Price anchoring is when the first price a buyer sees shapes how they judge every price after it.
A $199 plan makes a $79 plan look reasonable, and a crossed-out $120 makes $80 look like a deal, even though nothing about the product changed.
It works because people judge value by comparison, not in isolation, so whoever sets the first number sets the frame.
Below is what the research actually shows, what real A/B tests in our database show, and where anchoring fails or backfires.
You can browse every result on the test database, or start from the broader psychology hub.
How anchoring pricing works on real pages
On a store, SaaS pricing page or app paywall, an anchor is any number that shows up before or next to the price you want people to pay.
The three common forms are a reference price (compare-at or "was" price), a higher-priced option placed first or alongside, and a reframed unit (per month, per day, per ounce).
1. Reference prices: the crossed-out number
The simplest anchor is the original price struck through next to the sale price.
DRIP Agency tested this for Oceansapart: the control showed discounts in a way that needed mental math or a code, and the variant showed a crossed-out original price next to the final price on listing, product and cart pages.
That change added €43,876 per month in revenue, though the write-up does not state sample size or significance.
The anchor has to be honest. In the US, the FTC's Guides Against Deceptive Pricing (16 CFR 233.1) say a former price must be one the product was openly offered at "for a reasonably substantial period of time."
In the EU, Directive (EU) 2019/2161 defines the prior price in a price-reduction claim as the lowest price the trader charged in at least the 30 days before the discount.
An inflated "was" price is a dark pattern, and in many markets it is also unlawful.
2. Expensive options shown first
Putting a higher price in view first makes the next price feel smaller.
Elevate_ab reversed the bundle order on a product page so the most expensive bundle appeared first instead of the cheapest.
Revenue per visitor rose 8.47% (EUR 3.54 to 3.84), and conversion rate went from 5.6% to 5.74%.
It is a vendor write-up with no sample size, so treat it as a promising pattern rather than proof.
Microsoft 365 ran a similar test on its app paywall, listing the pricier plan first as the default.
Ramit Arora of Microsoft reported that both plans sold more, and the cheaper personal plan sold much more.
No numbers were published, but the direction matches the theory: the expensive plan worked as a reference point that made the personal plan look cheap.
A close cousin is the decoy effect, where a third option exists mainly to make another one look better.
Anchoring is about the number; decoys are about the comparison set.
3. Anchoring one plan against another
Plans can anchor each other. Convertibles worked with an Australian gin subscription brand that charged $99 for both monthly and bi-monthly boxes.
The variant raised bi-monthly to $109, kept monthly at $99, and added a "Go monthly & save $10/box" callout.
Revenue rose A$4,091 per month: conversion stayed flat, but more people picked the monthly plan.
Preu AI, an exam prep app in Chile, went the other way.
Its paywall showed only a monthly plan; the variant added an annual plan priced at 60% of the monthly price.
Proceeds per user rose 80% and trial conversion rose 22%.
The monthly price became the anchor that made the annual plan look like a bargain.
See more paywall examples on our paywalls page.
4. Reframing the unit
Breaking a price into a smaller unit changes which number becomes the anchor.
Mojo, a video editing app, showed its annual plan as a per-month price billed annually instead of a yearly total.
New revenue per paywall impression rose 45% in Brazil, 26% in Mexico and 8% in the USA, driven by more trials and more annual picks.
The spread across countries is a reminder that the same anchor lands differently in different markets.
DRIP Agency
Elevate_ab
ConvertiblesWhat the research says about the anchoring effect
The anchoring effect is the tendency for an initial number to pull later estimates toward it, even when people know the number is irrelevant.
It was named by Amos Tversky and Daniel Kahneman in their 1974 Science paper, "Judgment under Uncertainty: Heuristics and Biases."
The classic studies
| Study | What they did | What they found |
|---|---|---|
| Tversky and Kahneman (1974) | Spun a wheel rigged to land on 10 or 65, then asked what percentage of African countries were in the United Nations | Median estimates were 25% after seeing 10 and 45% after seeing 65 |
| Northcraft and Neale (1987) | Gave students and real estate agents identical property information except for the listing price | Both groups' value estimates moved with the listing price, and the agents denied being influenced (paper, summary) |
| Ariely, Loewenstein and Prelec (2003) | Asked people to write down the last two digits of their social security number, then bid on products | People with above-median numbers stated values 57% to 107% higher (paper) |
The real estate study matters most for pricing: anchors affected trained professionals who had a full information pack and toured the house. Expertise does not switch the effect off.
Replication: strong in estimates, weaker in willingness to pay
Anchoring is one of the better-replicated findings in psychology.
The Many Labs project ran 13 classic effects with 6,344 volunteers in 12 countries, and anchoring replicated consistently and came out stronger than the original study suggested.
The money version is shakier. Maniadis, Tufano and List replicated one of the Ariely, Loewenstein and Prelec experiments and found much weaker anchoring effects on what people would accept for listening to unpleasant sounds.
Their point was that a single striking result is weak evidence until others repeat it.
So the fair summary is this: anchors reliably move estimates of unknown quantities.
Whether an arbitrary number moves what someone will actually pay is less settled.
A relevant anchor on a pricing page, such as a real higher plan, is closer to the strong case than a random number is.
Anchoring bias: when it backfires, and how to test it
Anchoring bias is the same effect seen as an error: relying too heavily on the first number and adjusting too little from it.
For marketers, the risk is assuming the bias guarantees a lift.
Our database has clear cases where it didn't.
When anchors fail
- An anchor that's too high scares people off. iProspect tested "starting at" prices with a monthly payment figure on high-ticket filtration product pages that had shown no price before. The low starting price lifted lead submissions 15% at 95.26% significance, but the high-price variant was switched off early after doing worse than control in some categories.
- Savings displays can add friction. Conversion Fanatics tested a bulk product whose quantity options showed growing savings. Hiding the savings lifted conversions 25.1% (a per-unit price version lifted them 12.9%). No sample size was given, but it shows that more anchoring math is not always better.
- A repeated anchor stops being news. GrowthRock added "You Save #%" next to was/now prices for an apparel retailer where nearly every product had the same 30% discount. With 280,000 visitors per variation, all versions finished within 0.3% of each other.
How to test anchoring on your own pages
- Pick one anchor type per test. Plan order, a reference price and a per-unit frame are separate hypotheses. Testing them together tells you nothing about which one worked.
- Measure revenue, not just conversion. Anchors often shift plan mix more than they shift purchase rate, as in the gin subscription test. Use revenue per visitor or proceeds per user. Our price testing guide covers metrics and sample sizes.
- Only use real reference prices. If the "was" price was never charged, the test result is not worth the legal risk or the trust you lose.
- Segment by market. Mojo's lift ranged from 8% to 45% by country. An anchor that works in one market can be flat in another.
- Pair it with clean number formatting. Anchors interact with how prices end. See charm pricing for the evidence on .99 endings.
Treat anchoring as a strong hypothesis, not a rule.
The research says the effect is real; the test results say its size depends on the product, the price gap and whether the anchor is believable.