busemeyer, J. R. (2013). Not just for consumers: Context effects are fundamental to decision making. Psychological Science, 24(6), 901-908. https://doi.org/10.1177/0956797612464241
- Ariely, D. (2008). Predictably Irrational: The Hidden Forces That Shape Our Decisions. HarperCollins. (Book, no DOI.)