I have been freelancing on Fiverr since January 2023. Long enough that people ask me for advice about it now, which is funny, because for the first six months I had no idea what I was doing either. This is the honest version of what that work actually looks like, not the highlight reel.
Getting the first client is the hardest part, and not for the reason you think
Everyone assumes the bottleneck is skill. It is not. I could build the things people were asking for on day one. The bottleneck is that you are one unranked, unreviewed profile among thousands, and buyers filter by reviews before they filter by anything else. Nobody wants to be your first review.
What actually worked was underpricing two or three early gigs on purpose, telling the client up front that I was building a track record, and over-delivering hard enough that the review wrote itself. That is not a popular thing to admit, but it is the honest mechanism. Once I had five reviews at five stars, the algorithm started showing my gigs to people who never would have found them otherwise, and I could price normally again.
The brief you receive is never the brief you build
Almost every project starts as two paragraphs and a vague idea of “something like Stripe’s dashboard but simpler.” Nobody sends a spec. Part of the job, maybe most of the job, is turning that into something you can actually estimate and build, before any code gets written.
I learned to write back a scoped-out version of what I heard - pages, features, what’s in, what’s explicitly not in - and get a yes on that in writing before starting. Skipping that step cost me real money early on: a client who said “a simple booking form” meant a form, a calendar, email confirmations, an admin view, and Stripe deposits, and none of that was in the original price. I ate the difference that time. I do not anymore.
Scope creep is not malicious, it just needs a price tag
Most clients are not trying to get one over on you. They genuinely do not know where “small change” ends and “new feature” begins, because they are not engineers. “Can we also add login?” sounds like one sentence to them. It is not one sentence to me.
The fix was never to say no. It was to say yes, and here is what it costs and adds to the timeline. Once there is a number attached, most scope creep either gets paid for or quietly drops off the list on its own. The requests that survive that filter were the ones that actually mattered.
Some clients disappear, and you have to plan around that, not be surprised by it
At least once a year a client goes quiet mid-project. Sometimes it is their budget, sometimes their own client fell through, sometimes I genuinely never find out. Early on I took this personally, like I had done something wrong. I had not. It is just a normal failure mode of working with small businesses and solo founders whose own plans are less stable than a company payroll.
The practical fix was milestone-based payment, non-negotiably, on anything past a small gig. Build a piece, get paid for that piece, then continue. It protects both sides, but mostly it means a client vanishing costs me unfinished work instead of unpaid work.
What I actually build, most of the time
Not startups with a Series A and a design system. Small businesses that need a real website, and specifically ones that need something a page builder cannot do - contact forms with real backend logic, admin panels, integrations with something like Stripe or Twilio, occasionally a full internal tool a founder is tired of running out of a spreadsheet. React and Next.js for almost all of it, Node on the backend, because that is what lets me move fast without relearning a stack per client.
It is not glamorous work. Most of it will never show up in a portfolio, because most clients do not want their internal ordering tool on the public internet as a case study. But it is real, recurring, and it pays.
What surprised me after three and a half years
The technical part got easier fast. The non-technical part is still the whole job. Writing a scope clearly enough that there is no argument about it later. Saying a number out loud without hedging. Telling a client their idea has a cheaper version that gets them 90% of the value, even when the expensive version would have paid me more, because the ones you are honest with are the ones who come back.
That last part is the actual retention strategy. Repeat clients and referrals are where the real income is, not the first job from a stranger who found you in search results. Nobody refers you because you built exactly what they asked for. They refer you because you told them, correctly, what they should actually ask for.
I still do this alongside everything else I build. It is not the most exciting line on my resume, but it is the one that taught me the most about the parts of software that have nothing to do with writing it.