Note: This text is not legal advice. Have your specific implementation reviewed by a lawyer.
The pricing page with the most expensive plan on the left
You know this picture: a SaaS pricing page shows three plans. On the far left is “Enterprise, €499 per month”, next to it “Pro, €99” and “Basic, €29”. Reading left to right, you see €499 first. After that, €99 looks reasonable, almost cheap.
Enterprise
€499per month
- unlimited users
- SSO and audit log
- response within 4 hours
ProPopular
€99per month
- up to 25 users
- API access
- response next business day
Basic
€29per month
- up to 3 users
- no API access
- email support
A wheel of fortune that shifts estimates
The anchoring effect is named for the way a number we saw moments earlier ties our estimate to itself — even when it has nothing to do with the question.
Tversky and Kahneman demonstrated this in 1974 with a wheel of fortune. It landed on a number between 0 and 100 in the participants’ presence. They were then asked to estimate the share of African countries in the United Nations — first whether it was above or below the wheel’s number, then the value itself.
| Number on the wheel | Estimated share (median) |
|---|---|
| 10 | 25% |
| 65 | 45% |
Same knowledge, same question, 20 percentage points apart. A payoff for estimating accurately changed nothing.
Two details decide how much of this carries over to a pricing page.
First, participants had to use the number before estimating: they had to say whether the quantity sought lay above or below it. A number that merely sits nearby and that nobody attends to is a different thing.
Second, the 1974 article is a review piece, not a single study. For the wheel-of-fortune experiment it reports only the two medians — no sample size, no measure of dispersion, no significance test. So the famous original passage does not carry the effect. What carries it came later.
In pricing nobody spins a wheel anyway. The number is placed: as a strikethrough price, a list price, the most expensive plan, the first offer in a negotiation.
How well is the effect supported?
As broadly as almost any of these patterns — and still with three catches that the usual telling leaves out.
The foundation is one of the largest replication studies in psychology. In the Many Labs project, 36 laboratories repeated the same 13 experiments with 6,344 participants (Klein et al., 2014). Ten of the 13 effects appeared again. Anchoring appeared in all 36 samples, and markedly more strongly than in the original.
The three catches:
| Catch | What lies behind it |
|---|---|
| The setup was simplified | The two-step question was collapsed into one step. The authors’ own comparison test found that this version produces larger effects — so they qualified the strength themselves. |
| The spread was the widest | Of all 13 effects tested, anchoring varied most across samples. |
| None of the questions was a price | They asked about the distance from San Francisco to New York, the population of Chicago, the height of Mount Everest and the number of babies born per day. |
A number sitting nearby is not enough. Many Labs 2 (Klein et al., 2018) shows this on a case close to online retail: 6,826 participants estimated how well a smartphone would sell — called the P97 or the P17 depending on condition. The original had found d = 0.30; the replication found d = 0.04 (95 percent interval −0.01 to 0.09).
The d measures how far two groups sit apart, expressed in their own spread; 0.2 counts as a small effect. 0.04 is practically nothing, and the interval includes zero. So an incidental digit in a product name does not anchor. A price placed expressly as a comparison is a different matter.
How broadly the effect shows up is documented by a literature review. Furnham and Boo (2011) gather the areas in which it has been found: general knowledge, probability estimates, legal judgments, valuations, forecasting, negotiation.
How much such a review can carry depends on how it is built. This one is narrative: the authors organise and report, but they do not combine the studies into a single effect size, do not state a search procedure and do not appraise the quality of the individual studies. A meta-analysis would do all three. So it tells you that the effect turns up in many places — not how strongly.
Their own conclusion is accordingly more reserved than their introduction: on mood, personality and cognitive ability they explicitly call the findings contradictory and equivocal.
Two points from the review bear directly on pricing pages.
More extreme does not keep helping. Within plausible values the effect grows with distance; beyond them it does not. Wegener et al. (2001) even found an inverted-U pattern: very extreme anchors worked less well than moderate ones, because people start arguing against the value or ignore it altogether. For an enterprise tier that means an implausibly high price may anchor less than a high but credible one. Furnham and Boo also report contrary findings on this.
Experience offers little protection. Car dealers with all the necessary information, estate agents estimating prices and experienced legal professionals setting sentences were all pulled along by given numbers. The authors state that expertise does not appreciably reduce the bias — and that the experts considered themselves less susceptible than they were. Those primary studies are reported here at second hand.
For the effect itself the picture is clear. For the pricing page it is not: that the order of plans works the way a wheel of fortune works in the lab is plausible and unsupported. What argues against it is under limits and criticism — and it is not a short list.
Where anchors arise in pricing
- Strikethrough or list price: The crossed-out price serves as the yardstick.
- Order in pricing tables: The first price seen can set the yardstick. The assumption is plausible but not directly supported for pricing pages.
- Manufacturer’s suggested price: A stated guide price pulls price expectations upward.
- First offer in negotiations: In B2B sales, the first number named often has a strong influence on the outcome.
- Value anchor: The total value of a service is named before the price.
Practical cases
Test the order of plans
HypothesisTest whether the arrangement “most expensive plan first” changes the choice of the middle plan. Measure not only clicks but revenue per visitor and the cancellation rate after 90 days. An anchor that lures customers into an unsuitable plan merely moves the problem into retention.
Strikethrough price with a genuine reference
PlausibleAnyone advertising a discount needs a verifiable comparison price. Document your price history. In Germany, fixed rules apply to this when selling to consumers.
Value first, then price, in B2B offers
PlausibleOpen an offer with the expected benefit, for example “Your estimated effort today: €60,000 per year”. The price follows. The value anchor must be traceable and backed by assumptions, otherwise it looks like an empty number.
Legal framework in Germany
Since 28 May 2022, section 11 of the price indication ordinance (PAngV) requires that for price reductions on goods sold to consumers, the lowest total price of the last 30 days is stated and used as the basis for the reduction. A comparison with a manufacturer’s suggested price is a separate case. Whether and how the rule applies to digital subscriptions, services or pure B2B offers should be clarified by a lawyer.