011 min
Where the compromise effect shows up
Picture a product manager at a subscription app for meeting notes. The app sells two plans: basic at $10 a month and plus at $20 a month. The numbers in this example are made up. In the pricing test, 65% of buyers take basic and 35% take plus.
The manager wants more revenue from the same visitors. The team adds a third plan, pro, at $40 a month. Nobody on the team expects many people to buy it. After the change, 40% of buyers take basic, 50% take plus and 10% take pro.
Plus did not change. Its price, its features and its page are the same as before. Only the set around it changed. Plus used to be the most expensive plan, so a buyer could treat it as the extreme choice. Now it is the middle plan, and more buyers choose it.
The same thing happens whenever options can be put in order on one trait, such as price or size, and one option falls between the others. Shoppers, survey respondents and software buyers all show it.
021 min
Where the compromise effect comes from
Researchers measured this pattern in shoppers long before subscription pricing pages existed. In 1989 the marketing researcher Itamar Simonson argued that people who are unsure what they want tend to pick the option that is easiest to justify, to themselves and to other people.
From that idea he predicted a compromise effect, and his experiments supported the prediction. Brands gained share when they became the compromise option in a set. The effect was stronger among people who expected to explain their decision to others.
In 1992 Simonson and Amos Tversky gave the pattern a name, extremeness aversion. Their rule says an option becomes more attractive when it is in the middle of the set and less attractive when it is at an end.
Their best-known test used 35 mm cameras. Participants saw a Minolta X-370 at $169.99 and a Minolta Maxxum 3000i at $239.99. The two cameras were chosen equally often, 50% each. Other participants saw the same two cameras plus the top model, a Minolta Maxxum 7000i, at $469.99. Now 22% chose the cheapest camera, 57% chose the middle one and 21% chose the top one.
Among people who picked one of the two original cameras, the share choosing the middle camera rose from 50% to 72%. Nothing about that camera had changed.
032 min
Why the compromise effect happens
The camera result raises a question. Why would adding a camera that most people did not choose change how they judge the other two?
Simonson and Tversky started from loss aversion. Loss aversion is the finding that people weigh a loss more heavily than a gain of the same size. The two authors applied the idea to the comparison between options in a set, not to a gain or loss measured from a neutral starting point. Each option has advantages and disadvantages next to the others, and the disadvantages count for more.
Now look at the three cameras. The cheapest one is behind the others on quality. The top one is behind the others on price. The middle camera is never the worst on either trait, and it is not far from either end. Its largest disadvantage is smaller than the largest disadvantage of the two extremes. A buyer who dislikes disadvantages most will choose it.
A second reason is justification. Simonson found in 1989 that the effect was stronger when buyers expected to explain their choice. A middle option is easy to defend. It is not the cheapest, so it is not a risk on quality. It is not the most expensive, so it is not a waste.
A third reason is doubt. A buyer who does not know what they want can treat the set as a guide to what is reasonable. These explanations can all be true at once, and the studies cited here do not show which one matters most.
041 min
Tradeoff contrast and attribute balance in the compromise effect
Tradeoff contrast
Position is not the only force. Simonson and Tversky also described tradeoff contrast. A trade-off is what a buyer gives up on one trait to gain on another, such as paying more for better quality.
With two cameras there is one trade-off to judge. With three cameras there are three, and the buyer can compare them with each other. In the camera test the step from the cheapest camera to the middle one costs $70. The step from the middle one to the top one costs $230. Next to the second step, the first looks like a small price for the extra quality, so the middle camera looks like good value.
Computers and tires
The 1992 paper did not rely on cameras alone. The authors also tested the idea on personal computers that differed in memory and price, and on tires that differed in warranty length and price.
Attribute balance
Alexander Chernev added a further idea in 2005. His two experiments showed that attribute balance changes how strong extremeness aversion and tradeoff contrast are. Attribute balance means how evenly an option scores on each trait. A middle option is often moderate on every trait. So part of its appeal may come from being balanced, and not only from where it sits in the list.
051 min
What the compromise effect changes in markets and research
Sellers decide which options exist, so the effect reaches further than one shopper.
Soft drink sizes
Kathryn Sharpe, Richard Staelin and Joel Huber studied drink sizes in 2008. In three studies, each participant chose from several sets of sizes. People bought more when the smallest size was removed or a larger size was added. The old middle size became the smallest one, so it turned into an extreme and buyers moved up. The authors name extremeness aversion together with low sensitivity to price as the cause.
Lists in research
Researchers are affected too. In a 2015 working paper, a team led by Jonathan Beauchamp tested 550 people on a multiple price list. This is a list of rows in which a person picks the row where they switch from one choice to another. The team changed which options sat in the middle of each list. They report that estimates of risk preference are biased unless the analysis controls for the compromise effect.
062 min
How the compromise effect shows up in product, pricing and design
Back in the notes app, the manager can now read the pricing page differently. A team that lists a basic, a plus and a pro plan usually puts the plan it most wants to sell in the middle, and sets the top price high. The top plan does a job even if few people buy it. It makes the middle plan the compromise. Choosing which options a buyer sees, and in what order, is called choice architecture.
The effect can also work by accident. Say a hardware team offers three storage sizes because that is how the chips are sold. The middle size sells best, and the team reads this as proof that customers want the middle size. They plan the next release around it. But the sales may describe the list and not the need. If the team removed the biggest size, the middle size might lose its place.
Analysts meet the same problem. A product analyst who sees that most customers pick the middle plan should ask two questions. Would customers still pick it if the top plan were gone? Do the plan shares describe what people need, or only the shape of the list?
Designers meet it when they choose how many sizes or tiers to show. Each option added to a list changes which option is in the middle. That makes the list itself a design decision, and not a neutral display of the product.
072 min
How to guard against the compromise effect
A buyer or a team that wants to avoid the effect has to change the process, because the effect comes from comparing options with each other.
Write the need first.
Before opening the pricing page, write what you need: seats, storage, support hours. Then check each plan against that list alone. The needs do not change when a new plan appears, so the context cannot move them.
Remove one option and ask again.
Cover the top option and decide. Cover the bottom option and decide again. If your answer changes, the set was making the choice for you.
Score before you see prices.
A team picking a vendor can have each reviewer score the capabilities against a fixed list first. Price is added afterwards. The scores cannot be shifted by which vendor ends up in the middle on price.
These steps change how the choice is made, instead of asking people to notice a bias in themselves.
A seller who wants to use the effect fairly should test it. Show similar visitors the page with and without the top plan, then compare revenue per visitor. The share choosing the middle plan can rise while revenue stays flat, if the people who would have bought the top plan move down.
The fairness test is simple. Could a real buyer need the top plan? If not, the plan exists only to move buyers, and that is manipulation.
081 min
When the compromise effect does not hold
The camera test asked people to choose, but nobody paid. Does the effect survive real money?
Binding choices
Holger Müller, Benjamin Kroll and Bodo Vogt tested this in 2010. Their experienced consumers made unforced, binding purchases of real brands. The effect still appeared, but it was clearly smaller than in the hypothetical choices. The authors conclude that earlier studies on hypothetical decisions overstated the effect to a degree.
Durable goods
In 2016 Marcel Lichters and colleagues found that the effect also held for durable goods. The buyers paid real money for real brands, could evaluate the products before choosing, and could decide to buy nothing.
The middle is not the best
A common reading is that the middle option wins because it fits buyers best. The camera data say otherwise. The middle camera was a coin flip against the cheapest camera in Set 1. It became the favourite only when the top camera was added. Its price and quality had not changed, so its position made the difference.
Where it does not apply
The options must trade off against each other. If one option is better on every trait, there is no compromise to make, and a different effect applies. The effect is also useful for buyers whose real need is the middle option. Guarding against it would be a mistake for them.
091 min
The compromise effect vs. the decoy effect and nearby concepts
Simonson's 1989 reasoning explained the attraction effect, which is the decoy effect, and it also predicted the compromise effect. The two differ on one axis. A decoy is an option that is worse than another option on every trait. The formal name for that kind of option is an asymmetrically dominated choice. The compromise effect needs no dominated option. It needs only an option that sits between extremes.
| Concept | What is added or shown | What shifts the choice |
|---|---|---|
| This pageCompromise effect | ||
| PsychologyDecoy effect | ||
| MarketingPrice anchoring | ||
| PsychologyChoice overload |
Price anchoring sets a reference with the first price a buyer sees. A pricing page can use both: a high first price makes the plans below it look cheaper, and a middle plan is chosen as the compromise. Choice overload is a different problem. It concerns the number of options, while the compromise effect concerns their positions.
?3 questions
Questions people ask
Does the compromise effect need exactly three options?
How can a seller use the compromise effect fairly?
What is the difference between the compromise effect and extremeness aversion?
§7 sources
Sources
Show all 7 sourcesHide sources
Simonson, I. (1989). Choice Based on Reasons: The Case of Attraction and Compromise Effects. Journal of Consumer Research 16(2), 158-174. DOI 10.1086/209205
Simonson, I. and Tversky, A. (1992). Choice in Context: Tradeoff Contrast and Extremeness Aversion. Journal of Marketing Research 29(3), 281-295
Chernev, A. (2005). Context Effects without a Context: Attribute Balance as a Reason for Choice. Journal of Consumer Research 32(2)
Sharpe, K. M., Staelin, R. and Huber, J. (2008). Using Extremeness Aversion to Fight Obesity: Policy Implications of Context Dependent Demand. Journal of Consumer Research 35(3), 406-422. DOI 10.1086/587631
Beauchamp, J. P., Benjamin, D. J., Chabris, C. F. and Laibson, D. I. (2015). Controlling for the Compromise Effect Debiases Estimates of Risk Preference Parameters. NBER Working Paper 21792
Müller, H., Kroll, E. B. and Vogt, B. (2010). Fact or artifact? Empirical evidence on the robustness of compromise effects in binding and non-binding choice contexts. Journal of Retailing and Consumer Services 17(5), 441-448 (record page)
Lichters, M., Müller, H., Sarstedt, M. and Vogt, B. (2016). How durable are compromise effects? Journal of Business Research 69(10), 4056-4064 (record page)





