Lancer: He Shut Down a 15-Person Agency for a SaaS — $10K MRR in 60 Days

All dollar figures below are shown as disclosed in the source; the yen figures in parentheses are rough estimates at ¥150/$1.

Shut down your agency and turn what you struggled with there into a SaaS. Put into words, it’s a well-worn plot. But lining up the numbers Ivan Nedelkovski has published reveals something that mattered more than the idea itself. $10,000/month (about ¥1.5M) in monthly revenue 60 days after the commercial launch, now $20,000/month (about ¥3M). And almost all of that first $10,000 was brought in by a single referrer.

The timeline in numbers

What Lancer sells

Lancer is an AI agent for freelancers and agencies that automates writing Upwork proposals and surfacing promising leads. Pricing has two tiers. Pay-per-Lead is $149/month (about ¥22,000, $99/month billed quarterly), including 5 leads with additional leads at $19 each, one connected Upwork account, and unlimited proposals. Unlimited is $499/month (about ¥75,000, $333/month billed quarterly), with unlimited leads, up to 3 connected accounts, and done-for-you setup.

This pricing reflects who the product is really for: not individual freelancers, but “professionals who make a living on Upwork.” At $499/month, that’s essentially a fraction of the fee structure for someone earning $5,000/month on Upwork.

The tide turned the day one instructor switched over

As backstory, Ivan personally experienced the pain of writing Upwork proposals by hand. A $1,000 consulting gig, a $14,000 iOS app, a $10,000 MVP that grew into a $15,000/month retainer — that’s how he built up the agency’s revenue himself, giving him a high-resolution understanding of the problem.

But it wasn’t problem understanding that produced $10,000/month. What was decisive was one Upwork instructor switching sides. This instructor had originally been an affiliate for a competing service but, on seeing Lancer’s MVP, judged it “10x better” and switched who he recommended. Almost the entire $10,000 MRR reached 60 days after launch traces back to this single referral. Ivan himself has said that “finding someone who has distribution and hates the incumbent players was as important as the product itself.”

Why one referrer alone can generate $10,000/month

Breaking down the structure, what’s working here is that the market is unusually small. Ivan’s pre-launch market sizing found 18 million registered freelancers, but only about 30,000 actually active (2+ projects and $6,000+ earned in the last six months). Within that, the tiers break down to 30,000 earning $1,000+/month, 16,000 earning $2,000+, 9,000 earning $3,000+, and 5,000 earning $5,000+.

A base of 30,000 isn’t a scale you cast an ad net over. On the other hand, those 30,000 are densely concentrated around a handful of well-known instructors’ communities. In other words, in this market it isn’t the number of channels but the quality of “just one person” that dominates. Combined with a high per-customer price ($149-$499/month), even a small count like 30 users can hit $10,000/month. This path only replicates in a market where a small base and a high price point coexist.

The existence of a competitor also worked, paradoxically, in Lancer’s favor. To validate his idea, Ivan personally paid $500/month out of pocket to heavily use the leading competitor, stating plainly: “if their product were better, I wouldn’t have built Lancer.” That competitor had a three-year head start and had reached $2M ARR. Against the same 30,000-user base, assuming an average price of $200/month, $2M ARR implies roughly 830 paying customers, under 3% of the active user base. The competitor’s own numbers proved the market was far from saturated.

Four months and thousands of dollars burned on a design decision

What didn’t work is also disclosed concretely. Ivan initially copied the same design as the leading competitor: using Upwork’s “Agency Manager” feature, with third-party accounts sourced externally acting as proxies for users to send proposals.

This turned out to be a mistake. Because there was no direct access to the user’s own account, inbox automation was impossible, and inbound deal information (by Ivan’s estimate, about half of all opportunities) was lost. Furthermore, since the structure shared accounts across multiple agencies, it carried the risk of cascading bans. After losing “four months and several thousand dollars spent on accounts bought via cryptocurrency,” he switched to a design where users connect their own accounts directly.

The lesson Ivan draws is that the assumption “a competitor’s design, proven over three years, must be safe” doesn’t hold. A predecessor’s design is also the product of the constraints and past mistakes that predecessor is carrying.

A number that tells its own story: 3.3x users, 2x revenue

Easy to miss is the relationship between going from 30 users to 100, and the resulting revenue. User count grew 3.3x, but revenue only doubled, from $10,000 to $20,000. The average price fell from about $300 to about $200, down to two-thirds.

Given the two-tier $149/$499 pricing, a rough estimate suggests the initial 30 users skewed close to 40% Unlimited ($499), while the current 100 users skew closer to 10-20% Unlimited. This isn’t exact, given the quarterly discounts ($99/$333), but the trend reads as “the first to arrive were the top tier of the instructor community; what came after was a thinner layer.” Referral-driven inflow gets consumed from the highest-quality end first. Referrals now account for only about 10% of monthly revenue, and Ivan has diversified his acquisition into the free tool UpworkMRR.com, precise cold outreach built on that data, and posting on LinkedIn.

What transfers, and what doesn’t

What transfers easily is the discipline of validation: paying $500/month out of pocket to heavily use a competitor, counting the active tier by income bracket, and questioning a predecessor’s design choices. None of that requires capital or connections.

What doesn’t transfer is the precondition. Ivan ran a 15-person company and personally earned $500,000 on Upwork, which is what gave him the insider language of the instructor community. Being able to invest two full-time developers for six months on commercial v1.0 was also only possible because of the money and people left over after closing the agency. Looking only at the weekend MVP and reading “this can be done in 2 days” ignores the six months of work behind it.

The limits are equally clear. Paying customers number just 100, the market’s active tier is a small 30,000, and the business depends on the rules of a single platform, Upwork. Given the account-connection model, any policy change on Upwork’s side directly shakes the business’s foundation. What this case demonstrates is that “in a small, closed market, one person with distribution decides everything”, which is both a strength and, when that one person leaves, a fragility.

Related reading

• Two failures → $5M ARR → an eight-figure-dollar cash exit: ScrapingBee’s full journey
• How PDF.ai’s Damon Chen built customer acquisition as a solo developer