User generated marketing: how to collect customer content, what you may reuse, and where it belongs
The five kinds of customer content, the permission you need in writing, and the places it earns the most.
Published by AI Knows Us · Updated 13 October 2026
User generated marketing is marketing built from content your customers made: reviews, photos, videos, questions, before and after shots, forum answers. It works because it is evidence rather than a claim, and three things make it work in practice. You need a reason for customers to post, written permission to reuse what they post, and a place on your own site where it lives permanently. Miss the permission step and the campaign stalls at the legal review. Miss the third and you are building an asset on somebody else's platform.
The commercial case is simple. A buyer discounts what a brand says about itself and does not discount what another buyer says. That is why the same sentence converts better in a customer's words than in yours, and why a photo of your product in a real kitchen outperforms the studio shot.
The five kinds, and where each one belongs
| Kind | Where it comes from | Where it earns the most | What it proves |
|---|---|---|---|
| Reviews and ratings | Your own site, Google, the marketplaces, the industry directories | Product and service pages, and the comparison pages buyers read | That the thing works for people like the reader |
| Photos and video from customers | Social tags, review submissions, direct requests | Product galleries, ads, email, landing pages | Scale, fit, texture and real conditions |
| Questions and answers | Support tickets, live chat, comments, community threads | A question block on the product page, and your help centre | That you answer the awkward question honestly |
| Written stories and case notes | Interviews, post purchase surveys, long reviews | Sector pages and case studies | The outcome in the buyer's own language |
| Community content | Forums, subreddits, user groups, Discord | Wherever it already is, with your participation | That the product has users who care |
The last row matters more than its size suggests. Content on third party communities is where a great deal of buying research now happens, and the only thing to do about it is to be genuinely useful there under your own name.
How to get customers to make it
- Ask at the moment of satisfaction. That is a few days after delivery for a product, right after a result for a service, and immediately after a support problem is solved well. Not at the end of the month in a batch.
- Ask one specific question. "What were you worried about before you bought, and what happened?" produces usable content. "Leave us a review" produces four words and a star.
- Make it one tap. One link, no account creation, works on a phone, under two minutes.
- Ask for the photo separately. People who will not write a paragraph will send a picture.
- Feature the people who contribute. Credit by name and handle, and tell them when their content went live. That is the thing that produces the second contribution.
- Give a reason that is not a bribe. Entry to a draw open to all reviewers, early access, a feature slot. Never a reward conditional on the review being positive: the platforms prohibit it and consumer regulators act on it.
Permission, in writing, every time
The person who made the content owns the copyright in it. Tagging your brand does not transfer anything, and a platform's own terms let the platform use the content, not you. Settle this with a short written exchange at collection time.
- Ask for a licence in plain words and keep the reply. Name where you intend to use it: website, email, paid social, print, and for how long.
- Get a separate release for faces, including anyone else visible in the shot, and for children get it from the parent.
- Keep a record that joins the asset to the permission. A spreadsheet with the file name, the person, the date, the wording and a screenshot of their reply is enough, and it is what makes the asset usable two years later.
- Do not change the meaning. Trimming a review for length is normal. Cutting the reservation out of a mixed review is not, and it is exactly the practice regulators look for.
- Honour deletion requests quickly, and build the process now rather than when the first one arrives.
The search side, which most programmes miss
Customer content is also search content, and three specifics decide whether it helps you.
- Publish it on your own pages, rendered in the HTML rather than loaded from a widget that search engines never execute. If a review is only visible after a script runs, it does nothing for the page.
- Review structured data has rules. Google's review snippet guidance excludes reviews written or supplied by the business about itself, so mark up genuine customer reviews of a product or service and nothing else.
- Mark up links inside user content. Google's guidance on qualifying outbound links sets out the ugc value for links in comments and forum posts, and sponsored for paid ones. Applying it correctly protects the pages your community builds.
Reviews also feed the words you use everywhere else. Customers describe problems in language no internal team would write, and that language is what people actually search for. Read fifty reviews with a highlighter and you have a keyword list and a headline. For turning that into a plan, see keyword research.
Measuring it
Four numbers tell you whether the programme is working, and none of them is a follower count.
- Collection rate: content received divided by customers asked. This is the number you can improve fastest.
- Coverage: the share of your important products or service pages that have at least five pieces of customer content on them.
- Conversion difference: pages with customer content against comparable pages without it.
- Reuse: how many pieces made it into paid ads, email and sales decks. Content collected and never used is the commonest waste in this discipline.
This page covers the programme: collecting customer content, the permissions, and where to put it. For paid creator work, which is a different arrangement with different disclosure duties, see influencer marketing. For replying to reviews, see how to reply to Google reviews.
What gets asked about user generated marketing
Is user generated content the same as influencer content?
No. Influencer content is commissioned and paid, and it must be disclosed as such. User generated content comes from a customer who bought the thing. Paid creator content made to look like a spontaneous customer post is the one combination to stay away from.
Can I repost a customer's Instagram photo?
Ask first and keep the reply. A comment saying yes is a reasonable record for organic reuse, and for paid advertising get something more explicit that names the use and the term.
What do I do about a negative review?
Answer it publicly, briefly, without arguing, and fix the thing. A product page showing a four point something average with a few critical reviews answered well converts better than a wall of perfect scores, because the reader believes it.
How much content do I need before it is worth it?
Five pieces on each of your most important pages changes how those pages read. Start with your top ten pages by revenue rather than spreading thin across the catalogue.
Should I run a hashtag campaign?
Only with a collection and permission process behind it. A hashtag with no route from post to permission to page produces activity and no asset.
Where should reviews physically live?
On your own domain first, with the text in the page source, and syndicated to the platforms your buyers check second. Owning the primary copy is what keeps the asset when a platform changes its terms.
Where the customer's sentence ends up
Each piece of content you collect, a review or a testimonial or a photo caption, can be lifted into an answer. A buyer asks an assistant "is this brand reliable?", and the reply quotes something a customer wrote, perhaps from your site, perhaps from a forum.
Look at what is being quoted. Ask the question for your brand and read the reply as a prospect would. AI Knows Us records those replies across assistants and notes when the wording changes, which helps you decide what to publish next.