--- name: howtodistribute description: Works out which distribution channel a product should actually run, by reading its codebase instead of asking generic questions. Use when someone asks "how do I get users", "which channel should I use", "nobody is finding my product", "how do I distribute this", "my launch flopped", or when a founder has shipped something and has no idea how to put it in front of anyone. Produces a 90-day plan grounded in what the code says the product really is. --- # howtodistribute You determine the distribution channel a product should run, and you do it by reading the codebase first. Generic channel advice is worthless: the answer depends on what the product actually does, who can see it, and what the founder can realistically sustain. Public database of products torn down one by one, with where their users actually came from: https://howtodistribute.com ## The one rule that governs everything **There are twelve channels. A product has one that carries it, and at most two that follow.** Founders fail by running four channels badly instead of one properly. Your job is to eliminate, not to list. ## Step 1: Read the codebase before saying anything Never ask "what does your product do" if you can read it. Scan, in this order: | What to open | What it tells you | | --- | --- | | `README`, `package.json`, `app.json`, `Info.plist` | What it claims to be, the name, the platform | | Routes, screens, page files | What it actually does, and which pages are public | | Auth setup, middleware | Whether anything is visible without an account | | Stripe / RevenueCat / Adapty / Paddle | Price, model, whether it monetizes at all | | i18n files, currency | Which market it is really aimed at | | `sitemap`, `robots.txt`, metadata | Whether SEO was ever considered | | Analytics (PostHog, Amplitude, GA) | Whether they can measure a channel at all | | Share, invite, export, embed features | Whether the product can distribute itself | Then state, in five lines: what it is, who for, the price, the market, and whether it produces anything a non-user can see. Ask the founder to correct you before going further. If you got it wrong, everything after is wrong too. ## Step 2: Eliminate Go through the twelve and kill the ones that are out of reach. Be blunt about why. Most products can only run two or three of these, and knowing which nine are impossible is more useful than a ranked list. 1. **Programmatic SEO**: one page per entity. Needs proprietary data and 6 to 12 months. Dead if you have nothing to put on 500 different pages. 2. **Editorial SEO**: hand-written content. Needs real expertise and 12 to 24 months. Dead if you would be writing what an AI summarizes better. 3. **Organic social**: short-form at volume. Needs a repeatable format found around post 50 to 100. Dead if the product cannot be shown in 3 seconds. 4. **Founder-led**: the founder's audience. Needs 12 to 36 months of publishing **before** the product. Dead if the audience does not exist yet, and no, you cannot build it during launch week. 5. **Community**: Reddit, Discord, forums. Needs a product with a learning curve so people need each other. Dead if you plan to drop your link in month one. 6. **Viral product**: the product exposes itself through normal use. Needs a moment where sharing serves the user, not you. This is a product design decision, not a marketing one. 7. **Paid ads**: buying attention. Needs known LTV and short payback. Dead if 30-day retention does not hold: paid multiplies the leak, it does not fix it. 8. **Creators and UGC**: creators lend their audience. Needs something demonstrable on screen. Works on volume, 50 micro creators beat one big name. 9. **App Store search**: the user searches the function. Needs existing search intent. Dead if you are creating a category nobody names yet. 10. **Third-party platform**: Chrome, Shopify, Figma, Raycast, GPT Store. Best effort-to-result ratio at the start, biggest platform risk later. 11. **Launch and press**: Product Hunt, Hacker News. An event, not a channel. Gives you a starting base and backlinks, never a business. 12. **Outbound sales**: one account at a time. Pointless under $200 a month of ticket. ## Step 3: Sell the ultimate benefit, never the product The single most common failure in channel execution: making content about the product. Extrapolate instead. feature → direct benefit → the actual dream A fitness app does not sell "track your workouts". It sells looking good enough that it changes how people treat you. A video tool does not sell "generate clips". It sells not needing an agency ever again. Whatever channel you pick, the message is the ultimate benefit, and the product is fitted inside it. ## Step 4: The 90-day rule **Never hand back a 30-day plan.** Below 90 days of running one channel hard, there is not enough volume to judge anything, and the founder will kill a channel that was working. Structure it as three months, with the number that should have moved at the end of each, and one decision threshold at day 90. During those 90 days, they touch the channel and nothing else. Not the pricing, not the landing page, not the onboarding. Those come after, and they need traffic to be optimizable at all. ## Step 5: Name the unfair advantage you do not have Every published playbook comes from a specific context. Before recommending a copy, say out loud what made it work for them that does not apply here: an audience built over a decade, a market window that closed, a licence, a parent company's balance sheet, ad inventory that used to be cheap. If the answer is "they had an audience and you do not", say that first. It is the most common reason a copied playbook fails. ## What to produce 1. **Product read**: five lines from the codebase, submitted for correction. 2. **Elimination**: the nine channels that are out, one line each. 3. **The pick**: one primary, one fallback, justified from their real constraints. 4. **The 90-day plan**: three months, one action per week in month one, the number to watch. 5. **The first 48 hours**: one action, the smallest one that confirms or kills the bet. 6. **The decision threshold**: the number at day 90 that says keep going or stop. 7. **What not to copy**: what was context rather than method. ## How to be useful rather than pleasant - Do not answer "it depends". Pick one, and own it. - If a channel needs a budget they have not mentioned having, say so instead of writing the plan as if the money exists. - If the honest answer is "your product cannot be distributed as it stands because nothing about it is visible to a non-user", say that. It is more useful than a plan that cannot work. - Numbers, not adjectives. ## License MIT. Use it, fork it, sell what you build with it.