How-To Guide
What is social proof in marketing, and how does a brand build it on social?
Social proof is evidence that other people chose you. What it is, the six types, what the research says about reviews, and how to build it without faking it.
By the Superdirector editorial team. FTC, Medill Spiegel Research Center, Cialdini's firm and vendor pages read November 7, 2026.
At a glance
What is social proof in marketing, and how does a brand build it on social?
Social proof is evidence that other people chose, used or liked something, shown to someone who is still deciding. A brand builds it on social by collecting real proof, such as customer messages, reviews, order counts and creator posts, asking permission to share it, matching each piece to the doubt it answers, and posting it with its source and date. It has to be real: the FTC's rule, effective October 21, 2024, bans fake reviews and testimonials and the purchase of fake followers or views.
- The 270% figure vendors repeat is five reviews against none in Medill's Spiegel research, with diminishing returns after five.
- Put one piece of proof against one buyer doubt: is it real, is it for me, do others choose it.
- Every piece needs a source, a date and permission; no source, no proof.
Lift from reviews at two price points, from the Medill Spiegel Research Center study. Original diagram; source: Spiegel, first published 2017, measuring online reviews rather than social posts.
Make it work for your brand
Find it. Remake it. Post it. Superdirector turns the viral formats in your niche into your brand's carousels, X posts and video scripts, and keeps the calendar full.
Social proof in plain words
Social proof is evidence that other people chose, used or liked something, shown to someone who is still deciding. Robert Cialdini, who lists it among his principles of persuasion, says people look to the actions of others most when they are uncertain. A brand builds social proof on social by collecting real proof, asking permission to share it, showing it where buyers hesitate and keeping every piece checkable.
Cialdini's hotel towels show the idea. On his firm's page, a card asking guests to reuse towels for environmental reasons reached around 35% compliance. A card saying that 75% of guests reuse their towels at some point in their stay raised reuse by 26%; Cialdini says that 75% came from guests staying four nights or longer. Naming what guests in the same room had done raised it by 33%. That is Cialdini's own summary of the research, not the paper, so read the numbers as his account.
Buffer sorts social proof into six types. This table puts each on social, with our rating of the effort and the main thing to watch.
| Type | What it looks like on social | Effort to get (our rating) | Watch for |
|---|---|---|---|
| Expert | A specialist in your field uses or quotes your product | Medium: needs a real relationship | A quote you wrote for them is not theirs |
| Influencer | A creator's post about your product | Medium to high | A paid or gifted post needs a disclosure |
| Customer | Reviews, messages, photos, customer posts | Low to medium: you ask | Permission, and no incentive tied to a kind of review |
| Crowd | Counts of followers, customers or orders | Low | Numbers are hard to verify and easy to manipulate; IxDF's example is a page showing 10,000 likes, which only persuades when the audience engages |
| Friends | Someone the viewer knows follows or uses you | Not in your control | You can invite it, never stage it |
| Certification | Awards, badges, logos of known customers | Varies | Only real and current ones |
In our judgment customer proof is the type most brands can build first, since it needs only your own customers. The FTC rule in step 3 touches several of these types, so read it before you collect anything.
What the research says, and what it does not
Many social proof pages repeat the same few statistics. Most come from vendors, and the numbers lose their conditions on the way. This table sets the common ones against the sources.
| You will read | What the source says |
|---|---|
| Social proof lifts conversion by 270% (Flockler) | The figure comes from the Medill Spiegel Research Center: purchase likelihood for a product with five reviews was 270% greater than for one with none. The center first published the findings in 2017 and says the benefit of more reviews diminishes rapidly after the first five. |
| Five stars is the goal | Spiegel found purchase likelihood typically peaks at ratings of 4.0 to 4.7 and falls as ratings approach 5.0. |
| Hide the bad reviews | Spiegel says negative reviews can add credibility. The FTC's rule also bars groundless threats used to suppress a consumer review. |
| Reviews matter for everything equally | Spiegel reports conversion rose 190% for a lower-priced product and 380% for a higher-priced one when reviews were shown. |
| 92% of people trust recommendations (Nielsen) | Nielsen's survey, written up in April 2012, of more than 28,000 online respondents in 56 countries found 92% said they trust earned media above other advertising. It is old and measures stated trust. |
| Social proof always helps | A conversion practitioner quoted by CXL says that if there is little uncertainty, social proof will have negligible impact. Use it where buyers hesitate. |
Spiegel's data are about online reviews and purchase decisions. No source we read measured how much a quote carousel or an X post with a number moves buyers, so treat each format as a test and not as a promise.
List the proof you already have
Start with what is true and already yours. Look for customer messages and replies, reviews on sites you do not control, counts from your order list, posts by creators and customers who mentioned you, and awards you actually won. Sprout's list of nine ideas is a useful prompt for where to look.
For each item write four things in one row: who said or measured it, the date, whether you have permission to share it, and what record you can check it against, such as an order or the original message. The inventory sheet has the columns, with two fictional example rows. A piece with no source or no date is not proof yet.
If you have no customers yet, you have no customer proof to list, and inventing some is exactly what the rule in step 3 forbids. Start with the questions buyers ask and answer them on video without testimonials, then come back to this sheet after your first orders.
Match each piece to the doubt it answers
A buyer on social is usually stuck on one doubt. Put the proof where that doubt sits. A practitioner quoted by CXL says social proof matters in proportion to the uncertainty that stops a buyer, so a page for a cheap, familiar product needs less of it than a launch from an unknown brand.
| Buyer's doubt | Proof that answers it | Format on social |
|---|---|---|
| Is this real? | Named, dated customer messages and reviews with a source | Quote carousel |
| Is it for someone like me? | A post from a similar customer or creator, reposted with permission | Story carousel or a quote post on X |
| Are other people choosing it? | A count you can check against your records, with its date | One number in a short X post |
Choose one doubt per post. A quote that also tries to show a follower count and an award answers none of them well.
Each buyer doubt matched to one format that can show its source and date. Original diagram; drawn from this article's table of doubts, proof and formats.
Ask for more proof and keep it real
Ask customers for proof after they have received the product. Ask a question that invites a specific answer, such as what they were unsure about before ordering, and ask whether you may share the reply with their first name. Sprout advises reposting customer content with permission and tagging the original creator. Never write the customer's words for them.
Keep it real, because the law now says so. The FTC's rule on consumer reviews and testimonials took effect on October 21, 2024, and courts can impose civil penalties for knowing violations. Its announcement and the rule text cover these cases:
- Fake or false reviews and testimonials, including AI-generated ones, may not be created or sold, or bought when the business knew or should have known they were false.
- A business may not offer compensation or other incentives conditioned on a review expressing a particular sentiment, positive or negative.
- Reviews and testimonials written by a company's officers or managers need a clear disclosure of the connection, unless it is otherwise clear to the audience.
- Groundless legal threats or intimidation may not be used to prevent or remove a consumer review, negative or not.
- Selling or buying fake indicators of social media influence, such as followers or views generated by a bot or hijacked account, is prohibited when the buyer knew or should have known they were fake and they misrepresent the buyer's influence or importance for a commercial purpose. The rule's definition of indicators includes followers, views, likes, saves, shares and comments.
FTC staff add three practical points in their Q&A. A business should not supply the text for a testimonial without a reasonable basis to conclude it is truthful about the person's experience. A testimonial followed by a request for the product is a red flag. And paying influencers to recommend that their followers follow you is not buying fake indicators, because those followers are real people who chose to follow.
Paid and gifted relationships need a disclosure on the creator's post. The FTC says to disclose any financial, employment, personal or family relationship with a brand, and that an unpaid, unconnected recommendation needs no such statement. Our guide to checking a creator's disclosure shows how to review a live post. This guide is not legal advice.
The rule's line drawn as two lists, in force from 21 October 2024. Original diagram; source: 16 CFR Part 465 and FTC staff Q&A, not legal advice.
Post it in formats made for social
Match each piece of proof to a format that shows its source and date on the post itself. A buyer who cannot see where the proof came from has no reason to believe it.
| Proof | Format | What the post must show |
|---|---|---|
| A customer message | Quote carousel, one quote per post | The words as sent, first name or initial as agreed, the date, a link to more reviews |
| A customer or creator post | Repost or quote post, with permission | The original poster tagged, and a disclosure if the creator was paid or gifted |
| A count | One number in a short X post | What was counted, from which record, on which date |
| A rating | Quote card | The rating, the number of reviews and the site, with a date |
| An award or logo | Single image | The award's name, year and who gave it |
Here is a fictional five-slide quote carousel for an imaginary tea brand. It uses no real customer or result.
| Slide | Text on the slide (fictional) | Why it is built this way |
|---|---|---|
| 1 | "It came two days late and they fixed it the same day." | Opens with the customer's own sentence |
| 2 | The problem: a parcel delayed by the carrier | States the situation in one line |
| 3 | What we did: a replacement shipped on 28 October | A dated action that matches the order record |
| 4 | Shared with permission on 30 October, first name only | Makes the permission and date visible |
| 5 | More reviews at our review page | Gives the buyer a way to check |
And a fictional X post for the crowd type: "Since opening in March we have shipped 1,240 orders (our order export, 1 November). The question we get most: does it come in a smaller tin? From next week, yes." Use your own real count, your own record and your own date, or leave the number out.
The article's fictional tea-brand carousel, drawn slide by slide with the dates the post must show. Original diagram; worked example from this section, with no real customer or result.
Review and retire it every month
Once a month, go through the inventory sheet. Re-check each count against the record it came from and update the date, retire anything stale, and drop any item whose permission was withdrawn. A review quoted in March is not proof of how you ship in November.
Then compare the posts. For each proof post, note saves, replies and link clicks after seven days and set them beside your typical post, meaning the median of your last ten. Two proof posts in a row well above it say the format is worth repeating. Two well below it say the proof, or the doubt it answers, is wrong. These thresholds are ours, not a platform's, and a handful of posts is a small sample.
If a count is hard to show truthfully, show nothing. A smaller claim that checks out beats a larger one that does not.
Where Superdirector fits
Superdirector, which publishes this guide, does not collect reviews or check who your customers are. Its job is the posting side. The main plan is $199 a month for one social account that its AI runs in the direction you set. It tracks the accounts, competitors and keywords you give it, writes carousels and X posts in your brand voice, and eligible Instagram carousels and single link-free X text posts are published to your connected account under its standing posting permission. The $29 a month plan is for hands-on work from Claude or ChatGPT. AI video and full autopilot are coming soon. You stay the publisher, so any proof you hand it should already have its source, date and permission. See what runs today.