The Compromise Effect: Why People Choose the Middle Option

The tendency to choose the middle option in a set is not a preference. It is a decision strategy for people who do not know what they want.

Anika van der Berg · May 15, 2026

The Middle Is Not a Preference

If you have ever designed a pricing page with three tiers and noticed that the middle tier gets the most sign-ups, you have observed the compromise effect. You may also have concluded that you designed the tiers well — that the middle option represents the best value and customers are rationally selecting it. This conclusion is almost certainly wrong, or at least incomplete. The middle tier is not chosen because it is the best. It is chosen because it is in the middle.

The compromise effect, first formally described by Itamar Simonson in 1989, refers to the tendency for options that represent a compromise between extremes to gain market share when the set of available options expands. In its simplest form: given a choice between a cheap, low-feature option (A) and an expensive, high-feature option (B), people distribute their choices roughly evenly. But add a third, even more expensive option (C), and suddenly B — now the middle option — sees a significant increase in share, often at the expense of A rather than proportionally from both extremes.

This is a violation of what economists call "regularity" — the principle that adding a new option to a choice set should not increase the share of an existing option. If people had stable, well-defined preferences, introducing C should only take share away from A and B, never add to either. The fact that B gains share when C is introduced tells us something important: people's "preferences" are partly constructed from the choice context itself.

The Mechanism: Loss Aversion in Attribute Space

Why does the compromise effect occur? Simonson and Tversky proposed an explanation grounded in loss aversion. When choosing between options that differ on multiple attributes — say, price and features — selecting an extreme option means accepting a large loss on one attribute in exchange for a large gain on another. The cheap option saves money but sacrifices features. The expensive option maximizes features but costs more. The middle option involves moderate losses and moderate gains on both dimensions.

Because losses loom larger than equivalent gains — this is one of the foundational findings of Kahneman and Tversky's prospect theory, which I explore in detail in my essay on framing effects — the pain of accepting a large loss on any single dimension outweighs the pleasure of the corresponding gain. The middle option minimizes maximum loss, making it feel like the safest choice. People are not choosing it because they calculated that it offers the best value. They are choosing it because it feels like the least risky decision.

This is a subtle but important distinction. The compromise effect is not about value perception. It is about risk avoidance in multi-attribute choice. The middle option is selected because people are afraid of making a mistake, and the middle feels like the place where mistakes are smallest.

Boundary Conditions: When It Works and When It Does Not

The compromise effect is one of the more robust findings in behavioral decision theory, but it has meaningful boundary conditions that are often ignored in its application to pricing.

First, the effect is strongest when consumers lack expertise. Research has consistently found that domain experts are less susceptible to the compromise effect than novices. When people know what they want — when they have clear attribute preferences and can articulate trade-offs — the pull toward the middle weakens. This has direct implications for SaaS pricing: if your buyers are experienced purchasers of your category (e.g., a marketing director buying their fifth email automation tool), the compromise effect will be weaker than if your buyers are first-time purchasers navigating an unfamiliar category.

Second, the effect depends on attribute alignability. The compromise effect requires that options differ on attributes that can be naturally ordered and compared. When attributes are alignable — more storage, more users, more integrations — the compromise effect is strong because people can easily see that the middle option occupies a middle position on each dimension. When attributes are non-alignable — when different plans offer qualitatively different features rather than more or less of the same features — the compromise effect weakens because there is no clear "middle" to gravitate toward.

Third, justifiability matters. Research suggests that the compromise effect is amplified when people anticipate having to justify their decision to others. In studies where participants were told they would need to explain their choice, the preference for the middle option increased. The middle is easy to justify: "I didn't overspend, and I didn't go too cheap." Extreme choices require more elaborate justifications, and people generally prefer to avoid that cognitive work.

This finding is particularly relevant for B2B purchases, where buying decisions are almost always subject to scrutiny by managers, procurement teams, or budget owners. A B2B buyer choosing a SaaS plan may gravitate toward the middle tier partly because it is the easiest to defend in an internal meeting. "I chose the Professional plan — it has everything we need without the bells and whistles we'd be paying for in Enterprise" is a sentence that writes itself. Justifying the cheapest or most expensive option requires more effort.

The Three-Tier Pricing Page as Choice Architecture

Most SaaS pricing pages present exactly three tiers, and this is not a coincidence. The three-tier structure is designed to exploit the compromise effect, usually with the explicit goal of driving users toward the middle option. The middle tier often carries visual emphasis — a "Most Popular" badge, a highlighted border, a different background color — which serves as a nudge layered on top of the compromise effect.

But the design of those three tiers determines whether the compromise effect actually activates. I have reviewed pricing pages for dozens of SaaS companies over the past several years, and I see the same design errors repeatedly.

The most common error is making the middle tier too similar to the cheap tier. When the feature difference between the bottom and middle tiers is small and the price difference is significant, the middle tier does not feel like a compromise — it feels like a bad deal. Users compare the two most similar options, conclude the cheaper one is sufficient, and select it. The compromise effect requires that the middle option be clearly differentiated from both extremes, not just from one.

The second common error is pricing the top tier out of consideration entirely. When the Enterprise tier costs ten or twenty times more than the middle tier, or when it replaces a price with "Contact Sales," it ceases to function as the upper bound of a compromise set. Users mentally exclude it and treat the remaining two options as a binary choice, which eliminates the compromise effect. For the effect to work, all three options must feel like they are "in play" — all must be plausible choices for the user evaluating them.

A third error is offering too many tiers. Research on the compromise effect has primarily examined three-option sets. When the number of options increases to four, five, or more, the dynamics change. With more options, there are multiple "middle" positions, and the clarity of the compromise diminishes. Studies have found that the compromise effect weakens as the number of options increases, which aligns with research on choice overload more broadly.

An Experiment That Changed My Thinking

In 2023, I worked with a project management software company that was dissatisfied with its pricing page performance. They had three tiers: Starter at $12/month, Professional at $29/month, and Enterprise at $79/month. The Professional tier — their target — was getting about 40% of sign-ups, with Starter getting 45% and Enterprise getting 15%.

The conventional advice would have been to adjust the middle tier — perhaps by manipulating anchoring through price ordering, or by adding a decoy option —'s visual prominence or add a "Most Popular" badge. Instead, we looked at the attribute structure. The gap between Starter and Professional was substantial — Professional added team collaboration, integrations, and reporting. But the gap between Professional and Enterprise was mostly about scale (more users, more storage) and premium support. For most of their target users — teams of 5 to 15 people — the Enterprise features were irrelevant.

We restructured the tiers. The new Enterprise tier was priced at $49/month (down from $79) and included features that small teams might plausibly want — priority onboarding, advanced analytics. The goal was to make Enterprise feel like a realistic option rather than an abstract aspiration. The new structure made the Professional tier a true compromise between two options that both felt relevant.

The result, measured over eight weeks with roughly 6,200 unique visitors to the pricing page, was that Professional's share increased from 40% to 52%. More interesting, total revenue per visitor increased by 18%, because the lower-priced but more accessible Enterprise tier attracted users who previously would have chosen Professional. The middle option converted better, and the top option converted better too, because both were now plausible choices.

The lesson was instructive. The compromise effect does not require that the extreme options be bad. It requires that they be real — real options that real users might genuinely consider. When extremes are implausible, the compromise has no anchor points to work between.

The Ethical Dimension

There is an ethical question embedded in the compromise effect that deserves attention, even if it rarely receives it. If people choose the middle option not because they have evaluated it and found it best, but because the choice architecture made it feel safest, are they making a good decision?

Sometimes, yes. If the middle option genuinely represents the best value for most users — and if it was designed with that in mind — then the compromise effect is nudging people toward a good outcome. This is the benign case, and it is common.

But sometimes the middle tier is the target precisely because it has the highest margin, not because it offers the best value. The extreme options may be deliberately designed to be unattractive — a Starter plan stripped of essential features, an Enterprise plan loaded with features nobody needs — to funnel users toward a middle option that is profitable for the company but not optimal for the user. In this case, the compromise effect is being weaponized, and the user is making a worse decision than they would have made with better information or fewer options.

I do not think this distinction is merely academic. Companies that exploit the compromise effect to sell overpriced middle tiers generate short-term revenue but long-term problems: higher churn when users realize they are paying for a plan that does not match their needs, lower satisfaction, and the slow erosion of trust that comes from customers feeling like they were nudged rather than helped.

Caveats and Limitations

The compromise effect is well-documented in laboratory studies, but field evidence is thinner than the literature's reputation suggests. Most demonstrations use hypothetical choices or low-stakes consumer goods. The transfer to high-consideration SaaS purchases, where buyers may spend weeks evaluating options and consulting with colleagues, is plausible but not well-established empirically.

Additionally, the effect interacts with many other factors on a real pricing page — social proof, visual design, feature descriptions, competitive comparisons — and isolating its independent contribution is difficult. The case study I described above involved multiple changes to the pricing page, and attributing the improvement solely to the compromise effect would be an overstatement. What I can say with confidence is that the attribute structure of the tiers mattered at least as much as, and probably more than, the visual design.