The ad
I have a habit. When an ad in my Instagram feed promises to fix someone's business, I click through, read the whole site, and check whether the promise holds up. This is the first one I'm writing about.
This week it was an agency promising to put brands at the top of Reddit, and into the answers ChatGPT and Perplexity give, since those tools lean heavily on Reddit threads. Slick site. Confident numbers. A founder with a real track record.
I'm not naming them. This isn't about one company. It's about a practice, and why it bothers me more than most ads do.
What they were actually selling
Strip away the trademarked buzzwords and the offer is simple:
- A network of over a thousand old Reddit accounts, aged so they look like ordinary people.
- Those accounts join threads where buyers are comparing products, and nudge the verdict toward the paying brand.
- Success is measured by how many planted comments survive without moderators removing them.
One case study bragged about turning a perfume into the Reddit answer for a very specific question, then ranking first on Google for it. Their pitch, more or less: your buyers ask Reddit, AI repeats Reddit, so we write what Reddit says.
Meanwhile, their About page promises no fake accounts. Both can't be true.
Why it works
Reddit runs on one assumption: the person replying to you has nothing to sell. That's why millions of us add "reddit" to the end of a Google search. We've stopped trusting reviews, so we go looking for strangers.
That assumption is easy to exploit. A niche question like "long-lasting office perfume" might get fifteen replies in total. Four planted ones that agree with each other can turn it into consensus. Moderators are unpaid volunteers, and they can't vet every helpful-sounding stranger.
Then the thread outlives the campaign. Google ranks it. AI tools quote it. A comment planted once keeps answering buyers for years.
Why it's wrong
I've been on Reddit for 15 years. It's one of the last places online where you can ask strangers "is this actually worth it?" and get an honest answer, often a brutal one.
Nobody wrote the deal down, but everyone knows it: you don't pretend to be someone you're not to sell me something. People there will tell you your favourite product is overrated. They'll warn you off a scam. They'll spend twenty minutes on a stranger's problem for nothing in return. That trust was built by millions of real people over years.
Fake accounts spend it. An ad is labelled, so you discount it. A planted comment borrows the credibility of real users, and uses it at the exact moment someone is trying to escape ads.
It doesn't stay on Reddit either. When AI tools repeat those threads, a bought opinion turns into "the answer", delivered with a neutral face to people who never saw the thread. It also breaks Reddit's own rules, and in the US it's exactly what new fake-review laws target. But the law isn't really my point. Even if it were legal, it would still be lying.
Things I won't do
Reviewing ads means I'll see a lot of tactics. Some I'll steal happily. A few I'll never touch, whatever they promise:
- I won't pretend to be a customer of something I'm promoting.
- I won't buy followers, comments or upvotes.
- I won't talk about my own work anywhere without saying it's mine.
- I won't quote a statistic I haven't checked.
- I won't sell anyone growth that depends on fooling the people they want to reach.
This list will grow as the series does.
How to spot it
You can't catch all of it. You can catch a lot:
- The same brand recommended by "different" people within hours, in a thread nobody else mentions it in.
- Replies that read like product copy: the full product name, the exact concentration, "lasts all day, I get compliments constantly".
- Click the profile. Months of bland posts about football or cooking, then sudden passion for one brand.
- One account recommending the same thing across many unrelated threads.
The honest version of Reddit marketing exists, and it's boring. Show up as yourself. Say who you are. Be useful even when it doesn't sell anything.
If your product needs fake people to recommend it, the problem isn't your marketing.