Say a six-person web design agency decides to sell a template library for $29 a month. The spreadsheet looks encouraging. Two hundred members produces $5,800 in monthly recurring revenue, which is roughly one more retainer client without any extra sales calls. The decision looks obvious, and that is exactly the problem. When you evaluate a new revenue stream before launching it, the question “will this make money?” is really three separate questions wearing one coat.
Will it be profitable per unit, after every party who touches the money has taken their cut? Will it survive the gap between spending and getting paid? And will it leave the rest of the business better off, or will it quietly consume the attention that currently produces your highest-margin work?
An offer can pass the first test and fail the other two. That is the most common way new revenue streams go wrong. They are not bad ideas. They are decent ideas evaluated on one dimension.
Model the amount that lands in your bank account, not the price on the page
The $29 price is a headline, not a number you can plan with. What matters is the deposit: the money that actually arrives after platform fees, payment processing, billing fees, refunds, and payout charges.
Substack’s own pricing documentation is a clean illustration of how many layers exist in what looks like a simple product. The platform charges 10% of subscription revenue. For U.S. cards, it also cites Stripe card processing of 2.9% plus $0.30 and a Stripe Billing recurring fee of 0.7%. A custom domain is a one-time $50.
Run the arithmetic on an $8 monthly subscription. The platform takes $0.80. Card processing takes about $0.53. The recurring billing fee takes roughly $0.06. You keep about $6.61. Around 17% of the subscription never reaches you.
Now drop the price to $5. The same stack costs about $0.98, or close to 20%. Raise it to $15 and the stack costs about 15.6%. The percentage fees scale with price. The $0.30 does not.
Fixed per-transaction fees punish small, frequent tickets
This is the single most underrated fact in new-offer economics. Every pricing decision that increases the number of transactions makes fixed per-transaction fees more expensive, regardless of how good the margin looks in percentage terms.
A $4 digital download, a $7 monthly tier, a pay-per-use micro-charge: all of these look high-margin because the cost of goods is near zero. Then the thirty cents shows up on every single sale. At $4, that one line item is 7.5% of revenue before anyone has taken a percentage.
Two practical consequences follow. First, low-ticket models need either bundling or annual billing to be workable, because annual billing collects the fixed fee once instead of twelve times. Second, if you are comparing a $9 monthly plan against a $90 annual plan, the fee difference is real but smaller than the retention difference. Decide on retention and cash timing, and treat the fee saving as a bonus.
Platform fee stacks also vary by who you are and where your customers are. Patreon puts creator pages published after August 4, 2025 on its standard 10% plan, applies a 2.5% currency conversion fee when the payment currency differs from the payout currency, and charges $0.25 per U.S. direct-deposit payout. Two creators on the same platform, selling the same thing to the same audience, can have different economics based on when they signed up and where their supporters live. Stripe’s pricing pages also localize by region, so the widely quoted 2.9% plus $0.30 is U.S.-typical rather than universal. Check your own locale before you put a number in a model.
None of this is a reason to avoid platforms. It is a reason to build the fee stack into the model before launch instead of discovering it in month three. The same discipline applies to the costs already buried in your existing operation, which is a separate exercise worth doing properly: identifying the hidden costs in your business model usually changes what counts as a good margin.
Break-even tells you the floor, not whether to proceed
The U.S. Small Business Administration gives the formula in its plainest form. Break-even in units equals fixed costs divided by sales price per unit minus variable cost per unit. In sales dollars, it equals fixed costs divided by contribution margin, where contribution margin is the sale price minus variable cost, divided by the sale price.
The SBA’s calculator works on monthly totals, and it suggests adding about 10% to cover unpredictable miscellaneous expenses. That buffer is not padding. It is an admission that first-time cost estimates for a new offer are systematically low.
Here is where the formula becomes genuinely useful rather than academic: it forces you to classify costs honestly, and the fixed-cost line is where new revenue streams get flattered.
What usually gets left out of fixed costs
Subscription software for the new channel is easy to remember. The things people forget are the ones that do not arrive as an invoice.
Support time is the big one. A membership generates questions. A physical product generates returns. A new service tier generates scope conversations. If those hours come from someone who is currently billable, you have moved cost from one column to another and called it profit.
Then there is the setup cost that only happens once but has to be earned back: building the offer, writing the terms, configuring billing, and in some cases getting new legal, accounting, or sales-tax treatment in place. A new product category, a new country, or a shift from services to digital goods can all change your tax position. That is a question for your accountant before launch, not after the first quarter’s filings.
Shopify’s own tiers show how fixed and variable costs trade against each other. Basic is $29 a month with online standard card rates starting at 2.9% plus $0.30. Grow is $79 a month starting at 2.7% plus $0.30. Advanced is $299 a month starting at 2.5% plus $0.30.
Moving from Basic to Grow costs $50 more per month and saves 0.2% of card volume. That trade only pays for itself above roughly $25,000 in monthly card volume. Grow to Advanced costs $220 more for the same 0.2% saving, which needs something closer to $110,000 a month. If you are launching a new channel, you are almost certainly starting on the cheapest plan, and that is correct.
One detail matters more than the plan choice: Basic carries a 2% third-party transaction fee when you do not use Shopify Payments. At $25,000 in monthly volume that is $500, which dwarfs every plan upgrade decision above. The expensive line is rarely the one in the headline comparison.
Profit and cash are two different verdicts
A new revenue stream can be profitable on paper from the first month and still take six months to stop draining cash. The SBA’s guidance on projections points at why: for the first year, it advises quarterly or even monthly projections, with a prospective outlook over five years. Annual numbers hide the trough. Monthly numbers show it.
Three timing gaps cause the damage. You pay to build and launch before any revenue exists. You pay to acquire customers before they pay you, and on a subscription they pay you back in slices. And platform payout schedules, refund windows, and billing cycles all sit between a sale and usable money. Stripe made a one-time change to when product-usage fees appear on monthly tax invoices for U.S. accounts effective November 1, 2026, which is a small reminder that even when list pricing stays still, the timing of what hits your books can move.
So model two lines, not one. Monthly contribution profit, and the running cash balance attributable to the new stream. Find the deepest point of the cash trough and the month it turns positive. Then ask whether you can fund that trough without touching the working capital your core business needs. If you cannot, the answer is not “launch anyway and hope.” It is launch smaller.
Growth itself is a cash consumer, which is counterintuitive until you have lived through it. The mechanics are worth understanding separately, because more sales can create more cash-flow problems in exactly the moment the new stream looks like it is working.
Subtract what the new stream takes away
Almost every forecast for a new revenue stream is a gross number. The useful number is net new contribution: what the new stream adds, minus what it removes from everything else.
Two kinds of subtraction get missed.
Revenue cannibalization
If the agency sells a $29 template library, some share of buyers were going to become $4,000 project clients. A cheaper, faster option exists now, and some of them will take it. If one in fifty members is a displaced project client, the membership has to add a lot of members to stand still.
You cannot measure this precisely before launch. You can estimate it, and you should state the assumption out loud. The test is simple: ask whether the new offer solves the same problem as your highest-margin existing offer at a lower price. If it does, assume cannibalization is material until evidence says otherwise.
Attention and complexity
Complexity grows faster than contribution profit in a small business. A second revenue stream means a second set of customers, a second support queue, a second pricing conversation, a second reporting line, and a second thing to maintain when it breaks. None of those appear in a break-even calculation.
The honest version of the question is: what does this new stream cost us in the quality of the work we already do well? If the answer is “nothing,” you have probably not thought about it hard enough. Working out the true cost of serving a customer on the existing business first gives you a realistic baseline to compare against.
Model first, then run the smallest real test
There is a long-running argument between people who say model everything before launch and people who say forecasts are fiction so just ship something small. Both positions are half right, and the sequencing matters more than the debate.
Model first. Not because the model will be accurate, but because building it tells you which assumption is load-bearing. In almost every new revenue stream, one assumption carries the whole thing: conversion rate, retention, price point, or fulfillment time. The model’s job is to identify it. Then you test that one assumption cheaply, before you build anything.
The SBA frames pre-launch market analysis around industry outlook, target market, competitors, and trends. That work is worth doing, but it tells you whether a market exists, not whether your version of the offer will sell. Only buyer behavior answers that.
The cheapest useful test is usually manual delivery. Sell the thing to a small number of people and deliver it by hand, badly, with no automation and no platform. A membership can be a shared folder and an email. A new service tier can be three clients on a handshake. If you cannot deliver it badly by hand, you will not deliver it well at scale.
Manual delivery also surfaces the cost you most consistently underestimate: how long fulfillment actually takes. That number ruins more new revenue streams than weak demand does. A broader version of this discipline applies to any new venture, which is covered in more depth in how to evaluate a business idea before committing resources.
Choosing the model: recurring, one-time, or usage
Recurring revenue has a reputation it does not fully deserve. Predictability is real, and so is the effect on planning and valuation. But recurring also brings churn, ongoing support load, billing complexity, and the acquisition-payback problem where you spend today and recover slowly. A high-margin one-time or project model with no support tail can be the healthier business.
The deciding question is not which model is better in the abstract. It is whether the thing you sell has an ongoing reason to exist for the customer. If value is delivered once, a subscription is a billing structure pretending to be a relationship, and churn will tell you so within four months. If value accumulates or recurs naturally, recurring is the right shape, and it is worth designing the business model around recurring revenue properly rather than bolting a monthly charge onto a one-off product.
The interesting movement right now is hybrid. OpenAI renamed ChatGPT Team to ChatGPT Business on August 29, 2025, and introduced token-based billing for eligible Enterprise agreements, where usage is metered alongside contracted seat fees. That direction matters beyond software. It reflects a shift toward charging a predictable base for access and a variable amount for consumption, which gives the seller margin protection on heavy users and gives the buyer a lower entry price. If your new stream has highly variable delivery cost per customer, a flat price is a bet you may not want to take.
Set the floor with cost, set the price with value
Cost-plus pricing and value-based pricing are usually presented as rival philosophies. They are better used in sequence. Cost visibility is non-negotiable, because without it you cannot calculate break-even and you cannot know when you are selling at a loss. That gives you the floor.
The ceiling is set by what the customer believes the outcome is worth, and the only way to find it is to test prices with real buyers. Copying a competitor’s price is not value-based pricing. It is inheriting someone else’s cost structure, retention curve, and fee stack, none of which you can see. If your pricing already feels wrong on the existing business, the signals are worth reading before you replicate them in a new one: knowing when a pricing model no longer works is a prerequisite for designing a new one.
Assume the terms you launch on are not the terms you keep
Platform-based revenue streams carry a risk that rarely appears in evaluation checklists: the economics can change after you have committed. Patreon’s August 4, 2025 cutoff is the clearest recent example, putting newer creator pages on a different standard plan than earlier ones. Fee structures, plan names, packaging, and billing timing all move.
The practical protection is not avoiding platforms. It is building a margin of safety and knowing your exit cost. Two questions before launch:
If the platform’s total take rate rose by five percentage points, would this stream still be worth running? And if you had to move, what would you lose? Customer email addresses, payment relationships, reviews, and search rankings are the usual answers, and the ones you do not control are the ones that make you a tenant rather than an owner.
The decision, in four questions
When the modelling is done, the choice usually comes down to this.
What is the net contribution per unit after every fee, with the SBA’s 10% buffer for the costs you have not thought of yet? How deep is the cash trough, how many months until payback, and can you fund it without starving the core business? What does the existing business lose in revenue and attention, and is net new contribution still positive after that subtraction? And which single assumption is the model most sensitive to, and what is the cheapest way to test it this month?
A new revenue stream that passes all four is worth launching small. One that fails the third question is usually a distraction with good unit economics, which is the most expensive kind. If the strategic fit itself is the open question rather than the numbers, that deserves its own treatment: deciding whether a new opportunity fits your business comes before any of the math above.
The goal of this exercise is not certainty. It is to make sure that when the new stream disappoints you, it disappoints you in a way you can afford.
