The decoy effect --- adding a worse option to push customers toward a target --- became a foundational SaaS pricing tactic on the strength of Huber, Payne & Puto (1982) and Dan Ariely’s Economist example. In 2014, two large replication efforts found the effect mostly disappeared once you used realistic stimuli. Here is what the lab paradigm actually showed, what failed to replicate, and how to think about your pricing page without leaning on a finding that does not transfer.

A SaaS company opens a pricing-page redesign on a Tuesday morning. Three tiers. The middle one is highlighted, badged “Most Popular,” and priced so that it makes the top tier look like a deal. Somewhere in the deck explaining the design, there is a slide referencing “the decoy effect” --- the behavioral-economics finding that adding a clearly inferior third option to a two-option choice shifts preferences toward the dominating option. The slide cites Dan Ariely. There is probably a sketch of The Economist’s famous subscription page. There is the implicit claim that this is a powerful, well-established behavioral lever --- apply it correctly and conversion rates move.

This is one of the most widely-deployed behavioral-economics tactics in pricing design. It shows up in SaaS plan pages, e-commerce upsells, restaurant menu engineering, gym membership tiers, parking lots, real-estate brochures, almost anywhere multiple options need to be compared. The decoy effect is part of the conversion-optimization canon. Every CRO course covers it. Every pricing-strategy book mentions it.

It is also one of the clearest cases of a behavioral-economics finding whose real-world generalizability collapsed under serious scrutiny --- and where practitioners have continued applying it as if nothing happened.

The original 1982 paper is real. The lab effect is real, in the narrow conditions that produce it. Ariely’s MIT-student demonstration is real, in that the experiment occurred and produced the numbers he reported. What is not real, in the sense of “supported by the empirical record outside the lab paradigm,” is the general claim that adding a strategically-priced decoy tier to a realistic pricing page will reliably shift customer choice. Two large replication efforts in 2014 --- totaling more than 100 attempted demonstrations --- concluded that the effect mostly disappears once the stimuli become realistic, the attributes become qualitative, and the choice resembles something a real consumer would actually face.

This article is about what the lab paradigm showed, what failed to replicate, what the original researchers conceded in their 2014 response, and how to think about your pricing page when one of its foundational behavioral premises does not survive contact with realistic conditions.

What Huber, Payne & Puto 1982 Actually Tested

The founding paper is Huber, Payne & Puto (1982), “Adding Asymmetrically Dominated Alternatives: Violations of Regularity and the Similarity Hypothesis,” Journal of Consumer Research, 9(1), 90-98.

The paper had a specific theoretical target. Standard choice theory included a principle called regularity: adding a new option to a choice set should never increase the choice probability of any existing option. If you preferred A to B when those were your only choices, adding a third option C should at most cause some people to switch to C --- it should not cause anyone to switch toward A. Regularity was a basic assumption in rational-choice models.

Huber, Payne & Puto designed an experiment to test whether a particular kind of third option --- one that was clearly inferior to A on all dimensions but only inferior to B on some dimensions --- would violate regularity by increasing A’s share. They called this third option an “asymmetrically dominated alternative.” A dominates the decoy; B does not. A becomes the “target,” B the “competitor,” and the inferior third option the “decoy.”

The experimental stimuli were highly stylized. Participants made choices among options described by two numeric attributes: cars described by ride quality (e.g., 50 vs. 85 on a scale) and miles per gallon; restaurants described by driving distance and a quality rating; beer described by price and rated quality; lotteries described by probability and prize amount; films and TVs described similarly. The decoy was constructed by taking the target’s attribute pair and slightly worsening both, producing an option clearly inferior to the target but only ambiguously comparable to the competitor.

The headline finding: adding the asymmetrically dominated alternative increased the target’s share by about 9 percentage points on average across the product categories. Regularity was violated. The paper concluded that consumers do not evaluate options in isolation; the structure of the choice set itself influences preference. This was a real theoretical contribution, and it has held up as a description of what happens in tightly-controlled two-attribute numeric-choice paradigms.

Three things about the design are important to remember when evaluating later applications:

  1. The options were abstract --- pairs of numbers attached to a category name, not actual products with brands, images, descriptions, or context.
  2. The attributes were quantitative and comparable on common numeric scales, making dominance relationships clear and computable.
  3. The choice was hypothetical, with no economic consequence to participants.

These conditions matter because, as later replications would show, they are also the conditions under which the effect appears. Move away from any of them --- add pictures, use qualitative descriptions, make the choice consequential --- and the effect tends to shrink or disappear.

The Ariely Popularization And The Economist Example

The decoy effect became a household phrase largely through one book: Dan Ariely’s Predictably Irrational (HarperCollins, 2008), which devoted a chapter to it. The illustrative example was an advertisement Ariely had seen for The Economist magazine offering three subscription tiers:

  1. Web-only access --- $59
  2. Print-only --- $125
  3. Print + web --- $125

The print-only option is the decoy: clearly dominated by the print+web bundle (same price, less value). Ariely’s reported MIT Sloan experiment with 100 students found 16% chose the web-only option and 84% chose the print+web bundle when all three options were shown; when the decoy was removed and only the web-only and bundle options were presented, the split flipped to roughly 68% web-only and 32% bundle. Removing the decoy collapsed bundle preference.

This is a compelling demonstration. It is also worth being precise about what it is and is not. The 100-student MIT Sloan study is one experiment, with a specific population, on a specific product, with a specific dominance structure where the decoy and the target shared an identical price. The famous Economist page itself was an illustration Ariely encountered --- not a controlled marketing experiment The Economist ran and published, and there is no public record of whether The Economist tested the page or what their conversion data actually showed. The example became culturally inescapable, but its empirical basis is one experiment from one chapter of one popular book, not a body of field replication.

Around this single popular example, an entire pricing-design folk wisdom developed. “Add a decoy tier” became standard advice. SaaS pricing pages began showing three plans, with the middle one priced and described to make the top one look like a comparative bargain. Conversion-optimization courses taught the technique. The decoy effect went from a 1982 JCR paper about regularity violation in abstract choice sets to a widely-applied pricing tactic --- without the intermediate step of someone actually testing whether the effect transferred to realistic multi-attribute consumer choices.

That intermediate step finally happened in 2014.

The 2014 Replication Crisis For The Decoy Effect

The August 2014 issue of Journal of Marketing Research contained a remarkable exchange. Two papers presented large-scale replication efforts that failed to find the decoy effect under realistic conditions. The original authors responded. Another paper offered methodological commentary. The volume reads like a deliberate stress test of one of consumer psychology’s most-applied findings.

The first replication paper was Frederick, Lee & Baskin (2014), “The Limits of Attraction,” Journal of Marketing Research, 51(4), 487-507. The authors attempted to demonstrate the attraction effect (another name for the decoy effect) across many product categories and conditions. Their headline finding was that the effect --- robust in the original Huber paradigm with numeric two-attribute stimuli --- largely disappeared when realistic stimuli were used. When products were described qualitatively, when images were shown, when attributes were more complex than two numeric scales, the predicted shift toward the dominating option either failed to appear or sometimes reversed (the decoy could pull share away from the option it was supposed to enhance). The conditions for obtaining the attraction effect, the authors argued, were so restrictive that practical relevance to actual consumer marketing was doubtful.

The second was Yang & Lynn (2014), “More Evidence Challenging the Robustness and Usefulness of the Attraction Effect,” Journal of Marketing Research, 51(4), 508-513. The authors conducted 91 attempts to produce an attraction effect, spanning 23 product categories and 73 different decoyed choice sets. The headline number: only 11 of the 91 attempts produced a reliable effect --- far fewer than would be expected if the effect were a robust general phenomenon. They concluded that qualitative-verbal descriptions and pictorial depictions of products substantially reduce or eliminate the attraction effect, and that the conditions under which the effect reliably appears are essentially the abstract two-attribute numeric paradigm of the original Huber work.

The two papers together represent more than 100 attempted demonstrations across many categories, with a consistent finding: the decoy effect appears in the lab paradigm that generated it and fades or vanishes in conditions closer to actual consumer choice.

The implication for pricing-page design is uncomfortable. The conditions that produce reliable decoy effects --- abstract numeric attribute pairs, no images, no brand context, no economic consequence --- are the opposite of the conditions on a real SaaS pricing page, which has rich qualitative descriptions, visual hierarchy, brand context, and real money at stake. The lab paradigm and the application target are different choice environments, and the effect that holds in one does not appear to reliably hold in the other.

The Original Researchers’ Response (Huber, Payne & Puto 2014)

In the same 2014 issue, the original authors published Huber, Payne & Puto (2014), “Let’s Be Honest About the Attraction Effect,” Journal of Marketing Research, 51(4), 520-525.

The paper is admirable for what it concedes and instructive for what it defends. The authors did not dispute the replication findings. They acknowledged that the attraction effect is, in their words, fragile --- sensitive to stimulus characteristics, attribute structure, and presentation format. They acknowledged that the strong universal version that had filtered into popular marketing advice was overstated.

What they defended was narrower and reasonable: the original 1982 finding was a theoretical contribution about choice processes, not a marketing tactic. It demonstrated that regularity could be violated and that context affects preference --- both of which remain valuable findings about how human choice works. The fact that the effect is fragile to stimulus changes is itself informative about the cognitive mechanisms underlying it: the effect appears to depend on consumers’ ability to easily identify dominance relationships, which is straightforward in abstract two-attribute numeric paradigms and difficult in realistic multi-attribute consumer choices.

The honest version of their position, as best as can be reconstructed from the 2014 response, is something like: the attraction effect is real as a demonstration that context and choice-set composition matter; the effect’s robustness depends on the cognitive accessibility of the dominance relation; in realistic consumer settings where dominance is harder to compute, the effect is small or absent; the popular framing that treats it as a universal pricing-design lever was never warranted by the original work and is not supported by subsequent evidence.

Also in the same issue was Itamar Simonson’s commentary (“Vices and Virtues of Misguided Replications: The Case of Asymmetric Dominance,” 51(4), 514-519), which raised methodological objections to the replication studies --- arguing that some of the failures might reflect imperfect replication choices rather than the absence of the effect. This is a legitimate methodological debate, but it does not undo the central finding that the effect is fragile to exactly the kinds of changes that distinguish a lab paradigm from a real pricing page.

What’s Honest To Say About The Decoy Effect Now

Pulling the strands together, the defensible summary of the decoy effect as of the mid-2020s looks something like this:

Real in the original paradigm. Adding an asymmetrically dominated option to a set of abstract two-attribute numeric choices does, on average, increase the share of the option that dominates the decoy. This has been replicated and is not in doubt.

Fragile to stimulus richness. Across 91 attempts by Yang & Lynn covering 23 product categories with realistic descriptions and images, only 11 produced reliable effects. The effect appears to depend on the cognitive ease of identifying dominance, which falls sharply when attributes are qualitative, when images are involved, or when products have multiple non-comparable features.

Sometimes reverses with realistic stimuli. Frederick, Lee & Baskin documented cases where the decoy actually reduced the share of the option it was supposed to enhance. The directional certainty that pricing-page advice assumes --- “add a decoy and the target goes up” --- is not supported.

Probably small or absent on a typical pricing page. A SaaS pricing page has rich qualitative descriptions, feature lists, visual design, brand context, and real economic stakes. None of these are conditions under which the effect has been reliably demonstrated. The transfer from the lab paradigm to the pricing-page application is the kind of inference the 2014 replications specifically argued against.

The Ariely Economist example is one experiment. With a specific population (MIT Sloan students), a specific product, a specific dominance structure (decoy and target at identical price). It is suggestive but not the kind of replicated multi-context finding that would warrant the “powerful general lever” framing it has been given. Whether The Economist’s actual page tested better with the decoy is not part of the public record.

This is not “the decoy effect is fake.” It is “the decoy effect is a fragile lab finding that has been over-extrapolated into a confident pricing-design heuristic that the replication evidence does not support.”

What This Means For Your Pricing Design

The actionable translation, if you have been designing pricing pages around the decoy effect:

Stop assuming the decoy tier is doing work. There is no evidence-based reason to expect that adding an obviously-inferior tier to your three-plan pricing page is reliably shifting customer choice toward the tier it “dominates.” The effect that produces that prediction has not survived the conditions of a real pricing page.

This does not mean three-tier pricing is bad. Three tiers have lots of legitimate reasons to exist --- they let customers self-segment by needs, they create natural upgrade paths, they make the middle tier feel like a default choice --- none of which depend on the decoy effect. Keep three tiers if three tiers work for your customer base. Do not justify the structure with a behavioral lever the empirical record does not support.

If you want to anchor customer choice toward a specific tier, the mechanisms that have more empirical support than the decoy effect include:

  • Visual hierarchy and prominence. Making one tier visually larger, more colorful, or more central does shift attention and choice. This is well-supported by attention research and is testable directly.
  • “Most Popular” labels as social proof. Indicating that a tier is widely chosen leverages social proof --- a more robust behavioral effect than asymmetric dominance, with broader empirical support across realistic settings.
  • Default selection. Pre-selecting a tier (where allowed by your UX) leverages the default effect, which has substantially more replication support than the decoy effect (though see the caveats in the defaults article in this hub).
  • Annual pricing display prominence. If your annual plan is your high-margin product, displaying it as the default toggle creates a frame the customer must actively reject.
  • Clear feature-tier alignment. If the tier you want customers to choose is the tier whose feature set best matches their stated needs, well-designed feature lists do more work than any decoy ever did.

The honest framing is: anchor effects, social proof, and default effects have better empirical support than asymmetric dominance, and they probably do most of the work people attribute to “the decoy effect” on real pricing pages. Build around those.

How To Actually Test This On Your Pricing Page

The other actionable translation: stop reasoning from the lab paradigm and start testing your specific page.

The decoy effect’s collapse outside controlled lab conditions is a powerful argument for not trusting behavioral-economics generalizations to translate cleanly to your conversion environment. The way to know whether a pricing-page change moves conversion is to A/B test it on your customers, in your product, at your price points. Not to derive a prediction from a 1982 paper and assume it transfers.

Practical structure:

  1. State the hypothesis specifically. Not “we’ll use the decoy effect” --- “if we add a $99 print-only-equivalent tier, the share of the $125 bundle will rise by at least X percentage points.” Pre-commit to the prediction.
  2. Run the experiment with adequate sample size. Use a standard power calculation. Practical-significance thresholds matter more than statistical significance --- if you need 50,000 visitors per arm to detect a 1-point conversion lift, that lift is probably not worth the design complexity.
  3. Measure end-to-end revenue, not just plan selection. A decoy tier that shifts plan mix toward the high tier but reduces overall conversion is a loss. Selection share is not revenue.
  4. Watch for novelty effects. Initial lifts often fade as the page becomes familiar to repeat visitors or as winning copy gets absorbed across pages. Re-test winners after 90 days.
  5. Don’t generalize. A decoy that works on your pricing page does not mean decoys work in general. It means it worked on yours, in that period, for that traffic. The lesson is local.

If your A/B testing infrastructure can support this kind of test, you no longer need to defer to a fragile lab finding. If it cannot, the most useful thing you can build for your conversion stack is the ability to test, not the ability to cite Huber 1982.

Sources

  • Huber, J., Payne, J. W., & Puto, C. (1982). Adding asymmetrically dominated alternatives: Violations of regularity and the similarity hypothesis. Journal of Consumer Research, 9(1), 90-98. DOI: 10.1086/208899
  • Ariely, D. (2008). Predictably Irrational: The Hidden Forces That Shape Our Decisions. HarperCollins. (Chapter 1, including the Economist subscription example and the MIT Sloan student experiment.)
  • Frederick, S., Lee, L., & Baskin, E. (2014). The limits of attraction. Journal of Marketing Research, 51(4), 487-507. DOI: 10.1509/jmr.12.0061
  • Yang, S., & Lynn, M. (2014). More evidence challenging the robustness and usefulness of the attraction effect. Journal of Marketing Research, 51(4), 508-513. DOI: 10.1509/jmr.14.0020
  • Simonson, I. (2014). Vices and virtues of misguided replications: The case of asymmetric dominance. Journal of Marketing Research, 51(4), 514-519. DOI: 10.1509/jmr.14.0093
  • Huber, J., Payne, J. W., & Puto, C. P. (2014). Let’s be honest about the attraction effect. Journal of Marketing Research, 51(4), 520-525. DOI: 10.1509/jmr.14.0208

FAQ

Should I remove the “decoy” tier from my pricing page? Not necessarily. The empirical evidence does not support the claim that the tier is doing meaningful decoy work, but three-tier pricing has other legitimate reasons to exist --- customer self-segmentation, natural upgrade paths, defaults. The actionable change is to stop justifying the tier with the decoy effect and start testing whether your specific pricing structure actually maximizes revenue. Remove the tier if your A/B tests say it is not earning its place; keep it if it is, regardless of whether the decoy-effect story is right.

Does ANY pricing-design heuristic actually work? Some have better empirical support than others. Anchoring effects (showing a high reference price near a target price) have replicated reasonably well. Default selection has robust support. “Most popular” social-proof labels have support. Visual hierarchy and attention-direction effects are well-documented. The general pattern: simpler mechanisms (attention, social proof, defaults) replicate better than complex relational mechanisms (asymmetric dominance, contrast-induced shifts). Build on the simpler-mechanism side.

What about anchoring? Is that also a fragile finding? Anchoring has more replication support than asymmetric dominance, though the strongest forms of the original Tversky-Kahneman anchoring claim have also faced scrutiny. The relevant distinction is between numerical anchors (where a reference price genuinely shifts willingness to pay --- broadly replicated) and the stronger claim that any irrelevant number can anchor (more contested). For pricing design, numerical reference points (showing a struck-through “was $X” or comparing to a competitor price) have reasonable empirical support.

What about The Economist case specifically? The Ariely demonstration with MIT Sloan students is one experiment, with a specific population on a specific product. It produced a striking shift in stated preferences (68/32 to 16/84 when the decoy was added). It is suggestive evidence that the decoy effect can occur in realistic pricing contexts under some conditions. It is not, by itself, sufficient to claim the effect is a reliable general tactic. The Economist itself has not published the conversion data from its actual subscription page, so whether the famous pricing structure outperformed alternatives in the field is not part of the public record.

Why did this finding get so much practitioner traction if the evidence was fragile? Behavioral-economics findings that produce clean, counterintuitive, repeatable stories tend to escape the literature into popular marketing advice. The decoy effect has all of those properties --- clear demonstration, surprising violation of rational-choice intuitions, easy to apply. The replication evidence came late (2014), in a marketing-research journal that most practitioners do not read, and required understanding the difference between lab paradigm and field application. The replication signal has not reached the practitioner community with anything like the force the original story had.

What’s the difference between “the decoy effect doesn’t work” and “the decoy effect is fragile”? The honest version is the latter. The effect appears reliably in the original abstract two-attribute numeric paradigm and largely disappears in realistic multi-attribute consumer choices. So the finding is not fake --- it is real in the conditions that produce it. The error is the extrapolation: assuming that a fragile lab effect transfers to environments that lack the conditions under which it appears. The replication-era reading of this kind of finding is “stop generalizing from controlled paradigms to applied contexts without evidence the effect transfers.”

Are there any pricing contexts where the decoy effect probably does work? Possibly contexts that resemble the original paradigm: choices among quantitatively-comparable options with easy-to-compute dominance relations, low cognitive richness, low stakes, and limited brand context. Wine lists with two-attribute (price, rating) descriptions might be closer to the lab paradigm than a SaaS pricing page is. Even there, the effect is likely small. The general principle is that the lower the cognitive cost of computing dominance, the more likely the decoy mechanism is to operate --- but realistic high-stakes purchase decisions almost always involve more dimensions and more context than the conditions under which the effect reliably appears.

How should I update my mental model of behavioral-economics findings in general? The decoy effect’s trajectory --- lab finding, popularization, applied tactic, late replication failure --- is a recurring pattern across behavioral economics. The defensible posture is: treat lab demonstrations as evidence that an effect can occur under specific conditions, not as evidence it operates generally; demand replication evidence in conditions resembling your application context before designing around a finding; A/B test your specific application rather than reasoning from analogy; and read the original papers’ conditions, not the popular summaries. The replication crisis is, in part, a story about what happens when this discipline is absent.

replication-crisis decoy-effect pricing-design behavioral-economics evidence-evaluation

Free Tool

Built for Experimentation Teams

GrowthLayer is the experimentation platform I built for CRO teams --- test management, AI-powered insights, and pattern recognition across your entire program.

Explore GrowthLayer → (opens in new tab)

· Start Free →

Share this article

LinkedIn (opens in new tab) X / Twitter (opens in new tab)

Copy link

Go deeper

Methodology The PRISM Method Case Studies $30M+ in Results Work Together Services & Mentoring

Experimentation and growth leader. Builds AI-powered tools, runs conversion programs, and writes about economics, behavioral science, and shipping faster.

About LinkedIn Newsletter

← Previous

Foot-In-The-Door: The Persuasion Tactic That Works (Just Less Than Marketers Think)

Next →

Door-In-The-Face: The Persuasion Technique That Actually Replicates (With Specific Conditions)

replication-crisis decoy-effect pricing-design behavioral-economics evidence-evaluation
Share this article
LinkedIn (opens in new tab) X / Twitter (opens in new tab)
Atticus Li

Experimentation and growth leader. CXL-certified CRO practitioner, Mindworx-certified behavioral economist (1 of ~1,000 worldwide). 200+ A/B tests across energy, SaaS, fintech, e-commerce, and marketplace verticals.

About LinkedIn Newsletter