The social proof that works is specific, recent and from a person the reader can picture, and in the UK it must also be genuine, because fake or misleading reviews are now banned.
Most lists of social proof examples show what big brands put on their websites: customer counters, logo walls, star ratings. They rarely test whether any of it works, and they say nothing about UK law.
This guide sets out what the research does show, which formats suit a small business, how to collect proof without breaking the rules, and where to use it. It is part of our wider guide to social media marketing in the UK.
What the research says about reviews
The most cited study comes from the Spiegel Research Center at Northwestern University. It reports that a product with five reviews has a purchase likelihood 270% greater than one with no reviews, and that the benefit of additional reviews diminishes after the first five.
The same page says purchase likelihood typically peaks at ratings between 4.0 and 4.7 and then declines as ratings approach 5.0. It also reports that a verified buyer badge improves the odds of purchase by 15%. Spiegel's summary page has the detail.
Read it with care. The findings were first published in 2017, the page does not state a sample size or method, and the work was based on product sales, not local services. It tells you that a few reviews matter far more than none.
How recent the proof needs to be
BrightLocal's 2026 Local Consumer Review Survey asked 1,002 US adults about reviews of local businesses. It found 97% read online reviews, and 74% care only about reviews from the last three months. A third, 32%, look for reviews from the last two weeks.
xychart-beta
title "How recent reviews need to be (BrightLocal 2026, 1,002 US adults)"
x-axis ["Care only about last 3 months", "Look for last 2 weeks", "Swayed only by last week"]
y-axis "Share of consumers (%)" 0 --> 100
bar [74, 32, 18]
The sample is American, and BrightLocal sells local SEO software, so treat the numbers as a guide. The pattern is the useful part: a steady flow of recent proof beats a large pile of old reviews.
Which social proof formats to use
No study we found ranks social proof formats by how often customers read them. So this table is our judgement, based on what each format shows and what it risks.
| Format | What it shows | Main risk |
|---|---|---|
| Written review with full name, role or town | A real person with a specific experience | Needs permission; can look staged if all reviews sound alike |
| Screenshot of a customer message | An unscripted reaction | Needs permission and the customer's details hidden |
| Case study document | The problem, the work and the result | Needs the client's agreement and real numbers |
| Short video from a customer | Voice and face, hard to fake | Takes effort to collect |
| Count of customers or jobs | Scale | Must be true, current and defined |
| Third-party rating widget | Independent credibility | Shows your average, including bad scores |
| Customer's own post tagging you | Unprompted endorsement | You cannot control the wording |
Start with the first two. They are the cheapest to collect and the easiest for a reader to believe.
What makes a review believable
A believable review has detail. "Great service, highly recommend" is what a fake looks like. "They found the leak behind the bath panel in ten minutes and tidied up" is what a real customer writes.
Include the reviewer's first name and town or role where they agree. Anonymous reviews are easier to doubt.
Show some imperfect ones. A set of only five-star reviews looks edited, and Spiegel's finding that purchase likelihood peaks below a perfect 5.0 fits that instinct.
What the UK law says
The Digital Markets, Competition and Consumers Act 2024 bans fake and misleading consumer reviews. According to Lewis Silkin's guide to the law, the ban came into force in April 2025 and carries fines of up to 10% of a trader's annual global turnover.
The Competition and Markets Authority has already acted. After a three-month adjustment period, it reviewed the websites of more than 100 businesses for published policies on fake reviews and found 54 that might be failing to comply. Some had no relevant policy, and others had unclear or incomplete ones. Lewis Silkin's summary of the sweep covers the details.
The same source says the CMA expects businesses to have a policy on incentivised reviews, such as banning them or requiring them to be labelled and genuine. In practice that means no fake reviews from staff, friends or paid accounts, no payment for good reviews and no hiding the bad ones in a misleading way.
This is general information, not legal advice. Check the CMA's current guidance and take advice if you are unsure.
How to collect proof without breaking the rules
Ask every customer, not only the ones you expect to be happy. Ask soon after the work, while they remember it, and send a direct link to the place you want the review.
Do not pay for reviews or offer a discount in return for a good one. If you offer anything at all, label the review as incentivised and have a written policy.
Do not write reviews for customers or edit them to sound better. You may ask permission to quote a short part, and should quote it accurately.
Publish a short review policy on your website. The CMA sweep found that missing or unclear policies were the problem, and writing one takes an hour.
How to write the review request
Keep the request short and personal. Something like: "Thanks for choosing us last week. If you have a minute, a few lines on how it went would help other people in the area decide. Here is the link." Send it from a named person, not a no-reply address.
Ask for specifics with a light prompt, such as "what was the problem, and how did it turn out?". The answers come out more detailed than a bare request for a rating.
Send it once, and one reminder at most. More than that reads as pressure.
What to do with a bad review
Reply within a day, politely and without arguing the facts in public. Say sorry for the experience, offer a named contact and move the detail to a private channel.
Do not delete a review you are able to delete because you dislike it, and do not ask the customer to remove it in return for a refund unless that is a genuine resolution. Your policy should say how you handle complaints.
A calm reply to a poor review is itself social proof. Readers see how you behave when something goes wrong.
Using customer names, photos and messages
A customer's name, photo, job title and message are personal data. Ask for agreement before you publish any of them, tell them where it will appear, and keep a note of what they agreed to.
For screenshots of messages, hide phone numbers, email addresses and any other detail you do not need. For company clients, check whether the person speaking is allowed to speak for their employer.
If a customer withdraws agreement later, take the item down. Having a note of the original agreement makes that quick.
A simple case study layout
A case study does not need to be long. Six short parts do the job: who the client is, in a sentence; what the problem was; what you did; what changed, with a real figure; a line from the client in their words; and what happens next.
Use numbers only if the client agrees and you can back them up. "Enquiries up by roughly a third over three months, measured from their own form" is better than "dramatic growth".
If you do not have a figure, describe the change plainly. A real, modest result is more believable than a big claim with nothing behind it.
Where to use social proof on social media
Use it where it answers a doubt. On LinkedIn, a post quoting a client with their agreement, followed by the problem and what you did, works as a short case study. A document post can walk through the story in nine slides.
In local groups, let customers recommend you themselves, and answer politely when they do. Do not post as a customer. Our guide to Facebook groups and Nextdoor for local business explains the etiquette.
In paid ads, a short customer quote above the offer often reassures a cold audience. It needs the same permission and accuracy as any other use. Our paid media service can test proof formats in ads alongside the offer.
Industry notes
Ecommerce brands and retailers have the most reviews to draw on and the most to lose if they are not genuine. Keep verified-buyer reviews clearly marked, in line with the Spiegel finding on verified buyers.
Accountants, consultancies and other professional services firms often cannot name clients. Use anonymised case studies, or ask permission for a quote with a role and sector in place of a name.
Regulated professions have extra rules on testimonials and results claims. Check your regulator's current guidance before publishing any client quote.
How to measure whether it works
Test one change at a time. Add a customer quote to one landing page or ad and keep the other as it was, then compare enquiries after a few weeks.
A small business will not get statistical certainty, but you will see big differences. If a quote lifts enquiries clearly, keep it. If nothing changes, the proof was not what stood between the reader and the enquiry.
Why "followers up, sales flat" is the commonest agency failure explains how to tie any test back to enquiries.
Next steps
If you want help collecting reviews lawfully, building case studies or testing proof in your ads, talk to us.
