Robert Cialdini's principles of persuasion are seven shortcuts people use when deciding whether to say yes: reciprocity, commitment and consistency, social proof, authority, liking, scarcity and unity.
The first six come from the 1984 book Influence, and unity was added later.
In real A/B tests, these principles are useful sources of ideas, but none of them lifts conversion by default: they win when the cue is honest and relevant, and they lose when they add clutter or feel fake.
This page sets out each principle, what the lab research actually found (including the studies that did not replicate cleanly), and what store and SaaS tests in our database show when teams put the ideas on a live page.
The 6 principles of persuasion
Cialdini, now Regents' Professor Emeritus at Arizona State University, published Influence: The Psychology of Persuasion in 1984 (Wikipedia).
The six principles below use the definitions from Cialdini's own company, Influence at Work.
| Principle | Cialdini's definition | Typical page version | What our tests show |
|---|---|---|---|
| Reciprocity | People give back what they received first | Free samples, free content, unexpected gifts | Free samples and well-matched freebies won; a gift lost to a lower price |
| Commitment and consistency | People act in line with what they already said or did | Quizzes, yes/no openers, small first steps | Quiz-led flows often won; the lab effect is small |
| Social proof | When unsure, people copy others | Reviews, ratings, bestseller badges, customer logos | Placement decides it: the same reviews won in one spot and lost in another |
| Authority | People follow credible experts | Expert endorsements, press logos, certifications | Mixed: one expert framing won big, press bars and an expert badge did not |
| Liking | People say yes to people they like | Founder stories, real faces, friendly tone | Hard to isolate; one founder story test cut CPA |
| Scarcity | People want what they can have less of | Low stock, limited runs, countdowns | Real constraints won; generic "selling fast" labels were flat |
A useful way to read the table: each principle is a reason a shopper might trust a claim or act sooner.
If the shopper already has that reason, adding the cue does little.
If the cue gets in the way of the product or price, it can cost you.
The 7 principles of persuasion: adding unity
Cialdini introduced unity in the 2016 book Pre-Suasion and added it to the 2021 Influence, New and Expanded edition, where it gets its own chapter (ASU W. P. Carey News).
The idea: the more we see someone as part of our "we," the more they can influence us.
That is different from liking. You can like a salesperson without feeling they are one of your own.
On a website, unity shows up as shared identity: "for runners, by runners," member communities, causes the brand and buyer both care about.
It is the hardest principle to test, because identity messaging is usually spread across a whole site.
One example from our database: AFTCO added conservation-pledge messaging across the homepage, product pages, checkout and navigation.
The test run by Clean Commit came out flat at +0.7% conversion rate (+5.0% revenue per visitor at 79.1% probability, over 92,383 visitors).
Shared values may build the brand over time, but this test did not show a short-term conversion effect.
Cialdini's Influence: what research shows vs. what A/B tests show
Most of the evidence in Influence comes from field experiments and lab studies from the 1960s to the 2000s.
Many hold up in direction, but the effect sizes are often smaller than the famous versions suggest, and several depend on context.
Keep three gaps in mind before you copy a study into a test plan.
The famous numbers came from small samples
The best-known foot-in-the-door result, a commitment and consistency study, comes from Freedman and Fraser in 1966. Homeowners who first agreed to a small request were far more likely to accept a large "Drive Carefully" lawn sign: 76.0% versus 16.7% in the control group, with only 25 and 24 people in those two groups (original paper, Table 2).
A 1984 meta-analysis by Dillard, Hunter and Burgoon found foot-in-the-door and door-in-the-face effects to be small (r = .17 and .15), even under the best conditions, and both needed a prosocial topic (Penn State abstract).
Buying a product is not a prosocial request.
Effects depend on the setting
The hotel towel study by Goldstein, Cialdini and Griskevicius is a standard social proof example.
In the US, a "most guests reuse their towels" card raised reuse from 35.1% to 44.1%.
A German replication by Bohner and Schlüter found guests already reused towels at 70-90%, and the norm message did not beat the standard environmental message (PLOS ONE, 2014).
A Bayesian synthesis by Scheibehenne, Jamil and Wagenmakers in 2016 concluded the combined evidence still favored a small effect.
The lesson: when the baseline behavior is already high, the cue has little room to work.
Some results hold up well
Authority has one of the sturdier records. Jerry Burger's 2009 partial replication of Milgram's obedience study found 70% of participants were willing to continue past 150 volts, compared with 83% in the matching Milgram condition, a difference that was not statistically significant (American Psychological Association).
But obeying a lab experimenter is not the same as trusting a press logo on a product page, which is why tests matter.
A/B tests answer a narrower question than the lab: does this cue, in this spot, for this audience, change what people buy?
The rest of this page uses that standard.
Social proof in tests
Reviews are the most-tested Cialdini cue in our database, and placement decides the result.
Dave Diederen showed customer reviews on the checkout page and got +17.96% conversion rate and +23.13% revenue per visitor.
Yet Dave Diederen also hid written reviews above the fold on product pages and saw +13.36% conversion rate and +23.05% revenue per visitor, reporting that hiding them won in 8 of 10 tests across brands.
Clean Commit saw the same pattern from the other side: moving three short testimonials next to the add-to-cart area produced -9.2% conversion rate (89.3% probability the control was better, not significant).
Social proof helps where doubt is highest and hurts where it crowds out the product.
More on this in our social proof and bandwagon effect page.
Authority in tests
Authority cues are less reliable than people expect.
Dave Diederen hid an "As Seen In" press logo banner on the product page and got +5.63% revenue per visitor and +2.72% conversion rate.
Ruben tested a doctor or expert endorsement badge: it showed +17% conversion rate after one week at 99.6% probability, then settled at +1.66% after four weeks, which Ruben read as noise.
Authority did win when it answered the visitor's real question.
At Kaya Skin Clinic, reworking the text above an appointment form to offer an "Expert opinion" took conversion from 4% to 9.5% over 45 days, in a test credited to Ultraviolet Digital.
Convertibles found press logos directly below a homepage hero worth +$109,829/month, while USP-bar versions and lower placements lost.
Scarcity in tests
Scarcity works when the constraint is real and specific.
Oliver Kenyon added authentic scarcity messaging to product pages for a brand built on limited production runs ("Once sold out, no guarantees it will be produced again") and reported +$46,900 per month.
Stone Creek Coffee's limited-time copy for seasonal coffees, with no discount, gained +16% conversion rate in a Blend Commerce test, though no significance figure was given.
Generic labels did much less. Clean Commit's "high demand" cart message landed at +2.7% conversion rate, not significant over 141,998 visitors, and a redesigned mobile "selling fast" badge came in at -0.9%.
The research behind scarcity, including Worchel, Lee and Adewole's 1975 cookie-jar study, found that items in short supply were rated as more desirable (original paper).
Fake countdowns and invented stock levels cross into dark patterns.
Our scarcity and urgency guide covers where the line sits.
Commitment and liking in tests
Commitment shows up online as small first steps.
Desert Does It used an email popup that opened with a yes/no question for 5% off and reported about twice the email sign-up rate, though the baseline was not given.
Liking is harder to isolate, but Oliver Kenyon added the founder story from the ads to the product page and reported -15% CPA and +$26k/month.
That test is as much about message match as about liking.
The reciprocity principle
Reciprocity is the rule that we feel obliged to repay what someone gave us first.
The classic study is Dennis Regan's 1971 experiment at Cornell: participants who received an unrequested soft drink from a confederate later bought about twice as many raffle tickets from that person.
Regan also found that how much participants liked the confederate made little difference (summary of Regan, 1971).
A field version comes from restaurants. Strohmetz, Rind, Fisher and Lynn found that one mint per diner with the bill raised tips by about 3%, two mints by about 14%, and one mint followed by a second, personal offer of an extra mint by about 23% (Journal of Applied Social Psychology, 2002).
The details matter: the gift worked better when it felt personal and unexpected.
Two features make a gift reciprocity in Cialdini's sense: it comes first, and it has no strings.
A "free gift with purchase" is a conditional incentive, not reciprocity.
It can still work, but for different reasons, and it competes directly with price.
Reciprocity marketing: what our tests show
Reciprocity marketing means giving something useful before asking for anything: samples, guides, tools, advice.
The closest match in our database is TimberTech.
Surefoot tested a small, dismissible "Starter Kit" toaster offering a free sample kit, which raised sample kit orders by 73% on mobile and 12% on desktop, statistically significant.
The sample itself is the first step toward a large decking purchase.
Gifts tied to a purchase are a different story.
Oliver Kenyon reported +$68,000/month from a free Summer BBQ recipe e-book added to product bundles, a cheap gift matched to what buyers wanted.
But in another store, Oliver Kenyon lowered the price and removed the free gift ($82 versus $85 plus a gift) and got +46% CVR, because shoppers cared more about price than about the bonus.
Choice does not make a gift stronger, either.
In a MarketingSherpa test with Act-On, letting visitors pick one of three free guides instead of getting one showed no statistical difference, and most people picked the default guide anyway.
@oliverkenyon@oliverkenyon@oliverkenyon
@DaveDiederen@DaveDiederen
SurefootHow to test reciprocity on your own site:
- Give before the ask. A sample, a sizing tool or a genuinely useful guide placed before checkout matches the research better than a gift unlocked by spending.
- Match the gift to the buyer. The BBQ e-book worked because it fit the product. A generic freebie is just clutter.
- Test the gift against a price cut. If shoppers are price-sensitive, the money may be worth more to them than the bonus.
- Run it long enough. Ruben's authority badge looked like a big winner after a week and flat after four. Plan your sample size before launch, and expect many of these ideas to lose.
Browse more gift tests on the free gifts element page, or search every result in the test database and the rest of our psychology guides.