A small coffee roaster looks at twelve months of sales and finds one clear hero: the 12-ounce single-origin bag. It outsells everything. It gets the homepage slot, the ad budget, the extra shelf space. Two years later the business is bigger and the bank balance is flatter, and nobody can explain why.
The explanation is usually sitting inside that hero product. The unit economics every small business owner should understand start with the gap between “this sells a lot” and “this makes money,” and that gap is almost always wider than it looks on a revenue dashboard. A single-item shipment of a light, cheap product is one of the worst parcels you can send. It carries the same packaging, the same pick-and-pack labour, the same payment fee, and roughly the same postage as an order three times its value.
Unit economics is simply the practice of asking what one more of something earns you and what one more of it costs you. The hard part is almost never the arithmetic. It is choosing what “one” means, and then being honest about which costs genuinely belong to it.
Pick the unit before you touch the math
Nearly every guide on this subject skips straight to formulas. That is why so many owners run the numbers, get a comfortable-looking answer, and still cannot explain their own bank balance. If the unit is wrong, everything built on top of it is decoration.
The “one more” test
Here is the test I would use: your unit is the smallest thing that, when it happens one more time, causes new money to leave your business.
Run it out loud. One more candle sold off the shelf costs you wax, wick, jar, label. That is a product unit. One more online order costs you the product, the box, the postage, the card fee, and somebody’s three minutes at the packing bench, regardless of whether the box holds one candle or four. That is an order unit. One more subscriber costs you almost nothing this month but commits you to serving them for as long as they stay. That is a customer unit.
Notice that the test produces different answers for businesses that look similar on paper. A retailer with a physical shop and a retailer shipping parcels have the same products and completely different units, because shipping attaches cost to the basket rather than to the item.
What the unit usually is, by model
- Retail and wholesale: the product, by SKU. Averages across a catalogue hide the handful of items selling below their real cost.
- Ecommerce: the order. Shipping, packaging, payment fees, and returns all attach to the basket, not the item. Analyse SKUs too, but make the order your primary unit.
- Services and trades: the job or project. Not the hour. The hour is your capacity; the job is what you sell.
- Subscriptions and retainers: the customer, measured over their whole relationship, because the cost to win them lands once and the revenue arrives slowly.
- Restaurants, clinics, salons: the cover, the appointment, or the chair hour. Whatever your capacity is actually rationed by.
The U.S. Small Business Administration’s guidance quietly backs this up. Its break-even calculator is built around a single product or service, and it tells owners whose sales vary from month to month to calculate their products or services individually rather than lumping them together. That is a small instruction with large consequences. A blended number across a mixed catalogue tells you almost nothing you can act on.
There is a real argument in this space about whether to work at the product level or the order and customer level. I do not think it resolves in favour of one side. SKU-level discipline is what catches an individual item bleeding money inside a healthy-looking average. Order-level and customer-level analysis is what catches a business where every product is fine and every basket is still unprofitable. Pick your primary unit based on where cost actually attaches, then check the other level at least once a year.
Three numbers, in the right order
Once the unit is settled, the core calculation is short enough to do on paper.
The SBA defines contribution margin as the sale price per unit minus the variable cost per unit, divided by the sale price per unit. That gives you a percentage. The dollar figure sitting in the numerator, price minus variable cost, is the one I would actually write on the wall, because it tells you how many dollars each sale contributes toward the costs that arrive whether you sell anything or not.
From there, the SBA’s break-even point in units is fixed costs divided by sale price per unit minus variable cost per unit. In plain terms: how many of these do I need to sell before the month stops costing me money?
Those two lines are the whole foundation. Everything else in this article is about getting the inputs right.
Costs that arrive with the sale, costs that arrive with the month
“Variable” and “fixed” are technically correct and practically slippery, because owners keep sorting costs by how they feel rather than by how they behave. Try the plainer version. Some costs arrive with the sale: materials, postage, the card fee, the commission, the subcontractor, the installer’s drive time. Some costs arrive with the month: rent, insurance, software, your bookkeeper, the base salary of the person at the front desk.
Only the first group belongs in your unit calculation. The mistake that distorts everything is sorting a sale-driven cost into the monthly bucket because it happens to be paid monthly. Your ad platform bills monthly, but if ad spend scales with orders, it is behaving like a cost of the sale. Your customer support person is on salary, but if one product generates four times the support tickets of anything else, that time is a cost of that product in everything but accounting treatment.
Gross margin, contribution margin, net profit
Gross margin covers the cost of the goods themselves. Contribution margin covers everything that moves with the sale, which includes fees, shipping, returns, commissions, and touch time. Net profit is what remains after the costs of simply existing.
The old advice that gross margin tells you whether a product is worth selling no longer survives contact with how businesses actually operate. A 60% gross margin product can land at a negative contribution margin once you add a $9 parcel, a 3% card fee, a discount code, and a one-in-five chance of coming back. The distance between gross and contribution is where most small businesses lose money they never see. If you want to go further on that, it is worth reading how to identify hidden costs in your business model.
Where the money actually leaks
Ranking articles on this topic tend to stop at the formula. The formula is the easy part. Here is what quietly rewrites the inputs.
Fees and discounts
Payment processing, marketplace commission, and platform fees are costs of the sale, full stop. So are discounts. A standing 15% welcome code is not a marketing line item sitting somewhere else in your accounts; it is a permanent reduction in the price side of your contribution margin on every first order you take. Calculate those orders at the discounted price, because that is the price you actually charge.
Shipping, which moved under everyone’s feet
If your per-order spreadsheet was built more than a year or two ago, the postage assumption inside it is probably wrong. USPS shipping services price changes took effect on 13 July 2025, including increases of roughly 6.3% for Priority Mail, 7.1% for USPS Ground Advantage, and 7.6% for Parcel Select. Current posted USPS Ground Advantage pricing starts from $8.40 at the Post Office and from $7.33 commercial.
Look at that floor against a typical small-ticket order and the problem becomes obvious. On a $25 product, a single-item parcel can consume a third of the price before you count the box. Shopify’s operations guidance treats cost per order and shipping cost as a percentage of revenue as profitability metrics in their own right, alongside contribution margin, which is the right instinct. Shipping is not logistics admin. It is a line in your pricing.
Returns are a first-order cost, not an afterthought
This is the input that has shifted most and the one small businesses are slowest to model. The National Retail Federation projects total retail returns in 2025 will reach $849.9 billion, estimating that 15.8% of annual retail sales and 19.3% of online sales were returned. It also estimates that 9% of all returns are fraudulent.
Nearly one in five online orders coming back changes the arithmetic completely. The returned unit costs you the outbound postage you already paid, the return postage if you cover it, the inspection and restocking time, any damage to the item, and the payment fee on a refunded sale. Average those costs across every order you ship, not just the returned ones, because that is how they actually hit you.
Then there is the policy question. The evidence here genuinely pulls in two directions, and anyone telling you there is one right answer is not reading it carefully. On the demand side, DHL’s 2025 report says 64% of global shoppers have returned an item to an online retailer, 79% abandon their cart if their preferred returns option is not offered, and 55% mainly buy from retailers offering free returns. Shopify, citing NRF research, notes that 82% of consumers say free returns matter when shopping online. On the cost side, among retailers that charge for returns, 40% point to higher processing costs and 40% to higher carrier shipping costs.
My position: free returns are a conversion investment, not a customer service nicety, and you should treat them exactly like ad spend. You are buying a measurable lift in orders at a known cost per order. If you cannot state what that lift is, you are not investing, you are donating. Thresholds, category-specific policies, and higher prices on high-return categories are all legitimate ways to keep the conversion benefit without carrying the full cost on every sale. And for genuinely low-value items where getting the product back costs more than the product is worth, returnless refunds have become normal practice for a reason worth taking seriously: some units are cheaper to write off than to process.
Touch time, the service equivalent of shipping
Service businesses have the same leak with a different name. The quoted job assumes six hours. The real job includes the site visit, three rounds of revisions, two rescheduled calls, the invoice chase, and the follow-up email thread that runs for a month. None of that is in the quote. All of it is a cost of the sale.
Track touch time per job for one month, including admin and communication, and you will usually find that your contribution margin per job is a different number from the one you have been quoting against. It is the single highest-value measurement exercise available to a service business, and it feeds directly into knowing the true cost of serving a customer.
When acquisition cost belongs in the calculation
Stripe defines customer acquisition cost simply as what it costs to turn a prospect into a paying customer. The useful part is what comes next. If you charge $50 a month and spend $500 to acquire a customer, it takes ten months to break even on that acquisition. The customer is profitable. The tenth month is where profitability begins.
That example explains a failure mode that catches growing businesses repeatedly. You can have healthy contribution margins, genuine demand, and a rising customer count, and still run out of cash, because every new customer is a ten-month loan you are making from your own working capital. Growth accelerates the lending. This is the mechanism behind the uncomfortable truth that more sales can create more cash-flow problems.
So should acquisition cost sit inside your unit economics? My answer is staged. Get contribution margin right first. An owner-operator who has not yet separated sale-driven costs from monthly costs will produce a blended acquisition number that is noisy, hard to allocate, and more likely to mislead than help. Layer in acquisition cost once marketing spend becomes material, and when you do, follow the approach Stripe describes for subscription teams: segment it. Paid against organic, new against repeat, channel against channel. A blended figure averages your best channel with your worst and tells you to keep funding both.
One thing to put down and not pick back up: the idea that a 3:1 lifetime value to acquisition cost ratio is a universal target. It is repeated everywhere and anchored almost nowhere outside subscription software. For a landscaping company or a bakery, the far more useful questions are how many months the payback takes and whether your cash position can survive that wait. Payback period is a cash question. Ratios are a storytelling question.
Per unit, per hour, or per dollar: optimise what is actually scarce
Here is the refinement that separates owners who have run these numbers for years from those running them for the first time. Contribution margin per unit is the right measure only when units are the thing you are short of. Usually they are not.
Ask what your real bottleneck is. If it is your own hours or your team’s hours, the number that matters is contribution margin per hour of capacity, and a $4,000 project at 40% margin can be strictly worse than two $900 jobs at 35% that take a quarter of the time. If the bottleneck is cash tied up in stock, the number is contribution dollars per dollar of inventory per month, and a thin-margin product that turns over eight times a year beats a fat-margin product that sits for nine months. If the bottleneck is physical space, it is contribution per square foot.
The rule underneath all of this: divide margin by whatever you run out of first.
This is also what exposes the accounts that look impressive and quietly destroy value. The largest client on your books, the one who negotiated the deepest discount and generates the most emails, is often the one you would struggle to replace profitably. The same applies to the hero SKU with the highest order count and the highest return rate. Revenue rank and profit rank are different lists, and most owners have only ever looked at the first. Working out which part of your business is genuinely the most profitable often reorders everything.
How often to rerun the numbers
Quarterly is a reasonable default for a full pass. But the calendar matters less than the triggers. Rerun your unit economics whenever a supplier raises prices, whenever carrier rates change, whenever you add a sales channel with its own fee structure, whenever you introduce a standing discount, and whenever you change your returns or delivery policy. Any of these can move contribution margin by several points without a single line of your price list changing.
One final habit worth building. When a number comes back ugly, resist the instinct to immediately cut cost. Sometimes the answer is to raise the price, sometimes it is to change the unit by pushing average order value above the point where shipping stops dominating, sometimes it is to stop selling the item, and sometimes the right move is to accept a thin margin on something that reliably leads to profitable repeat purchases. Unit economics does not tell you what to do. It tells you which decisions you were making blind, which is usually the more valuable information. If the numbers keep coming back thin no matter how you cut them, the problem is probably further upstream, and it is worth asking whether your pricing model still works at all.
Ecommerce reached 16.6% of all U.S. retail sales by the fourth quarter of 2025. More of the economy now runs through channels where fees, postage, and returns are structural rather than incidental. The owners who hold margin through the next few years will be the ones who stopped treating those costs as operational noise and started treating them as part of the price.
