In July 2024 I built JobReferral over a weekend. The idea was simple: companies post jobs with referral rewards, and people earn by referring friends who fit.
The first few days were encouraging. Within days, 66 people had signed up as referrers and 51 recruiters had joined the waitlist. Then I hit the wall every solo builder hits. Building the product had been the easy part. Getting it in front of the right people, consistently, was the real job.
So I ran an experiment I had been curious about for a while: paying micro-influencers to spread the word.
This is the honest write-up. It didn’t go well, but I learned more from it than from a lot of things that did.
Why micro-influencers
The case for micro-influencers sounds convincing. Big creators are expensive and their audiences are broad. Smaller creators, with around ten thousand followers, are supposed to have closer relationships with their audience, higher engagement, and lower prices. For a tiny product with a tiny budget, that seemed like a good fit.
JobReferral needed two kinds of people: professionals who might refer friends, and recruiters who might post jobs. A few voices that those people already trusted, talking about the product in their own words, felt like a faster way in than ads.
How I set it up
I used a micro-influencer marketplace rather than reaching out to creators one by one. The plan was:
- Hire two or three influencers with roughly 10,000 followers each.
- Give them a clear brief and scripts, so they didn’t have to figure out the product from scratch.
- Measure whether it moved the waitlist enough to be worth repeating.
On paper, a marketplace should make this easier. It handles discovery, payments, and coordination, and you get to focus on the message.
In practice, almost every one of those steps was harder than expected.
What went wrong
Communication was poor
Updates were rare. There were no clear timelines. I had to follow up several times just to get a list of influencers who might be relevant. For a small experiment, the coordination overhead was out of proportion to the size of the bet.
The right influencers were hard to find
This was the biggest problem. JobReferral was a professional product, aimed at people thinking about jobs, hiring, and careers. Most of the influencers available on the platform were in beauty, fashion, and lifestyle. Their audiences were real, but they weren’t my audience.
That may have been specific to the platform I used. But it points to something more general: follower count is the easiest thing to filter on, and the least useful. What matters is who those followers are and why they follow that person.
Everything took longer than planned
Even with scripts and every detail provided upfront, getting the first campaign live was slow. When you’re trying to learn quickly, a delay of a couple of weeks isn’t a minor inconvenience. It means your experiment is measuring a moment that has already passed.
How did it go?
Not great.
The results were underwhelming. My honest guess is that spending the same amount on targeted Facebook or Instagram ads would have produced more interest, and would have given me cleaner data about what was working.
I didn’t throw the idea out entirely. I could see it working at a different stage, with the right creators, once the product was fully live. But for that moment, with that product and that platform, it was the wrong tool.
Why it didn’t work
With some distance, I think the experiment failed for reasons that had little to do with influencer marketing as a whole.
The audience didn’t match the product. A professional, B2B-flavored product needs voices that professionals trust. A marketplace optimized for consumer brands couldn’t supply them.
I outsourced the most important decision. Choosing who speaks for your product is a strategy decision, not an operations task. Handing it to a marketplace saved me effort and cost me the outcome.
The test was too slow to teach me anything. By the time the first campaign was live, the product and my priorities had already moved on.
I had no baseline. Without a small paid-ads test running alongside, “underwhelming” was a feeling, not a measurement.
How I’d run it next time
If I tried this again, for any early-stage product, this is the checklist I’d use:
- Start from the audience, not the follower count. Make a list of people your users already listen to. They might have five thousand followers or fifty thousand. Relevance matters far more than reach.
- Go direct for the first test. Reach out to two or three creators yourself. You’ll learn how they think about their audience, which is half the value of the exercise.
- Ask for proof of audience. Ask about audience location, roles, and what their best-performing posts were about. If a creator can’t tell you who follows them, that’s your answer.
- Agree on timelines in writing. A one-week test that slips to a month isn’t a one-week test anymore.
- Give creators room to say it their way. Scripts are useful for facts, but an audience can tell when someone is reading an ad. Share the key points and let them explain it in their own voice.
- Track every link. Use a unique link or code per creator, so you can see exactly what each one brought in.
- Run a paid baseline next to it. Spend a similar amount on targeted ads in the same week. Then you’re comparing two channels, not guessing.
- Decide in advance what success looks like. Pick a number, such as waitlist signups or recruiter sign-ups, before you start, and decide what you’ll do if you hit it or miss it.
The bigger lesson
The real takeaway wasn’t about influencers. It was about distribution.
When building gets cheaper and faster, the hard part of a new product moves somewhere else: finding the right people, earning their attention, and doing it again next week. No single channel solves that for you. You learn it by running small, fast, honest experiments, and by writing down what didn’t work as carefully as what did.
If you’ve tried micro-influencers for a professional or B2B product, I’d genuinely like to hear how it went. Especially if it went better than mine.
Originally published on LinkedIn.