Anchoring
The first number a person sees pulls every number they estimate after it.
When people judge an unknown quantity, they start from whatever value is already in front of them and adjust away from it. The adjustment is usually too small, so the final judgement stays close to the starting point. The anchor works even when people know it is arbitrary and even when they are told to ignore it.
How it shows up in software
Every pricing page is an anchoring: The first number you see shapes every judgement that follows. exercise. The enterprise tier a small team will never buy still sets the frame they judge the mid tier against, and a struck-through original price sets the frame for the sale price. Anchors also run in non-money contexts: a storage meter that tops out at 2TB makes 50GB feel small, and a default tip of 20 percent moves the whole distribution of tips.
Using it well
- Order pricing tiers so the highest-value plan appears before the one most customers choose, and label plans by what they unlock rather than by size alone.
- Anchor on the cost of the problem, not the cost of your product. Show the hours a workflow burns today next to the subscription price.
- When asking for a numeric input (budget, team size, goal), pick a default that reflects real usage rather than a round number chosen to flatter your revenue.
- Check what anchor your empty state: What a screen shows before there is any content in it. set. A dashboard that shows a 0 next to an unexplained target teaches a user they are failing before they start.
Where it turns manipulative
- Fake reference prices. Printing a 'was' price the product never actually sold at is a false reference pricing dark pattern: An interface built to get an outcome the user would reject if it were stated plainly. and is regulated in the EU under the Omnibus Directive and in several US states.
- Anchoring on a competitor number you cannot source, or on a made-up 'industry average', is a lie wearing a chart.
- Pre-filling a donation or tip field with a high amount and making the lower options hard to find shifts money by making the alternative laborious, not by making it clearer.
Where you have seen it
Amazon
Deal listings show a crossed-out list price above the current price, so the discount is read against the higher number.
Slack
The pricing page presents Business+ and Enterprise alongside Pro, so the Pro per-seat price is read next to more expensive options.
Square point of sale
The tipping screen offers three preset percentages before the custom amount option.
What the research says
- Tversky and Kahneman, 1974Well evidenced
Participants spun a wheel that stopped on 10 or 65, then estimated the share of African nations in the UN. Median estimates were 25 after the low spin and 45 after the high spin, despite the spin being visibly random.
- Klein et al., Many Labs 1, 2014Well evidenced
Four classic anchoring problems were re-run across 36 samples and thousands of participants. Anchoring was among the most consistently reproduced effects in the whole replication set.
- Chapman and Johnson, 1999Mixed evidence
Anchors shifted judgement even for participants who were paid for accuracy and warned about the anchor, which is why the effect is treated as an attention effect rather than a lack of effort.
Classic anchoring with an explicit, relevant number is solid. Claims about subliminal or incidental anchors, where the number is never consciously processed, sit closer to the priming literature and have a much weaker replication record. Do not build a product argument on those.
Grades are a judgement about the evidence, not about how useful the idea is. Plenty of contested effects are still worth knowing, as long as you do not cite them as settled.