The anchoring effect

In short

A jacket’s tag shows $400 struck through, with $190 below it. If you don’t know what the jacket is worth, $190 may seem cheap. After all, your only point of comparison is the $400 you saw first.

That first number is called the anchor, and this is the anchoring effect. The estimate starts at the anchor and usually stays close to it.

In interfaces, you see this effect most easily on pricing pages. Below I first describe two experiments, then the honest and the manipulative use of anchors.

Tversky and Kahneman’s wheel-of-fortune experiment

Amos Tversky and Daniel Kahneman described the concept in 1974 in an article in the journal Science. Its title is “Judgment under Uncertainty: Heuristics and Biases”.

The article covers three mental shortcuts people use when judging under uncertainty. One of them is adjustment from an anchor. According to the authors, people often take a starting value and reach their answer by adjusting it.

This adjustment is typically insufficient. That is why different starting points lead to different estimates, and the estimates are biased toward the starting value.

In the experiment, participants were asked to estimate various quantities stated as percentages. One example was the percentage of African countries in the United Nations. For each quantity, a wheel of fortune was spun in front of them, giving a number between 0 and 100.

Participants first said whether this number was higher or lower than the true value. Then they gave their estimate by moving up or down from that number. The group that saw 10 on the wheel gave a median estimate of 25%; the group with 65 estimated 45%.

The number from the wheel had nothing to do with the question and still clearly affected the estimates. According to the article, paying for accurate estimates didn’t reduce the effect either.

A second experiment shows that the anchor doesn’t have to come from outside. Two groups of high school students estimated, within 5 seconds, a product written on the blackboard. One group saw the sequence 8×7×6×5×4×3×2×1, the other the same numbers from 1 to 8 in ascending order.

For the descending sequence the median estimate was 2,250, for the ascending one 512. The correct answer is 40,320.

The authors’ explanation: people answering quickly compute the first few steps and estimate the rest. Because the adjustment is insufficient, both groups stayed far below.

Where the anchor sits in the interface

Someone judging a price often has no yardstick for it. Is $59 a month for an invoicing app expensive? To know, the person looks for a point of comparison and takes the first one they find on the screen.

The point of comparison can come from several places: the struck-through former price, the most expensive plan in the table, or a bundle with a “total value”. A preselected amount on a donation form can play this role too. All of them put a number in the person’s head.

The anchor itself isn’t the problem. Giving a point of comparison makes the decision easier. The problem begins when the number serving as the anchor isn’t real.

Live example: the same pricing table, three states

The table below belongs to a fictional invoicing app. The Standard plan in the middle costs $59 a month in every state. Only the numbers you see around that price change.

The same price, three states

1. No anchor

Starter

$29 / month

• 1 user
• 50 invoices a month
• Email support

Standard

$59 / month

• 3 users
• 500 invoices a month
• Bank account reconciliation
• Email and phone support

Business

Request a quote

• Unlimited users
• Unlimited invoices
• Custom integration
• Dedicated contact

What does this table make you think?

Your only point of comparison is the Starter plan at $29. Standard costs a little more than twice as much and can therefore look expensive. The Business plan shows no price, so you don’t know the top of the price range. You have little information to judge $59.

2. Honest anchor

Business

$149 / month

• Unlimited users
• Unlimited invoices
• Custom integration
• Dedicated contact

Standard

Price before the reduction: $69

$59 / month

$10 off Since October 1, 2026. $69 was the price right before the reduction.

• 3 users
• 500 invoices a month
• Bank account reconciliation
• Email and phone support

Starter

$29 / month

• 1 user
• 50 invoices a month
• Email support

What does this table make you think?

Your eye lands first on $149. That is a real plan you can buy. Next to it, $59 looks reasonable. The struck-through $69 is real too: the date is shown, and it was the price right before the reduction. There is an anchor, but one you can check.

3. Fake anchor

Starter

$29 / month

• 1 user
• 50 invoices a month
• Email support

Standard

$147

$59 / month

60% off This price is for today only!

• 3 users
• 500 invoices a month
• Bank account reconciliation
• Email and phone support

Business

Request a quote

• Unlimited users
• Unlimited invoices
• Custom integration
• Dedicated contact

What does this table make you think?

The struck-through $147 makes you believe you are saving $88. Yet in this demo the Standard plan was never sold for $147 on any day. The price was $59 from the start and will still be $59 tomorrow. No date, no source. The anchor works, but with a made-up number.

Prices in US dollars, billed monthly, plus applicable tax.

The pricing table of a fictional invoicing app; the plan names, prices and dates are made up. Switch the state of the table with the buttons. The Standard plan costs $59 a month in every state. Below the table, you read what it suggests to you.

In all three states you pay the same amount. In the second and third states, $59 looks cheaper. The difference: you can check the numbers in the second state, but not those in the third.

Honest use in the interface

There are several ways to use an anchor honestly on a pricing page. The same condition applies to all of them: behind the number serving as the anchor, there must be a product that is sold or a price that was actually charged.

• Plan order. You can put the most expensive plan at the start of the table or next to the recommended plan. If it is really sold, that is information: the person sees the top of the price range.
• Real former price. If the discount is real, show the old price with a date. The sentence “$69 until October 1” can be checked.
• Bundle comparison. Next to the bundle price, write what the parts would cost separately. If every item in the total is also sold on its own, the anchor holds.

In plan tables you may also meet another effect: the decoy effect. I describe it with a live table in the pricing page pattern.

Manipulative use: the fake strikethrough price

The most common manipulation is a struck-through price the product was never sold at. Another form: raise the price shortly before the sale, then “reduce” it. In both cases, the person counts on a saving that doesn’t exist.

Designs like these are called dark patterns. I have collected the types in the guide to dark patterns.

A fake anchor rarely comes alone. It comes with a countdown or the sentence “14 people are looking at this right now”. The latter is an example of made-up social proof. The anchor supplies the number; the fake crowd supplies the time pressure.

Rules for former price comparisons

How a former price may be used in an ad is regulated in many places. What applies to your offer depends on the country, the type of offer and whether you sell goods or services.

In the US, the FTC’s Guides Against Deceptive Pricing cover this in 16 CFR 233.1, “Former price comparisons”. According to the guide, a former price is a legitimate basis for a price comparison if it is the actual, bona fide price at which the article was offered to the public on a regular basis for a reasonably substantial period of time.

The same section describes the opposite case: an artificial, inflated price set so that a large reduction can be advertised afterwards. Its example is a retailer who sells pens at $7.50, raises them to $10 for a few days, then advertises “Were $10, Now Only $7.50!”. The guide calls this a false claim.

In the EU, Article 6a of Directive 98/6/EC applies. Any announcement of a price reduction must state the prior price, meaning the lowest price the trader applied during a period of at least 30 days before the reduction. Member States can set different rules for some cases, such as goods that spoil quickly.

Clarify the legal situation with a specialist; this note isn’t legal advice.

For design, this means: the campaign team doesn’t choose the struck-through number; it comes from the price history. Keep that history so you can prove the former price.

How to check your own pricing page

1. For every struck-through price, ask: was this price actually charged before the reduction, and can you prove it? If there is no proof, remove the number.
2. Put the start date next to the discount label. A short note such as “since October 1” is enough.
3. Check whether the most expensive plan in the table can really be bought. A plan that is only there to make the others look cheap is a decoy.
4. For bundles “worth $X”, check the individual price of each item.

For the structure of the price block on a product page, see the product page pattern. More principles are in the list of laws of UX.

My note

When I look at a pricing page in an audit, I first note the struck-through numbers. Then I ask where each one comes from. If the answer is “we set it for the campaign”, that is a finding.

An honest anchor helps; I don’t advise hiding it. Instead of hiding the most expensive plan, put it in the table with its real scope. When the person sees the price range, the decision gets easier.

You pay the price for a fake anchor later. When the person sees the same product on another website at the same price, they stop trusting the label. After that, they view the other numbers on the website with suspicion too.

If you want your pricing page checked from outside, the UX Audit covers this check too.