Selling Your n8n Workflow Once vs. Earning Every Time It Runs

# Selling Your n8n Workflow Once vs. Earning Every Time It Runs

You built a workflow that works. Now you want it to earn. You have two models to choose from, and they are not close.

Sell it once: package the workflow, list it on Gumroad, collect a flat fee per download. Revenue stops when sales stop, and every buyer keeps a copy of your JSON forever.

Earn per run: list the workflow where it charges every time it executes, and collect a markup on each run for as long as teams keep using it.

The one-time sale is the current default because it is the obvious one. It is also the one that leaves the most money on the table. Here is the math, the mechanics, and the cases where selling once still wins.

The current default: package it, list it, sell it once

Walk through how builders monetize n8n today and the path is well-worn. The official n8n template library is free — thousands of workflows you import in one click, no payment, and creators earn only through an affiliate program, not direct sales. So the money moves off-platform. Builders export the workflow as JSON, bundle a few together, and sell the bundle on Gumroad.

Browse Gumroad's n8n listings and you see the pattern. Single templates and small bundles run around $15 to $97. Larger "mega bundles" of 200 to 2,000+ workflows list higher, often discounted from a struck-through anchor price. The product is a JSON file (or a folder of them) plus setup notes. The buyer pays once, downloads, and imports.

This model has real advantages. It is simple. It works today. You get paid immediately, in full, at checkout. There is no ongoing support obligation baked into the price. For a builder testing whether anyone will pay for their work at all, a Gumroad listing is the fastest way to find out.

But two things are true about a one-time sale that are easy to overlook when the first payment lands.

Gumroad takes a cut, and it is not small on discovery. Per Gumroad's own pricing, a direct sale — someone arriving from your own link or profile — costs 10% + $0.50. A sale that Gumroad's Discover marketplace sends you costs 30%. So on a $97 direct sale you net about $86.80. On a $97 Discover sale you net about $67.90. That is the price of distribution when the platform, not you, brings the buyer.

The artifact is a file, and files copy. A workflow is JSON. Once you deliver it, the buyer has it permanently, can run it on unlimited jobs, can modify it, and — nothing technical stops them — can pass it along. You have no meter on its use and no visibility into whether it ran once or ten thousand times. Your revenue from that buyer is capped at the moment of sale. Their value from your work is not.

The real math: flat fee vs. per run over a year

Put numbers on it. These are illustrative — plug in your own — but the shape holds regardless of the exact figures.

One-time model. You sell a workflow for $97 as a direct sale. After Gumroad's cut you net roughly $86.80. Say you move 50 copies over a year. That is about $4,340 — a good year for a single listing. Then it flattens. Year two starts at zero again unless you keep finding new buyers, and every one of last year's 50 buyers already owns it outright.

Per-run model. You list the same workflow so it charges a markup each time it executes, and you set that markup — say $0.30 a run. One team that runs it 300 times a month pays you $90 from that one buyer, every month. Ten teams at that volume is $900 a month — about $10,800 a year — and it does not reset in January. It carries into year two and compounds as more teams adopt it.

The one-time column is not wrong. It is just bounded. The per-run column is the same expertise, uncapped. Revenue stops being a function of how many units you sell and starts being a function of how much work your workflow does.

This is the seller-side mirror of what buyers already pay for. On the buyer side, per-run pricing is how marketplace agents are billed — a few cents to a few dollars per task, as covered in AI agent pricing in 2026. Every cent a buyer spends running your agent is a cent your markup can sit on top of. The one-time model throws that recurring stream away for a single upfront payment.

What happens to your reputation after the sale

On a template marketplace, the transaction ends at download. The buyer's experience — whether the workflow ran clean, whether it saved them ten hours or broke on their first job — never comes back to you as a signal that helps your next sale. There is a review box, and that is the extent of it. Your reputation does not compound; it resets with each new listing.

A per-run marketplace inverts that. When the workflow earns from usage, usage is the metric, and usage is public. A high run count is proof the thing works. Ratings accrue against the same agent over time instead of scattering across one-off product pages. Your track record becomes an asset that makes the next buyer easier to win — the reputation compounds instead of evaporating at checkout.

That is the difference between selling a file and running a service. A file is done when it downloads. A service earns its standing every time it executes.

What it takes to relist a workflow as a per-run agent

The gap between the two models is smaller than it looks. You already did the hard part — you built and refined the workflow. Moving it to a per-run model is mostly repackaging, not rebuilding.

• Define the scope tightly. A per-run agent should do one job with a clear input and a clear output, so a buyer knows exactly what a "run" is and what they are paying for. If your Gumroad bundle crams twelve loosely-related workflows together, split them. Each becomes its own listing.
• Make the output judgeable. Buyers pay per run when they can tell a good run from a bad one at a glance. Return a clean, structured result, not a raw dump the buyer has to interpret.
• Set your markup. This is your price per run. You are not guessing at a bundle price a stranger will pay once — you are setting a per-execution rate that has to clear the value of one unit of work.
• Document the trigger and the inputs. Same setup notes you would ship with a Gumroad bundle, minus the "here's how to import the JSON" part, because the buyer never touches the JSON.

For most workflows, no code rewrite is required. The logic that worked as a template works as a per-run agent. What changes is the wrapper: scope, inputs, outputs, and price.

When selling once still makes sense

Per-run is the better model for most workflows worth selling, but not all. Sell once when:

• You are validating demand. A $27 listing tells you fast whether anyone will pay for this at all. Cheap to set up, cheap to learn from.
• The workflow is a one-shot tool. Some workflows run once per buyer — a one-time migration, a setup script, a bulk cleanup. There is no recurring usage to meter, so there is no recurring revenue to capture. A flat fee is the honest price.
• The buyer specifically wants to own and modify it. Some buyers want the raw JSON to fork and make their own. That is a sale of the artifact, not a service, and a one-time fee fits.
• You want zero ongoing relationship. A file sale ends at download. If you do not want to stand behind the workflow as it runs in other people's businesses, do not price it as if you do.

The test is simple: does a buyer run this once, or run it every week? If it is every week, a one-time fee is charging for one week and giving away the other fifty-one.

How Augex fits

Augex is a marketplace built on the per-run model. You package the workflow you have already refined as a deployable agent, set your markup, and earn every time a team runs it — the markup is yours, and the platform handles distribution and billing the buyer separately. The reputation compounds where usage is the metric: the more teams run it, the stronger the signal for the next buyer. Your expertise stops being a one-time solution and starts earning indefinitely. That is the whole idea behind *your expertise, unlimited reach* — the workflow you would have sold once earns from every team that needs exactly what you built.

FAQ

Can you make passive income from n8n workflows?

You can earn ongoing revenue, though "passive" oversells it. A one-time Gumroad sale pays once and then needs new buyers to keep earning. A per-run model pays every time the workflow executes, so revenue continues as long as teams keep running it — but the workflow still has to work, stay current, and be worth running. The income recurs; the upkeep does not fully disappear.

How much can you sell an automation workflow for?

On Gumroad, single n8n templates and small bundles commonly list from about $15 to $97, with large mega-bundles priced higher. That is a one-time figure. Under a per-run model there is no single "price" — you set a markup per execution, and total earnings depend on how many teams run it and how often. A workflow run hundreds of times a month across several buyers can out-earn a one-time bundle within the first year.

What is the difference between selling a template and selling an agent?

A template is a file. You sell it once, the buyer owns the JSON, and you have no meter on how they use it. An agent is a service. It charges each time it runs, you earn a markup on every execution, and your reputation accrues against the same listing over time. Same underlying workflow — different revenue model and different ceiling.

Is there a marketplace for AI agents?

Yes. Agent marketplaces let builders list a workflow as a deployable agent and earn per run, while buyers deploy it in minutes without building it themselves. This is distinct from a template library like n8n's, where workflows are free to import, and from a file store like Gumroad, where you sell the JSON once.

Sources

• Gumroad — Pricing — the fee a seller pays: 10% + $0.50 on a direct sale, 30% on a Discover marketplace sale.
• n8n — Workflow templates library — the official template library is free to import in one click, with creators earning through an affiliate program rather than direct template sales, which is why paid one-time sales happen off-platform.
• Gumroad — n8n workflow bundle listing — an example of the one-time n8n bundle listings sellers use today.
• Augex — AI agent pricing in 2026 — the buyer-side per-run pricing (cents to dollars per task) that the seller markup sits on top of.