OctoLoops
8 min read

Keep stacking channels

Gojiberry reached $4M ARR in a year on five channels added one at a time, and the order is not a story about ambition: it is a cost ladder, which is why outbound and Reddit come first and paid ads come last

Quick answer: Gojiberry hit $4M ARR in a year on five channels added one at a time, never dropping the previous one: outbound, then Reddit, then content, then partnerships, then paid. The order is set by cost, not ambition. Beat one channel until it works, then add the next.

Dylan, a founder of Gojiberry, posted the channel history of a company he says passed $4M ARR in its first year and got into YC along the way. The headline is a shrug: there is no magic distribution channel. The list underneath the shrug is worth considerably more, because it is not a list of channels that worked. It is a list of channels in order, and every step up is gated by a revenue number.

His ladder, as he states it:

Read that as five separate stories and it is a survivor's anecdote, interesting and useless. Read it as one story about sequencing and it becomes something you can act on this week.

The line he says people miss

"None of these stages replaced the last one. They stacked. Outbound never stopped. Content never stopped. We just added the next lever once the current one was clearly working."

That reframes the whole list. The MRR labels are not phases you pass through and leave behind, like towns on a road. They are the moments a new thing got added to a pile that keeps growing. By $150k MRR the company is running all five at once, which is why the last rung is also the one where they start hiring: five channels is not a solo workload, and the hiring is a consequence of the stacking rather than a separate strategy.

The failure mode this describes is quiet and extremely common. You find something that works, run it for six weeks, get bored, and go do the shiny new thing instead of the boring effective thing. Then the numbers sag and you conclude the first channel stopped working. It did not. You stopped. A channel that is producing is not a channel that is finished, and the only good reason to spend less time on it is that you have hired someone to spend the time for you.

The order is a cost ladder

Here is the part the thread does not say out loud. Sort those five rungs by what each one costs to start, cheapest first, and you get his exact order. It was not a strategy so much as an inevitability.

We have run the same ten channels through the same disqualifying questions for 33 products, and paid ads came back as a recommended first move zero times out of 33. Gojiberry reached $150k MRR before touching it. Those two facts are the same fact: most channels are wrong for most companies at any given moment, and the thing that changes is not the channel, it is you.

The gate is revenue, not the calendar

"Once the current one was clearly working" is doing a lot of work in that sentence, and it is the only trigger he gives. Not a quarter, not a date, not a feeling. Clearly working has a testable meaning: you can predict roughly what a week of effort produces before you spend it. Ten emails gets two replies and one call, most weeks. Two Reddit posts gets four trials, most weeks. When the output of an hour has become boringly predictable, you have a machine rather than an experiment, and machines can be run at lower attention while you go and build the next one.

The inverse test matters more. If you are thinking about adding a channel because the current one is hard, that is not the signal. That is the moment before it starts working, which looks identical to the moment it will never work and feels considerably worse. Adding a second thing there gives you two channels run badly, and two run badly produce less than one run properly, because attention does not divide cleanly.

What does not transfer

Three honest caveats, because the thread is a single company in a single market and the numbers are his rather than a benchmark.

First, 10M organic Reddit views is an outcome, not a plan. Nobody requisitions that. He also volunteers that the traffic quality was not amazing and that what it did was flood a trial funnel, which means the rung only converts if you have a self-serve trial for it to flood. If your product needs a demo and a conversation, the same posts produce a traffic spike and no revenue, and you would be better off using communities the narrower way: finding the individual threads where someone has described your problem and answering those.

Second, YC arrives at rung four and he says the intensity went vertical. Some part of the partnerships rung is a network that came with the batch. Copying the sequence does not copy the address book.

Third, the MRR thresholds are descriptions of one company's economics, not thresholds you should hold yourself to. If your ACV is $50k, rung one may carry you past $75k MRR on its own and you should let it. The order transfers. The numbers on the gates do not.

The sentence buried in rung one

The most quietly loaded phrase in the whole thread is in the first line: they did outbound "using our own early product on ourselves." They were not emailing a purchased list. They pointed their own tooling at their own market and wrote to people showing intent, which is the difference between an outbound rung that gets you to $6k and one that gets you a spam folder.

That is also the difference this site keeps arguing about. Everybody bought the same lead list, so the list is not the asset. The evidence that a specific company has just become worth talking to is the asset, and a founder doing rung one by hand can find it: they read, they notice, they write four lines about the thing they noticed. Ten to fifteen emails a week with half of them replying is the shape of that, and it is small on purpose. Rung one is not a volume play. It is a filtering play that happens to produce revenue.

Where you actually are

Work out your rung by revenue rather than by preference, because preference will send you straight to rung three. Writing content is pleasant and asking strangers for their time is not, so the founder with no customers gravitates to the blog, which pays out in six months if it pays out at all. The ladder says do the unpleasant cheap thing first precisely because it is the only one that works with nothing.

If you are at zero, run founder-led outbound until it is predictable. If it is predictable, add the community rung and be a regular in the room before you need anything from it. Then, and only then, start writing. And when you add rung two, do not quietly stop doing rung one, which is the entire point he was making.

The first two rungs are the ones OctoLoops was built to carry. Give it your product and it works out who your buyer actually is, then keeps watching for the moment someone becomes worth contacting: the Reddit thread describing your problem, the funding round, the job post, the launch. Every one arrives with the event that qualified it and a link to the source, plus a draft you send from your own inbox. It will not climb the ladder for you. It makes the bottom two rungs cheap enough that you get to the third with revenue behind you.