Left digit changes
With cent ending
€2.99
left digit: 2
Round
€3.00
left digit: 3
Left digit stays the same
With cent ending
€3.59
left digit: 3
Round
€3.60
left digit: 3
Evidence
Two conditions have to coincide. Thomas and Morwitz (2005) showed across five experiments that the left digit has to change and the comparison price has to be close. When it was two dollars away instead of one, the effect vanished despite the digit change. The same patterns appeared with quality ratings and with numbers carrying no product context at all.
Why it happens is open — the two explanations contradict each other. Bizer and Schindler (2005) showed that people drop the right-hand digits, but only among participants with no motivation to calculate carefully. Anderson and Simester (2003) rejected exactly that explanation against sales data from three field experiments and consider a price signal more plausible: the 9 tells customers the price is low.
The overall picture finds small effects. A preregistered meta-analysis of 69 studies (Troll et al., 2024) arrives at g = 0.13 for purchase decisions and g = 0.28 for the perceived price image. The g measures the size of an effect: 0.2 counts as small. That a change in the left digit is what matters is supported for the price image; for purchase decisions the same test did not hold up.
Limits
The effects are small and highly heterogeneous. After correcting for publication bias, nothing of the purchase effect remains. It appeared only when both price formats were visible side by side in the first place. For premium products it may fail to appear or reverse. A .50 ending did not work in those same field experiments — the ending has to touch the digit in front of it.