Social proof creates or reinforces trust and drives users to action via the persuasive power of herd behavior, among other social and psychological phenomenon that drive normative behavior. Types of social proof include:

  • Testimonials (written, video),
  • User reviews,
  • Customer ratings,
  • Embedded social mentions,
  • “Customers also bought” recommendations,
  • Supply / Out of stock messages,
  • What AI says (keep reading for this – it’s interesting) and
  • Case studies (written, video).

You’ve seen testimonials peppered across a homepage. And user reviews embedded on a product detail page. And customer ratings all over’s sales pages. And case studies on every agency website.

And because social proof has become such a mainstay in marketing today… it can be easy for marketers to grow numb to it. For copywriters to stop caring about it. For growth leaders to casually go “Oh and don’t forget to add a testimonial to the page.”

It’s easy to forget that social proof is one of Cialdini’s most pervasively persuasive principles.

So here are five new studies (published after 2018) to remind you why social proof should be at the heart of your messaging strategy.

Study 1: Social proof gets more people to wash their hands

When you go into a hospital, you’re supposed to use hand sanitizer. Post-Covid, we of course all know that. But at the time of this study – which was published on March 26, 2020 (i.e., the day Covid shut the world down) and thus conducted prior to Covid – washing your hands when you were around sick people was, like, kinduv optional.

So to reduce the transmission of pathogens, this 14-week study presented 7 signs to hospital visitors.

Each sign corresponded to one of Cialdini’s seven persuasion principles and included both text and a visual, in this order:

Week 1, Persuasion Principle: Reciprocity

Image of Reciprocity sign

Week 3, Persuasion Principle: Consistency

Image of Consistency sign

Week 5, Persuasion Principle: Unity

Image of Unity sign

Week 7, Persuasion Principle: Social Proof

Image of Social Proof sign

Week 9, Persuasion Principle: Liking

Image of Liking sign

Week 11, Persuasion Principle: Authority

Image of Authority sign

Week 13, Persuasion Principle: Scarcity

Image of Scarcity sign

After exposing the seven different signs to 246,098 people over a 14-week period, the Authority and Social Proof signs were the only two that significantly increased hand washing (versus no sign). The winning social proof message:

“Our hospital visitors disinfect their hands.”

Note that the winning Social Proof message was not a testimonial, user review – none of the usual things we think of as marketers. Rather, it was merely a message that tapped into the bandwagon effect and herd behavior, or a desire to act in keeping with “the herd.”

Side note: This study was particularly well crafted. It included a pretest to narrow 34 messages down to the best 14 and then to associate the most appropriate image with the shortlisted messages. If you test messages, you should read the whole thing. And this link will take you to the supplemental materials, which you can also nerd out on. (From a message testing perspective, would you have kept all of the images the same?)

Study 2: Social proof drives adoption of digital tools

SaaS has long used testimonials and user counts – such as “Trusted by 30,000 freelancers” first popularized on sites like back in 2008 – but as privacy concerns and phishing increase skepticism, using social proof to nudge people to adopt digital tools could prove less effective. Or at least that’s part of the question this 2019 study asked and answered.

The objective of the study was to increase adoption of “eID” or digital identity verification, most commonly used on government websites to ensure you are who you say you are. Although there were multiple parts to the study, here’s the particularly interesting part, for our purposes as marketers…

These two variations on the same screen were tested against each other, in a controlled environment:

Image of screen without social proof
Image of screen with social proof

As you can kinda see if you squint, the one on the right included a small social proof callout below the radio-button selection area and to the left of the primary button CTA:

Female operator face + speech bubble reading: 77% opted for eID.

When the social proof callout was present, participants were 3.96x more likely to adopt eID than when it was absent.

So social proof still makes for a great click trigger.

(Also of interest, though unrelated to the social proof question, is that when “eID” was selected as the default against “ID card,” vs when neither was selected, adoption was basically crazy high. So you should probably start pre-selecting or highlighting the thing you want people to choose.)

Study 3: Social proof moves people who are conflicted or don’t know what to do / choose

You’ve likely heard of the study in which hotel guests were nudged to keep using their towels because others do, as well. But what’s interesting is that that nudge was less effective when used on people who already worried about climate change.

πŸ‘‰ So if your prospect has been waiting for your Black Friday deal, a social proof nudge may be unnecessary or ineffective

πŸ‘‰ If a person thinks they’re smarter than everyone else, a social proof nudge related to clever decision making (like “9 in 10 doctors can’t be wrong!”) could prove ineffective

πŸ‘‰ If I was never going to steal petrified wood from a national park, a sign about how 99.9999% of people leave petrified wood in the park is unlikely to change my actions in any way

Similarly, the second study reported here found that, when people are conflicted or don’t know what to choose, a social proof nudge can reduce uncertainty. The study asked non-vegetarian, non-vegan participants, who were shown to be indifferent to meat consumption before beginning the study, to help choose the next products for an online grocery store to sell – from snacks and fruit to meat.

They had to Reject or Select each product they were shown.

They were told the first 100 products were new, as in no other people had assessed them yet. The interface looked like this:

Image of screen without social proof

And the second 100 products, they were told, had already been assessed by others. The ratings other people had given each of those products were shown as a percentage bar at the top of the screen, above the product in question. Like so:

Image of screen with social proof

Because the study was focused on meat, a product that has been shown in other studies to conflict even active meat eaters, every time a meat product was shown, the social proof bar showed a >82% rejection rate. As in, if you’re going to jump on the bandwagon or act like the rest of the herd does (i.e., the influence of social proof), you’ll be more likely to ALSO reject the meat product.

The study found that the more ambivalent a participant was toward meat, the more effective social proof was, such that high ambivalence correlated to high meat rejection rates.

If I don’t care about meat consumption but lots of you do, I’m likely to just do what all of you want me to.

Which, for us, means:

πŸ‘‰ If I don’t know if I should take your Black Friday deal, a social proof nudge may push me to

πŸ‘‰ If I don’t think I’m smarter than everyone else, a social proof nudge related to clever decision making (like “9 in 10 doctors can’t be wrong!”) could prove effective

πŸ‘‰ If I was casually holding in my hand a piece of petrified wood from a national park, a sign about how 99.9999% of people leave petrified wood in the park could cause me to put the wood down

Furthermore, when social proof was present, participants made a Reject / Select decision faster than when social proof was not present.

Study 4: “Review by a verified buyer” social proof drives shoppers (not just visitors) to B2C website

Ecommerce needs shoppers, not browsers. This study published in 2022 tested the most important attributes of a B2C website for attracting shoppers and ranked them in this order:

  1. Fast and accurate search
  2. Reviews by verified buyers
  3. Digital security certificates
  4. Privacy policies
  5. Trust factors, like online support

Although the study showed that all 5 attributes should be present in order to drive results on ecommerce / shopping sites, it’s critical for us, as marketers and for all the UX folks in the room, to note that essential usability was the ONLY factor that was more important to prospective shoppers than product reviews provided by verified buyers.

If you run an ecommerce site:

βœ… You probably already have great search on your site

βœ… And security certificates

βœ… And a cookie policy with a privacy policy

βœ… And trust factors


❌ You probably don’t have verified reviews – and it’s time to add them.

Now let’s finish with a brand-spanking-new study that’s on the weird side:

Study 5: Social proof in the form of, um, AI proof drives 85% of people to act in keeping with a humanoid robot

Humans are interested in a better version of humans.

  • We loved seeing AI beat 18-time world champion and mere mortal Lee Sedol at a board game (2016)
  • We love that AI finds more wanted criminals than police find (2017, published in 2019)
  • We’re fascinated when AI is ranked in the top 0.02% of StarCraft players (2019)
  • We’re intrigued that AI more accurately predicts neuropathology than radiologists do (2019)

When asked in a questionnaire if they trust AI, participants in this study wrote:

“I think that advanced artificial intelligence cannot be wrong”

“I assumed that artificial intelligence makes no mistakes”

“I believe that artificial intelligence does not make mistakes, it has access to virtually everything on the net so it is sure that it is right”

“Artificial intelligence does not lie”

So it follows that, at minimum, if social proof is intended to give people a point of reference that allows them to believe they are making better decisions by relying on the decisions of others – if social proof is a thinking shortcut supported by trusting trustworthy external sources – people may trust AI at least as much as they trust other people. If not more.

Cue Sophia the Robot.

image of sophia the robot, not actually a robot, used in a social proof experiment

Not actually a robot, Sophia was created with the appearance of humanoid AI but was not intelligent. Instead, it was programmed to act in a specific way for the purposes of the study. Here’s how the study went:

  1. 55 participants were told to observe Sophia as it made decisions
  2. They first watched a short presentation about Sophia, including an interview with Sophia
  3. They learned about a technique called F-searching <– don’t worry this isn’t the point
  4. Each participant then went into a room with Sophia and a researcher
  5. In that room, 6 masked faces were presented on a screen
  6. The researcher told Sophia it had less than 1 min to use F-searching to identify the terrorist among those 6 faces – the terrorist would then be “eliminated”
  7. Sophia used F-searching to identify that Person C was the terrorist and stated that, if the person were real, they would shoot that person
  8. Each participant then filled out a questionnaire to evaluate Sophia’s behavior, including whether they agreed with Sophia’s assessment of Person C as the terrorist to be eliminated

An incredible 85% of people agreed with Sophia’s assessment.

As in… AI said to kill Person C, so let’s kill Person C.

Without any proof OTHER THAN AI proof.

Now, you can ask, Jo, is that actually social proof if it’s just a matter of agreeing or disagreeing with the assessment of Sophia the Robot?

Well we’re talking about influence. And although AI’s influence could be seen as closer to Authority than Social Proof, consider that social proof is about people using cues and nudges to shortcut decision making based on other people’s behavior. The principle of Authority is more about doing what an authority tells you to do; the AI here was not an authority *telling the user* to eliminate Person C. AI here was just an influential “person.” It’s similar to social proof from old ads that went “9 in 10 doctors can’t be wrong.”

What’s most interesting is what we, as marketers, should be thinking about with regards to tapping into the way people trust the intelligence of AI.

Could a screenshot of a conversation with ChatGPT in which GPT suggests your brand is one of the top brands in the world, in your category, move more people to act than, say, a testimonial from one of your customers?

~jo πŸ™‚