How to work out which distribution channel is actually yours

Most distribution advice comes from people whose product was nothing like yours.

That's the whole problem in one line. You read a good post about how some company found its first ten thousand users, you copy the channel, and three months later you have a blog with forty readers and no idea what went wrong. The advice was fine. It was just about someone else's product.

There are a handful of frameworks that help with this, and they're worth knowing. They also answer different questions, and people tend to reach for whichever one they read most recently. So here is how I'd actually narrow the list, and where each framework earns its place.

Start with the arithmetic, because it deletes most of the options

In 2014 Christoph Janz wrote Five ways to build a $100 million business, which maps price straight onto motion. It is the fastest sanity check I know of, and it is five rows long.

Read across a row and the channel question mostly answers itself. Charge $10 a year and there is no version of events where a salesperson phones anyone, so the product has to carry its own growth. Charge $100k a year and you need a thousand customers in total, which makes a viral loop a waste of engineering time. Janz's own line about the rabbit tier is that there is no silver bullet, and the closest thing to one is inbound marketing.

The tiers assume venture scale, because the whole piece is about reaching $100M. Almost nobody reading this is doing that, so run the same division at whatever number you actually want. The table below is just Janz's arithmetic at smaller targets.

Customer counts on their own look reasonable everywhere in that table. Churn is what separates the rows.

Say you charge £5 a month and you want £100,000 a year. That's 1,667 customers. Fine, ambitious but imaginable. Now add churn. At 5% a month you're losing roughly 83 customers every month, so you need 83 new paying customers every month, forever, to stay exactly where you are. Not to grow. To hold still.

Charge £99 for the same product and the target becomes 85 customers, with about four a month to maintain it. Same product, same target, completely different problem.

Then work out your ceiling. If you want to earn back acquisition cost in six months, a product at £9 a month gives you £54 to spend winning a customer. Paid search and paid social are closed to you at that number. They work perfectly well for other people. They don't work for £54.

Peter Thiel described the awkward middle of this spectrum as the dead zone, where a deal is too small to justify a salesperson and too large to win through advertising. Products priced there have the hardest job, because neither end of the spectrum is available to them.

Almost nobody does this sum before picking a channel. It takes two minutes and it usually removes about half the list.

Then ask what your product actually gives the channel

This is the idea I'd tattoo on something if I were that sort of person. Brian Balfour, in Four Fits For $100M+ Growth, puts it plainly: "Products are built to fit with channels. Channels do not mold to products."

Everyone works the other way round. Build the thing, ship it, then go and find distribution. By then most channels have quietly closed, because a channel needs fuel and the product isn't producing any.

Some examples of what that means in practice.

SEO needs public pages. Not a blog you write by hand once a fortnight, though that helps. It needs pages that appear because people use the product. Pinterest is the classic version of this: users make boards, the boards are public, search engines index them, new users arrive and make more boards. If your whole site is nine marketing pages, you don't have SEO as a channel yet. You would have to build a product feature to earn it.

Virality needs something a non-user can see. A share button bolted onto a private tool does nothing. David Skok's work on the maths makes the point sharper: the variable that matters most isn't how many people each user invites, it's how fast the loop goes round. In his worked example, over twenty days, a two day cycle produces 20,470 users and a one day cycle produces over twenty million. Same invite rate. The only difference is frequency.

Partnerships need somewhere for a partner to attach. An API, an integrations directory, something. Without that there's no conversation to have.

Chris Dixon's "come for the tool, stay for the network" is the classic move for building the fuel deliberately: a single player tool gets you critical mass, the network keeps people there. Worth reading the pushback from Marc Bodnick alongside it, who argues that if the social part isn't in the product from day one it rarely appears later. Google Buzz, Apple Ping and Dropbox Carousel all had enormous existing user bases and none of it helped.

Then ask who is going to run it

This one is missing from every framework I have read, and for a solo builder it's the gate that closes the most doors.

Direct sales, trade shows, sustained partnership work and offline advertising all quietly assume there's a person whose job is that channel. If you're also writing the code, you don't have that person. A channel can be perfectly affordable and structurally supported and still be unavailable to you, because there are only so many hours.

That sounds obvious written down. It doesn't stop people building an outbound sales motion in month one and then wondering why the product stopped moving.

Whatever you pick will stop working

Andrew Chen's Law of Shitty Clickthroughs is the most useful thing on this list and the least fun. Banner ads on HotWired in 1994 got a 78% clickthrough rate. Facebook ads in 2011 got 0.05%. That's a 1,500x decline over the life of a channel.

Three things cause it. People learn to ignore whatever is new. Competitors copy anything that works, quickly. And as you scale past early adopters you start acquiring people who were always going to respond less.

There's an uncomfortable corollary. Once a channel strategy is written up in a popular post, it's already past its best, because that's exactly when everyone copies it. The published playbook is a lagging indicator by definition.

So the thing worth building is the habit of finding the next channel before it gets crowded, and spotting early when the current one starts to slide. Any particular channel is temporary.

Where the frameworks still earn their keep

None of the above makes the classic frameworks useless. They're just answering a different question.

Bullseye, from Gabriel Weinberg and Justin Mares' Traction, is a search procedure. Brainstorm all nineteen channels, run cheap parallel tests on the best two or three, then focus everything on whatever shows signal. Weinberg's budget guide is at most a thousand dollars and a month per test, and his advice on test design is to keep it low fidelity: run four Facebook ads rather than forty, because you're checking whether a channel has a pulse, not optimising it. Use Bullseye once the arithmetic has already cut the list down.

Growth loops, from Reforge, are how you tell a channel that compounds from one that needs constant feeding. A funnel takes input at the top and produces output at the bottom, and next month you need the same input again. A loop takes the output and reinvests it as input. When you're comparing two channels that both look viable, ask which one's output feeds back in.

Lenny Rachitsky's studies of how consumer apps got their first 1,000 users and how B2B companies found their first ten customers are the best empirical grounding available, and they carry an uncomfortable lesson. Almost none of what worked was a scalable channel. Slack begged and cajoled friends at other companies. Figma found designers on Twitter and cold emailed them. Airtable posted the beta on Hacker News. Manual, specific, unscalable work. The scalable channel comes later, and looking for one on day one is solving a problem you don't have yet.

The short version

Three questions, in this order.

Can you afford this channel, given your price and your churn? Does your product produce what this channel needs to run on? And is there anyone available to run it?

A channel has to clear all three. Most of the arguments people have about distribution are really arguments about one gate while ignoring the other two.

I got tired of doing this by hand, so I built DistroFit to do it. Put in your URL, your price and what you want to earn. It looks at your site, runs the sums, and tells you which of the nineteen channels are open, which are closed, and which would open if you changed something specific about the product. That last group is usually the interesting one.

It's free, there's no account, and it doesn't try to estimate your traffic, because nobody can do that accurately for free and it's better to say so.

If you'd rather read first, the channel pages go through all nineteen one at a time.