The Psychology of Social Proof

At a midsized hotel in the American Southwest, a towel-reuse sign told guests that most other guests reused theirs. In the 2008 study by Noah Goldstein and Vladas Griskevicius, that message produced 44.1% reuse, compared with 35.1% for an environmental appeal.

The message gave guests information about a choice other people had made. To borrow that approach for copy, you need to decide which people's behavior could help your reader.

Picture a pricing page, invented for this lesson, that says "Join 212 happy teams." A procurement lead at a large company reads it and wonders whether anyone her size has bought. We'll follow her decision through the examples below, which are also invented.

The mechanism in plain terms

Nielsen Norman Group describes social proof as people using others' behavior to guide their own. The hotel sign supplied a possible norm: this is what guests here do.

The original hotel's second experiment illustrates how shared circumstances can define a group. One sign said 75% of guests who had stayed in this room had taken part.

That sign got 49.3% reuse, against 42.8% for the other crowd signs combined. Yet when the researchers surveyed a separate group of 53 people, being a guest in a particular room ranked well below citizenship, gender and caring about the environment as part of their identity.

Before requesting a demo, the procurement lead wants evidence that the tool can support her planned rollout: 30 people in finance and purchasing handling about 500 purchase approvals a month. A count of teams leaves that workload unclear.

Write the decision beside the evidence: "Can a company our size work with this?" That question gives you a way to choose among proof elements. Other persuasion principles and cognitive biases address different reasons someone might hesitate.

How the reader uses it

From doubt to a borrowed decision

1. I am unsure what to do

• Uncertainty makes other people's choices relevant.

2. I look for people in my situation

• The group has to fit the decision in front of the reader.

3. I check whether I believe the signal

• A specific, checkable signal has to answer the actual doubt.

4. I follow them, or walk away

• Following the crowd is possible, and never guaranteed.

Your copy can lose the reader at any of these steps.

Three questions for any line of proof

1. Which doubt must the evidence answer?

Name the doubt your evidence needs to answer. Your objection research gives you the list to match against.

Start with the procurement lead's size question. Even if the count of 212 is correct, "happy" doesn't tell her which teams resemble hers. She needs an account from a company like hers that she can examine.

Here is an account invented for this lesson:

"Our logistics firm has 240 employees. After six months in finance, we added purchasing; 30 people across the two departments now route about 600 purchase approvals a month through the tool."

The lead can compare active users and monthly approvals with her planned rollout. The account describes two departments; it leaves company-wide use untested.

2. Is the group relevant to this decision?

For the procurement lead, a customer managing approvals across departments offers a comparison she can use. Pick a reference group through what the buyer needs to do, then check whether readers recognize the connection.

3. Will a skeptic believe the signal?

Define whether "212 teams" counts current paying customers or everyone who ever opened an account. For a customer story, collect the work it describes and permission to publish the customer's words.

Prepare a proof request before collecting quotes

Unfocused request

Tell us how much you love the product.

Suggested interview prompt

What made you hesitate before trying it, and what happened when you did?

Keep the answer in the customer's own terms, including qualifications they consider relevant.

When it does nothing

In 2014, Bohner and Schlüter studied towel-reuse messages at two German hotels, recording 724 opportunities for reuse in the first study and 204 in the second. The counts describe opportunities for reuse. Reuse was high overall, and both messages beat having no message.

The crowd wording did no better than the plain environmental appeal, though. Mentioning guests from the same room didn't reliably help either.

Keep the earlier hotel's percentages out of the pricing-page forecast. The German findings leave open the possibility that a clear request does much of the work. Compare proof against an equally clear request so you can see what the extra information contributes.

Bohner and Schlüter, field studies in two German hotels

Finding

In the two German hotel studies, norm wording did not outperform the standard environmental appeal.

Caveat

The 2014 replication limits how confidently you can transfer the earlier hotel result to a new setting.

Source

PLOS One (2014)

When can social proof backfire?

NN/g reports a usability participant who saw a share count on an article and decided it wasn't popular enough to read. The site meant to signal popularity, and the reader took away the opposite.

The procurement lead's doubt takes a similar turn. The page offers "212 happy teams" as reassurance; she reads the count as a possible sign that the product serves smaller buyers. Ask a reader what the number suggests before making it more prominent.

A widget slot is optional; leave it unused when the evidence is irrelevant.

A count can raise a new doubt

A share count can invite a reader to judge popularity. A customer account can help them examine a particular experience. Choose the evidence after identifying the decision it needs to support.

A customer account lets the reader examine work a bare count leaves unexplained.

Don't answer the size question with a bigger-looking metric. A renewal figure shows customers staying and says nothing about their size, so request a real account like the invented one above and keep its limits in.

Handling proof you aren't proud of yet

Do

• Keep a dated record of what each count measures
• Use a relevant customer account with permission
• Remove a share widget if it raises an unhelpful popularity question
• Describe the group accurately

Avoid

• Invent a larger crowd
• Treat the largest number as the strongest evidence
• Suggest that an unwanted behavior is the expected choice
• Imply an old customer relationship is current

Writing the crowd, line by line

Draft the introduction to the customer account around the relevant work: "See how one customer routes approvals across departments." That line earns its place only if the account shows the workflow and gives enough context to judge its scale.

Beside the draft, record which customer approved it and which passages support the description. Replace "enterprise-ready" with the specific work the account establishes, so the procurement lead can compare it with her own.

Put that account where the decision happens. Testimonials and proof on landing pages covers placement, and proof on product pages covers review displays.

Place the account beside the demo request, with its user count and approval workload visible. The procurement lead can then compare the relevant work while deciding whether the tool deserves a closer look.

Where the line is

Customer accounts must describe real experiences. In the U.S., 16 CFR 465.2 covers reviews and testimonials that materially misrepresent whether someone exists, used the product or had the experience described.

Section 465.8 covers fake indicators of social media influence under specific conditions. Check it before presenting a bought follower or engagement count as popularity.

Under 16 CFR 255.2, testimonials can imply expected results, and that implication needs support. Exceptional results may require disclosure of generally expected results; "Results may vary" alone won't resolve it.

Keep the two-department qualification visible in this account. If the surrounding copy implies company-wide performance or broader results, seek legal review before publishing.

Before any proof goes live

• I can name the doubt this evidence addresses
• The reference group is relevant to that decision
• The count has a defined unit and measurement period
• Customer words retain their meaning and have permission for this use
• Any implication about expected results has adequate support
• The test measures the action the page asks readers to take

Match each proof element to a doubt

The psychology topic explores other reasons a reader might hesitate. For the 212-team page, keep the test focused on the procurement lead's size question.

Compare the current page with a version that adds the verified, relevant customer account beside the same "Request a demo" button. Keep the pricing and offer identical. Measure completed demo requests from companies in the size range you serve, with that range defined before the test.

In our invented example, the procurement lead reads the account's 30-user, 600-approval workload and requests a demo for her planned rollout. She asks to see her own approval steps in that session and leaves any wider rollout undecided.