How to Use Social Proof in Ads Without Breaking the Rules
Reviews, ratings and testimonials are some of the strongest proof you have. Here is how to use social proof in ads without fake claims, rejections or legal trouble.
Social proof is the reason a stranger believes your ad. When someone sees that other people like them bought the product and were happy, the risk of trying it feels smaller. That is why reviews, ratings and testimonials show up in so many long-running ads.
It is also one of the easiest places to get into trouble. Inflated review counts, cherry-picked ratings, testimonials that promise results most customers will not get, and "as seen on" logos that overstate a mention are all common. They can get ads rejected, accounts restricted, and in some cases draw regulator attention.
This guide covers the kinds of social proof that work, how to use each one honestly, and what to avoid on Meta and on ChatGPT ads.
Why social proof works
A cold buyer has no relationship with your brand. Your claims are, by default, the claims of a seller. Social proof shifts the source of the claim from you to other buyers, which feels more credible.
It works best when it is:
- Specific. "Finally a pillow that does not go flat after a month" beats "Great product!"
- Relevant. The person giving the proof resembles the reader or shares their problem.
- Verifiable. The reader could check it if they wanted to.
It works worst when it is vague, generic or obviously curated. Readers have seen a lot of fake-looking reviews and have become good at spotting them.
The main types of social proof
| Type | Example | Main rule |
|---|---|---|
| Star rating | 4.7 out of 5 | Must reflect all eligible reviews, not a filtered subset |
| Review count | 2,300 reviews | Must be real and current |
| Customer quote | "Stays cool all night." | Genuine, permissioned, typical of experience |
| UGC video | A customer using the product | Disclose paid relationships |
| Customer count | 50,000 customers served | Backed by records |
| Expert endorsement | Recommended by a named professional | Real, disclosed, relevant expertise |
| Press mention | Featured in a named publication | Accurate description of the coverage |
| Awards | Named award, year | Real award, not pay-to-play without disclosure |
Every row has the same underlying rule: it has to be true, and you have to be able to show it.
Rules that apply on every platform
Ad platforms and consumer protection rules converge on a few principles. Specific laws vary by country, so treat this as general guidance, not legal advice. Our guide to ad claims compliance and substantiation goes deeper.
Reviews must be genuine
No fake reviews, no reviews written by staff or family presented as customers, no reviews bought in exchange for a positive rating. In the US, the FTC finalized a rule in 2024 specifically targeting fake reviews and testimonials, which raised the stakes for this.
Ratings must not be cherry-picked
If you have 900 reviews averaging 4.2 stars, you cannot advertise the 4.9 average from your best-reviewed variant as if it applied to the whole product. If you exclude reviews (say, for spam), the rule for doing so should be consistent and defensible.
Testimonials should reflect typical results
A customer who lost a lot of weight or made a lot of money is not representative if most customers do not. Results-heavy testimonials in health, weight management and income categories are where most problems happen, and these categories are restricted or have extra rules on most ad platforms.
Paid relationships must be disclosed
If you paid a creator, sent free product, or offered a commission, that relationship needs to be disclosed in a way the viewer will notice. On Meta, branded content tools exist for this purpose.
Endorsements must be real and accurate
Do not imply a public figure, publication or organization endorses you if they do not. "As seen in" should only be used for real coverage, and should not suggest a recommendation if the coverage was neutral.
Social proof on Meta ads
Meta gives you room for rich social proof: UGC videos, review screenshots, quote cards, star graphics and testimonial carousels. A few practical notes:
- Quote cards work well in static ads. Use a real review, attribute it to a first name and initial with permission, and keep it short.
- UGC is social proof and demonstration at once. A real customer showing the product working is often stronger than any quote. See our guide to UGC ads on Meta for briefs and scripts.
- Comments on the ad are also social proof. Moderate spam, but do not delete genuine criticism as a habit. Answering it publicly can build more trust than hiding it.
- Avoid personal-attribute callouts. Meta's policies restrict ads that assert or imply personal attributes, such as health conditions. "Customers with eczema love it" can trip this; "Made for skin that reacts" generally does not.
Social proof on ChatGPT ads
ChatGPT ads are short. The chat card has a headline of 3 to 50 characters and a body of about 100 characters, plus an image. That leaves room for one compact piece of social proof at most.
OpenAI's ad policies explicitly prohibit fake reviews, invented statistics and deceptive or unverifiable claims, and ads are reviewed against those rules. They also prohibit impersonating public figures or other brands. So any social proof you use has to be real and supportable from your landing page.
What fits well in a chat card:
- A verifiable rating and count. "Rated 4.6 by 1,200 verified buyers" if your review platform shows exactly that.
- A plain customer count. "Shipped to over 20,000 homes" if you have the records.
- A specific, believable quote fragment. Hard to fit, but possible in the body.
A hypothetical example for a home goods brand:
- Headline (31 chars): A rug you can put in the washer
- Body (94 chars): Rated 4.6 stars by 1,200 verified buyers. Machine-washable, pet-friendly, free 30-day returns.
The proof is specific and checkable, and the landing page should show the same rating and count near the top. If the page shows a different number, you have a message match problem and a potential policy issue. The ChatGPT ad policy checker flags risky phrasing before you submit.
How to collect social proof you can actually use
If your proof is thin, fix the supply rather than stretching what you have.
- Ask at the right moment. Request a review after the customer has used the product, not the day it ships. For consumables, that might be two weeks; for mattresses, a month or more.
- Ask a specific question. "What problem did this solve for you?" gets more usable answers than "Leave a review."
- Get permission for ad use. Add a checkbox or a follow-up email asking whether you can quote them in ads.
- Collect video. Offer a small, disclosed thank-you for a short unedited video of the product in use.
- Keep records. Save screenshots and export review data so you can substantiate any number you advertise.
Proof when you have little or no social proof
New brands can still be credible. Concrete specifics and risk reversal carry a lot of weight:
- A guarantee or long return window
- Materials, ingredients, sourcing and process details
- Real certifications you hold
- A clear demonstration of the product working
- A founder story told plainly, without unverifiable claims
These are covered in more detail in the promise, proof, offer framework.
A social proof checklist before you launch
- Is every number current and traceable to a source you can export?
- Does the rating include all eligible reviews?
- Do you have permission to use each quote?
- Is any paid or gifted relationship disclosed?
- Would the testimonial's results be typical for most customers?
- Does the landing page show the same proof as the ad?
- Have you avoided naming competitors or implying endorsements you do not have?
How SecondWin handles proof in ChatGPT ads
SecondWin writes ChatGPT ads using only proof it can find on your site, such as the ratings, guarantees and details you already publish. Every ad is checked against OpenAI's ad policies and SecondWin's own rules, which block unproven claims, competitor names and made-up facts. If the proof is not on your page, it does not go in the ad.
Start with a free analysis of your site to see the proof SecondWin would use and the messages that keep running in your niche. You can compare plans on the pricing page.
FAQ
Can I use customer reviews in my ads?
Yes, if the reviews are genuine, you have permission to use them, and they are not misleading. Quote real customers accurately, do not edit their words in a way that changes meaning, and avoid testimonials that suggest results most customers will not get. Keep records of the original reviews so you can show where each quote came from if an ad platform or regulator asks.
Can I put my star rating in a ChatGPT ad?
You can, as long as it is accurate, current and verifiable, and your landing page shows the same rating. OpenAI's ad policies prohibit fake reviews and unverifiable claims, so the rating should reflect all eligible reviews rather than a filtered subset. Keep it short, since the chat card body is only about 100 characters.
Do I need to disclose that a creator was paid?
Yes. If you paid a creator, gave them free product, or they earn a commission, the relationship should be clearly disclosed where viewers will notice it. On Meta, use the branded content tools. Disclosure rules vary by country, but undisclosed paid endorsements are a common target of consumer protection enforcement and platform policy.
What social proof can a brand-new store use?
Use proof that does not depend on customer volume: a clear guarantee, specific materials or ingredients, real certifications, a product demonstration and an honest founder story. Start collecting reviews from your first customers right away, ask for permission to quote them, and add them to ads as they come in. Never fill the gap with invented reviews.