Amazon 1-Click: A Patent on Removing a Step, and a Spending Effect That Keeps Shrinking

Most interface design adds something: a prompt, a banner, a badge. The most valuable interface patent of the internet’s first decade was for taking something away.

What was actually patented

US Patent 5,960,411, Method and system for placing a purchase order via a communications network, was filed on 12 September 1997 and granted on 28 September 1999. The named inventors are Peri Hartman, Jeffrey P. Bezos, Shel Kaphan and Joel Spiegel; the assignee is Amazon.com, Inc.

The operative language of the first independent claim is worth reading, because it shows exactly what the property right covered:

“A method of placing an order for an item comprising: under control of a client system, displaying information identifying the item; and in response to only a single action being performed, sending a request to order the item along with an identifier of a purchaser…”

The patent’s status is now Expired - Lifetime, with an anticipated expiration of 17 September 2017.

The invention, in substance, is the absence of a shopping basket. Every element already existed: stored addresses, stored card details, a server that could match one to the other. What was claimed was the removal of the confirmation steps between wanting the thing and having ordered it.

The courts were not persuaded it was novel

Amazon sued Barnesandnoble.com over its “Express Lane” feature and won a preliminary injunction in the district court, which found likely infringement and thought the validity challenges weak.

The Federal Circuit disagreed. In Amazon.com, Inc. v. Barnesandnoble.com, Inc., 239 F.3d 1343 (Fed. Cir. 2001), decided 14 February 2001, the court vacated the injunction:

“BN has mounted a substantial challenge to the validity of the patent in suit. Because Amazon is not entitled to preliminary injunctive relief under these circumstances, we vacate the order of the district court.”

The standard the court applied is worth noting, because it explains what was and was not decided: “If BN raises a substantial question concerning either infringement or validity… the preliminary injunction should not issue.” Amazon had shown likely infringement. What defeated the injunction was that Barnes & Noble had raised a serious question about whether the patent should ever have issued.

The prior art the court examined included the CompuServe Trend System, the Web-Basket ordering system, Magdalena Yesil’s 1996 book Creating the Virtual Store, the Oliver’s Market ordering system, and US Patent 5,708,780.

This was a decision about a preliminary injunction, not a final ruling of invalidity, and the case settled the following year. But the patent that is remembered as an unassailable moat was, on appellate review, substantially challenged on its first serious test.

Did removing the friction change what people spend?

This is where the story is usually told badly. The received version is that one-click ordering unleashed impulsive spending because the pain of paying vanished. The measured version is smaller and moving in the wrong direction for that account.

Schomburgk, Belli and Hoffmann (2024), in the Journal of Retailing (100, 382-403), meta-analysed the “cashless effect” - the claim that less tangible payment methods increase spending.

Their corpus: 392 effect sizes from 71 papers, 11,257 unique participants and 338,513 transactions across 17 countries, spanning studies from 1978 to 2022.

The pooled result: g = 0.135, 95% CI [0.068, 0.203], which they describe as “small, but significant.”

Their bias and heterogeneity checks: funnel plot, Egger’s test and trim-and-fill together suggested publication bias was “unlikely”; heterogeneity was high at I² = 86.20%.

The moderators are where it becomes interesting:

Moderator	Finding
Conspicuous consumption	Stronger effect, g = 0.551
Pro-social consumption	Negligible, g = 0.023
Hedonic vs utilitarian goods	No significant difference
Decoupling or transparency alone	No main effect; only their interaction mattered
Time	Significant weakening, b = -0.006, p = 0.025

The authors’ conclusion on that last row: “the cashless effect has generally weakened over time,” which they attribute to consumers becoming familiar with cashless payment and adapting to it. Their advice to retailers is unusually candid - capitalise on it “before it diminishes.”

So the effect of making payment easier is real, is about a seventh of a standard deviation, is concentrated in visible, status-linked purchases, and has been shrinking as the thing became normal.

That is a considerably weaker claim than the one one-click ordering is usually credited with. And it is worth being precise about what this evidence is and is not: nobody has run a randomised test of one-click checkout specifically. The cashless-effect literature is about payment tangibility, which is adjacent to removing a confirmation step, not identical to it. The connection made in this article is an argument, not a measurement.

What the design actually did

If the spending effect is modest, what was the patent worth?

Two things that are not the same as making people spend more per purchase. It removed occasions to reconsider - the basket page, the review page, the confirmation - each of which is a moment where an order can be abandoned. And it made Amazon’s stored credentials into a reason to buy from Amazon rather than from a site where you would have to type a card number again. The patent’s value was arguably less in psychology than in switching costs.

Which is why it fits into a pattern this site has documented from the other direction. Friction is not applied evenly. It is removed on the way in and reinstated on the way out - the design known as the roach motel, where signing up takes a single action and cancelling takes a phone call. One-click ordering and a four-screen cancellation flow are the same design philosophy pointed in opposite directions, and only one of them has ever been patented.

Scarcity and urgency cues are a separate mechanism with its own regulatory record, and consent banners are a third. The common thread is that in each case the interface decides how hard the next step is.

Limitations and disputes

• No study tested one-click checkout itself. The behavioural evidence here concerns payment tangibility in general. The link to single-action ordering is inference.
• The 2001 decision was preliminary. It vacated an injunction on the basis of a substantial validity question; it did not rule the patent invalid, and the case did not proceed to a final judgment on validity.
• We have not reported the outcome of the later USPTO reexamination of the patent, because we could not verify its record directly.
• We have not reported what Apple paid to licence the patent, or on what terms, for the same reason.
• Heterogeneity in the cashless-effect meta-analysis is high (I² = 86%), which means the pooled figure conceals wide variation between settings.
• The meta-analysis disclosed no funding source in the copy we read.
• This is a description of a design and its evidence, not advice about payment methods or spending.