How Conbersa Discovered the Real Distribution Problem (Going From Agency to SaaS)

The founder distribution problem is that almost every brand and creator I have ever worked with rents their distribution from one person, one platform, or one algorithm, and when any of those three change, the audience walks away with them. I learned this the hard way running a UGC agency to $200k ARR before pivoting to build Conbersa, the agentic platform for managing real social media accounts at scale. This is the candid version of how that pivot happened: the pattern I kept seeing across clients, the YC office hours question that finally made the bottleneck obvious, the moment the agency revenue stopped feeling like progress and started feeling like a research budget, and the call I made to stop selling services and start building infrastructure. If you are a founder running a services business and feeling the same itch, this is the post I wish I had read 18 months earlier.

How the Agency Started and What It Actually Did

I started the UGC agency because the demand was obvious. Every consumer brand and SaaS company I talked to wanted more short-form video. None wanted to hire a creator in-house, train them, manage workflow, or absorb production overhead. Agencies that produced UGC at scale were closing deals fast, and we did the same.

The model was straightforward. We had creators and clients. We took briefs, produced clips, posted them on accounts the client owned or accounts our creators owned, and reported on reach. Unit economics were fine for a services business: 30 to 50 percent gross margins, $5k to $15k monthly retainers, enough word-of-mouth to keep the pipeline full.

We hit $200k ARR in the first year. That is the level at which the structural problems became impossible to ignore.

The Pattern I Kept Seeing Across Every Client

Once you run enough client engagements, the pattern stops feeling like coincidence and starts feeling like physics.

Every successful client engagement was carried by one creator. One person with the right voice, the right face, the right instincts about hooks and pacing. When that creator was on, distribution looked like magic. When that creator took a vacation, got sick, or wanted to quit, distribution cratered within two weeks.

This was not a creator-management problem we could solve by hiring more people. It was a structural property of how the agency made money. The leverage point in every engagement was a single human, and humans quit.

I would watch us close a client at $10k per month, ride them up to $15k as the program scaled, then watch the whole thing unravel within a quarter when the lead creator left to pursue their own brand. The audience the client thought they were paying us to build was attached to the creator, not the brand. The minute the creator left, so did the audience.

Brands consistently underestimated this. They thought they were paying for content production. They were paying for rental rights on a person's audience for as long as that person felt like working with us.

The YC Office Hours Question That Broke the Model Open

We ended up in YC office hours at the point where the agency was technically growing but the churn pattern was eating us alive. The partner running our session asked one question that I have not stopped thinking about since.

The question was: what happens to your business if your top creator quits tomorrow.

I knew the answer immediately. Revenue would collapse within 60 days. The clients tied to that creator would churn. The pipeline would slow because referrals were tied to that creator's reputation. We would lose roughly 40 percent of ARR in a quarter and would not recover within the year.

That is not a business. That is a leveraged bet on one person's career trajectory dressed up as a recurring revenue model.

The follow-up question was sharper. It was: what is the actual asset you are building.

I had no clean answer. We were building a roster of creators who could leave anytime. We were building client relationships that depended on those creators staying. We were building a brand reputation that was, again, attached to the creators rather than to the agency. None of those compound. None of them survive a key person leaving.

That conversation is the one I look back on as the moment the pivot became inevitable, even if I did not know it for another four months.

What I Eventually Realized the Real Bottleneck Was

I spent the next four months trying to fix the agency without rebuilding it. We hired more creators. We documented the production process so creators were more interchangeable. We built internal tools to manage briefs and posting schedules across more accounts. None of it worked, because none of it addressed the actual constraint.

The actual constraint was not creator throughput. It was distribution infrastructure. Every client wanted multi-account programs (50, 100, 500 accounts each posting variant content) and the infrastructure to do that without shadowbanned accounts was nonexistent off the shelf.

I started cataloging the problems we had to solve manually for every engagement. The list was long.

Each account needed an isolated device environment so platforms could not link them via fingerprint correlation. Each needed a dedicated geographic IP because shared IPs cascaded into network-level flags. Each needed warmup discipline because skipping warmup meant new accounts got shadowbanned in their first week. Each needed identity isolation so phone numbers, emails, and device IDs did not link the network. Each needed content variation pipelines so duplicate detection did not cascade through the portfolio.

We were rebuilding that stack from scratch for every client. And we could not charge for it directly, because clients did not understand what they were paying for. They thought they were paying for content. We were charging for content while subsidizing the infrastructure work that made the content reach anyone.

That is when the second piece clicked. The work no client will pay for directly is exactly the work that creates compounding leverage. If we could not charge for the infrastructure inside the agency, the infrastructure had to become the product.

The Decision to Stop Selling Services and Start Building a Platform

The math on the pivot was uncomfortable. We had $200k of agency ARR that paid the bills, a team trained on services delivery, and a pipeline of new client work I was walking away from. The platform we wanted to build did not exist. There was no product, no pricing model, no early customers waiting in line.

I called every YC partner I could get on the phone, and the message was consistent. Agency revenue is a research budget for the product underneath, but only if you actually use it as research. Keep treating it as the business and the agency keeps extracting time and energy from the product the agency exists to inform.

So we made the call. We told existing clients we would honor their contracts but were not taking new work. We pulled the team off services delivery and onto product development. We documented every infrastructure problem we had solved manually and turned it into a product spec.

The first build of Conbersa was, in retrospect, embarrassingly close to the manual stack we had been running for clients. Each account ran in an isolated device environment with a unique fingerprint, a dedicated geographic IP, its own warmup pipeline, identity isolation, and content variation hooks. The stack was not new science. It was the disciplines we already ran, finally written into software so a brand could run their own programs without rebuilding the infrastructure layer.

The first paying customer was not a former agency client. It was a brand that had tried to run their own multi-account program, watched 30 accounts get shadowbanned in three weeks, and was looking for the infrastructure layer they were missing. They told us, almost word for word, the same problem statement we had been solving manually for clients all year.

What Owned Distribution Actually Means and Why It Matters

The framing that emerged from the pivot, and the one that still drives how I think about Conbersa, is the difference between rented distribution and owned distribution.

Rented distribution is when your reach depends on a single creator, a single platform, or a single algorithm continuing to work the same way tomorrow as today. If the creator quits, you lose the audience. If the platform changes its rules, your distribution collapses. If the algorithm shifts, your reach evaporates.

Owned distribution is when your reach lives inside infrastructure you control: portfolios of accounts you own, devices and identities you operate, content variation pipelines you maintain, and audience relationships that survive any individual creator leaving or any platform tweaking its ranking model. The asset is the system, not the person.

Almost every brand I talk to is on rented distribution and does not realize it. They have a top-performing creator, a primary TikTok handle, or one Reddit subreddit driving most of their traffic. When I ask what happens if the creator leaves or the platform changes, the answer is silence followed by a fast topic change.

The pitch for multi-account distribution and TikTok distribution at scale is not that more accounts are inherently better. The pitch is that more accounts plus owned infrastructure means the brand survives any single point of failure: a creator quitting, a platform shifting, an algorithm changing. The brand owns the distribution because the distribution lives inside infrastructure the brand controls.

What I Would Tell Founders Running Services Businesses Today

Three things I would have told myself 18 months earlier.

The pattern across clients is the product. If the same problem shows up in every engagement and you keep solving it manually, that is not coincidence. That is a product begging to be built. Stop solving it manually and start writing the spec for the version that solves it once for everyone.

The work no client will pay for directly is the leverage. Services revenue pays for the work clients understand. The work clients do not understand but cannot succeed without is exactly the work that compounds. If you cannot charge for the infrastructure inside the engagement, the infrastructure is the product.

The pivot question is when, not whether. If you have read this far and the description is matching your business, the pivot is probably inevitable. The only question is how long you let agency revenue subsidize the product before you commit. The longer you wait, the harder it gets to walk away.

How Conbersa Operates Today

Conbersa is an agentic platform for managing social media accounts on TikTok, Reddit, Instagram Reels, and YouTube Shorts. Each account runs in an isolated device-grade environment with a unique fingerprint, dedicated geographic IP, and persistent identity. Brands and agencies run their own multi-account distribution programs on top of the platform without rebuilding device isolation, IP routing, identity management, and warmup pipelines from scratch.

The honest framing: Conbersa is the infrastructure layer the agency was rebuilding manually for every client, written into software so brands can run the programs themselves. We are not the content team. We are not the strategy consultancy. We are the layer underneath that makes content and strategy actually reach an audience without the program collapsing on platform detection. Brands running UGC at scale and anti-detection infrastructure sit at the center of who we built it for.

The pivot was the right call not because services businesses are bad. It was the right call because the problem we were solving was infrastructure, the leverage point was infrastructure, and the durable asset was infrastructure. Services revenue funded the discovery. The product is what survives when the discovery is done.

If you are a founder reading this and the pattern feels familiar, the question is not whether your services business contains a product. It probably does. The question is how long you wait before you build it.