A consultant sells a $4,000 operations audit. It takes about eighteen hours to deliver. There are no materials, no subcontractors, no software beyond what she already pays for. On paper the margin looks close to 85%. Twelve months and forty audits later, the company has almost nothing left over. Nobody stole anything. The offer was exactly as profitable as it looked.
The difference between a profitable offer and a profitable business is the gap between what one sale earns and what the whole operation has to pay for. One sale can clear a healthy margin while the company that produced it quietly loses money every month. That is not an edge case. For young businesses it is the normal situation, and misreading it is how founders end up scaling something that gets worse with volume.
An offer is a transaction. A business is everything that transaction has to carry.
The price you charge has to survive four subtractions before it becomes money you keep. People who say “my offer is profitable” have usually only done the first one.
What it costs to keep the promise
This is the obvious layer: materials, subcontractors, software licences used only for that client, hours of actual delivery. It is also where the single most common distortion lives, which is the founder’s unpaid labour.
Take the audit above. Eighteen hours of senior work went into it. If you priced that labour at what you would have to pay a competent replacement, the 85% margin shrinks immediately. Use $100 an hour as a placeholder and $1,800 of the $4,000 is gone before anything else is counted. Whether that number is right for your market is your problem to defend, but the exercise is not optional. Call it the phantom salary test: put a market wage for every founder hour into the cost line and see what survives. If the offer only works at zero founder pay, you do not have a profitable offer. You have a job with good gross margins, which is a perfectly reasonable thing to own as long as you know that is what you bought.
The rounding-error tax
Between the price on your invoice and the money in your account sits a set of costs that are each too small to notice and collectively large enough to decide the year.
Stripe’s standard domestic card pricing is 2.9% plus 30 cents per successful transaction. On a $4,000 payment that is $116.30. If the card is international, Stripe adds 1.5%, which is another $60. If currency conversion is required, add 1%, another $40. The same sale, bought by a client two countries away, costs $216.30 to collect instead of $116.30. That is 5.4% of the price gone before you have done any work.
Refunds are worse than most people model them. Stripe states that payment processing fees from the original transaction are not returned on most refunds under standard pricing. So a refunded sale is not a cancelled sale. It is a sale you paid to make, delivered against, and then gave back. Disputes carry their own charge: Stripe’s dispute received fee is $15 per dispute. Two disputes spread across a hundred sales is thirty cents per sale, which sounds like nothing, and that is exactly the point.
Add rework, support time you did not quote for, the discount you gave to close, and the two hours of scope you absorbed to keep a client happy. None of it appears in your pricing spreadsheet. All of it appears in your bank balance. This is the territory of hidden costs inside the business model, and the reason offer-level margin and company-level margin drift apart over time.
What it cost to find the buyer
Acquisition cost belongs to the offer, not to some separate marketing bucket. The honest version includes the attempts that failed. If you wrote six proposals to win two engagements, the cost of the four losing proposals belongs in the cost of the two wins. Agencies and consultancies routinely skip this, which is why their offers look profitable and their quarters do not.
Once acquisition is inside the number, you have contribution margin: what a sale actually contributes toward everything the business owes regardless of whether it sells anything. This is the number that should govern the decision to do more of something, and it is the heart of unit economics for a small business.
What the sale has to carry on its back
Rent, admin, accounting, the tools you pay for whether or not anyone buys, the half of your week spent on things no client sees. Fixed costs do not care how elegant your offer is. They are absorbed, or they are not.
This is where a genuinely profitable offer can sit inside a loss-making business. Forty audits a year at a real contribution of $1,200 each generates $48,000 to absorb overhead. If overhead is $70,000, the offer is fine and the business is insolvent. The fix is not a better offer. It is either more volume, a different cost base, or a different model.
The tests arithmetic cannot do
Once the four subtractions are done you have a number. The number is still not an answer, because two of the most important questions about a business are not arithmetic questions at all.
The hundredth sale test
Your offer is proven by sale one. Your business is proven by sale one hundred, and specifically by the hundredth sale you did not personally touch.
Early customers are not a sample. They are the people most eager to buy, served by the most motivated person in the company, during the period of maximum attention. Almost everything that destroys margin at scale is invisible in that group. So run the sale forward and ask three concrete questions. Who delivers it when you are busy, and what does the quality drop cost you? Does the delivery stay standard, or does each new buyer bring a variation that turns a product back into custom work? And is the hundredth buyer as good as the first, or have you worked your way down to people who need more convincing, more hand-holding and more discount?
A model where every new sale adds bespoke work is not a scalable business no matter how good the margin looks today. That is a separate judgement from profitability, and worth making deliberately when you assess whether your business model can actually scale.
Borrowed margin
The second non-arithmetic question is where the profit comes from and who controls it.
On 5 March 2024, Google’s core update introduced new spam policies covering expired domain abuse, scaled content abuse and site reputation abuse. On 19 November 2024 it updated the site reputation abuse guidance and stated plainly that moving content to a subdomain or subdirectory does not resolve the underlying issue. Businesses whose entire acquisition engine sat on top of those practices did not become less efficient at selling. Their offers were unchanged. The rules under them moved.
The general principle matters far beyond SEO. If your margin depends on a rule you did not write, a platform you do not own, a channel price you do not set, or one relationship you cannot replace, that margin is rented. It can be excellent rented margin. Plenty of good businesses rent. But rented margin should be valued differently from owned margin, and a business with one traffic source, one marketplace, one large client or one founder who closes every deal is carrying a risk that never shows up in the offer’s P&L.
The same offer can belong to a good business or a bad one
Here is the part that most explanations skip. Offer profitability is not a fixed property of the offer. It is largely determined by two structural choices made at the business level: how you charge, and how you buy your costs.
Pricing architecture decides whether your best customers become your worst accounts
Intercom’s published pricing combines seat-based charges with usage-based AI pricing: Essential at $29 per seat per month, Advanced at $85, Expert at $132, and its Fin AI Agent priced from $0.99 per outcome. That second layer is not a pricing gimmick. It is a solution to a structural problem.
Under pure seat pricing, a customer who triples their support volume costs you more to serve and pays you exactly the same. Their account margin erodes as they succeed with your product. Tying part of the price to the thing that drives your cost keeps revenue and cost-to-serve moving in the same direction. If you sell anything where heavy users cost materially more than light users, your pricing architecture is the main lever you have, and it belongs to the business model rather than the offer. Working out the true cost of serving a customer is what tells you whether you have this problem at all.
The reverse case is just as instructive. Mailchimp’s free plan includes up to 250 contacts and up to 500 sends per month, with Essentials from $13 a month and Standard from $20 on the displayed base tier. The free plan has no offer-level profitability whatsoever. It is an acquisition cost wearing a product’s clothes, and it is defensible only if conversion and retention pay for it at the business level. Notice what the caps are doing. They are not product decisions. They are a budget. A deliberately unprofitable offer with a hard ceiling on how unprofitable it can get is a very different thing from an unprofitable offer with no ceiling.
You can buy margin with commitment, and you pay for it in flexibility
Amazon’s EC2 On-Demand pricing is pay-as-you-go with no long-term commitment and a 60-second minimum, with low-cost T2 instances starting at $0.0058 per hour. AWS says Savings Plans can cut EC2 costs by up to 72% against On-Demand, and that Spot Instances are available at up to a 90% discount.
Read that as a lesson about cost structure generally rather than about cloud computing. The same workload can be a pure variable cost or a largely committed one, and the choice changes which failure mode you are exposed to. Commit, and your margin per sale improves at volume. Commit, and a quiet quarter now hits a fixed cost that does not shrink with your revenue. Every discount for commitment, whether it is a cloud plan, a bulk purchase order or a twelve-month lease on bigger premises, is a bet on your own forecast. The offer looks more profitable afterwards. The business has become more fragile.
Recurring revenue does not settle the argument
The standard advice is that recurring revenue is simply a better model. It is more useful to be precise about what it actually buys you.
HubSpot reported 2025 subscription revenue of $3.06 billion, a gross margin of 84%, and a GAAP subscription margin of 85.4%. Those are the numbers people point to when they argue that recurring software is the superior business model, and at that scale it is hard to argue. But look at another line from the same filings: average subscription revenue per customer went from $11,343 in 2024 to $11,414 in 2025. That is growth of less than 1% per customer.
An 85% margin tells you the offer is strong. It does not tell you where next year’s growth comes from. In that case it is coming from customer count and retention, not from each relationship getting richer. That is a business-level fact that no amount of offer analysis would have surfaced.
So my position on the recurring-versus-one-time debate is that it is usually framed around the wrong variable. Recurring revenue buys predictability and the confidence to spend on acquisition. It does not buy profit. It also converts a one-off sales obligation into a permanent service obligation, which is a real cost that founders discover in year two. A one-time high-ticket offer can produce better cash and far lower cost-to-serve. The variable that decides which is better for you is what happens after the money arrives, not how often the money arrives. If you are going that route anyway, the question is how to design the model around recurring revenue rather than bolting a subscription onto a service that was never built for it.
When it is safe to scale a profitable offer
Both camps in this argument are half right, and the resolution is sequence rather than preference. Offer economics come first as a gate. Business economics come second as the verdict.
A negative contribution margin is never fixed by volume. If each sale loses money after direct costs, leakage and fully loaded acquisition, scale is just a faster way to reach the same ending. That gate is non-negotiable. But passing it proves very little on its own. A profitable offer is evidence, not proof.
Before you put real money behind something, I would want three further things to be true. The margin has survived a step up in volume that you did not personally manage, because that is the moment delivery quality and support cost show their real shape. Cash converts fast enough to fund the next cohort of customers without borrowing, since contribution margin arriving in sixty days cannot pay for advertising that leaves your account today. This is precisely how more sales create cash-flow problems in otherwise healthy companies. And you can say out loud what happens if the largest single dependency, whether that is a channel, a client, a platform or you, disappears on Monday.
One more thing worth saying. You will be told that the ratio of lifetime value to acquisition cost must be at least 3:1, or that payback must land inside twelve months. These rules get repeated constantly and almost never come attached to a sector, a date or a source. Treat them as conversation starters rather than thresholds. The right payback period for a business with twelve months of runway and no credit line is nothing like the right one for a company with committed capital, even if the offer is identical.
Two sets of books in your head
The habit worth building is simple. Keep two views running at once and never let one answer the other’s question.
The offer view answers whether you should sell this again tomorrow. It is fast, it is granular, and contribution margin is its verdict. The business view answers whether you should build around it, and it only resolves across a quarter or a year, once overhead absorption, retention, cash timing and dependency risk have had time to show themselves.
The standard planning templates have always known this. The SBA’s business-plan framework asks you to describe what you sell and what makes it unique in one part of the document, and to include income statements, balance sheets and cash-flow statements in another. Those are deliberately separate sections because they test separate things. The trouble is that founders find the first part much more enjoyable to write, and plenty of businesses never really finish the second. If you want to go further on this, the related distinction between revenue growth and profitable growth works on the same logic one level up.
Nothing here argues against a strong offer. A strong offer is the hardest thing to build and the thing most companies never get right. It just means that when someone tells you their product makes money on every sale, the correct response is not congratulation. It is a question about the hundredth one.
