Choosing the Right Business Model

You have an idea. What you do not have is a reason to build it as one kind of business rather than another. That choice matters more than the idea. One insight — say, that small contractors lose jobs because they quote too slowly — can be run as a done-for-you quoting service, a fixed-price package, a monthly retainer, a software product, or a marketplace connecting contractors and homeowners. Five different companies, with different costs, timelines, skills and ways of failing.

Most advice here is a list of models with a paragraph each. That is not the hard part. The hard part is matching one to your capital, your time, your skills, your appetite for risk and what you want out of this.

Six models cover nearly everything: service, productized service, subscription, ecommerce, SaaS and marketplace. Blends exist, but they almost always start as one and become another. Here is each, then how to choose.

Service business

How it makes money. Skilled hours for fees. A design project, a roof, a bookkeeping engagement. Revenue is rate multiplied by hours delivered.

What it costs to start. Commonly $200 to $2,000: an entity, insurance if the work needs it, a domain, a simple site, tools. If you own the tools already, almost nothing.

Time to first revenue. Days to weeks. The only model here where you can plausibly be paid in a fortnight, because you sell the work before building anything.

How hard it is to grow. Hard, in a specific way. Growth means raising your rate or adding people, and adding people means managing delivery instead of doing it. Margins per job run 50 to 80 percent before your own labour is costed, but capacity is capped by hours. One person commonly ceilings between $150,000 and $400,000 a year, and the step up is a different job.

What kills it. Undercharging, scope creep and feast-or-famine sales. It dies slowly: too busy delivering to sell, then too broke to refuse bad work.

Who it suits. Anyone with a marketable skill, limited capital and a need for revenue soon — and anyone unsure the demand is real. Nothing validates a market faster than someone paying you to fix the problem by hand.

Productized service

How it makes money. The same delivery sold as a defined package. "Brand identity, $2,500, three weeks, two rounds of revisions." No custom quoting, no discovery call per deal.

What it costs to start. As a service, plus defining the package and building a page that sells it. Call it $500 to $3,000 with a proper site.

Time to first revenue. Two to six weeks. Slower than raw services because you must decide what the offer is before selling it, a discipline most founders skip.

How hard it is to grow. Much easier than bespoke work. Fixed scope lets you document delivery, hire against it and predict margin. You still sell hours, but the same hour repeatedly, so it gets cheaper each time. Margins commonly land 40 to 70 percent once someone else delivers.

What kills it. Packaging something the market wants customized, or pricing on the average job when a quarter of them are twice the work. One bad scope definition wipes out the margin on every sale.

Who it suits. Service operators drowning in quoting, and first-time founders with a repeatable skill who want a company rather than freelancing.

Subscription and retained service

How it makes money. A monthly fee for ongoing work or ongoing access to you. Managed IT, bookkeeping, a marketing retainer, a maintenance plan.

What it costs to start. Same as a service. The cost is the harder sale, not the money — you are asking for a decision about the next twelve months, not the next two weeks.

Time to first revenue. Four to twelve weeks. Longer cycle, and often a one-off project first to earn the right to the retainer.

How hard it is to grow. The best growth profile available without writing software. Revenue compounds — every month starts from last month, not from zero — and $10,000 a month recurring is worth several times $10,000 in project work. Growth is limited by delivery capacity and churn.

What kills it. Churn and invisible value. Clients cancel when they cannot see what they are paying for, usually three or four months in, when the work has become routine. The other killer is unbounded scope priced as though it were bounded.

Who it suits. Anyone already doing the same work repeatedly for the same clients. Rarely the right place to start cold, almost always the right place to end up.

Ecommerce brand

How it makes money. You buy or make a product for one price and sell it for a higher one. Margin is the whole game.

What it costs to start. $5,000 to $50,000 for a real attempt with inventory. Dropshipping starts under $1,000 but strips out most of the margin. Budget for product, packaging, photography, a store, and — the line people forget — ad money spent while you learn what converts.

Time to first revenue. Six to sixteen weeks. First revenue is not the milestone that matters; profitable revenue is, and that commonly takes two to three times longer.

How hard it is to grow. Scales well in units, badly in cash. Every extra sale consumes stock paid for in advance, so growth eats working capital. Gross margins of 60 to 80 percent sound generous until acquisition, shipping, returns and platform fees take their share; net margins of 5 to 15 percent are common.

What kills it. Acquisition cost rising above contribution margin, and inventory bought against optimism. A brand with no repeat purchase is renting customers from an ad platform at the platform's price.

Who it suits. Founders with capital to risk, product knowledge, and either a differentiated product or an audience. A poor fit if the plan is reselling what a hundred other stores list.

SaaS

How it makes money. Recurring fees for access to software. Build once, sell many times. The margin everyone wants — 70 to 90 percent gross — is real once you get there.

What it costs to start. $15,000 to $80,000 for a credible first version if you are paying for the build, and more if it touches payments, compliance or integrations. Writing it yourself costs six to twelve months of your life instead, which is not cheaper.

Time to first revenue. Six to twelve months is normal. Faster when the founder already has customers waiting, which is the biggest single predictor of whether a SaaS attempt survives.

How hard it is to grow. Excellent and slow at the same time. Revenue compounds, but early growth is agonising: two hundred customers at $49 a month is under $10,000, and two hundred paying customers is a serious achievement.

What kills it. Building for eleven months before speaking to a buyer. Churn — losing 5 percent of customers a month replaces your entire base every twenty months just to stand still. And competing feature-by-feature with a funded incumbent who gives away what you charge for.

Who it suits. Founders who can build it themselves, founders with deep operational knowledge of one workflow, or founders with capital and the patience to spend a year before the business is real. If you are none of those, read the section below first.

Marketplace

How it makes money. You take a cut of transactions between two sides — buyers and sellers, clients and providers — or charge listing and lead fees.

What it costs to start. $30,000 to $150,000 and up for the platform, plus a marketing budget funding two audiences at once. The build is the smaller half.

Time to first revenue. Nine to eighteen months to anything meaningful, and first revenue is a poor signal here. Twenty transactions is not yet a marketplace.

How hard it is to grow. Brutal early, extraordinary late. Once liquidity exists — enough supply that demand gets served fast, enough demand that supply keeps showing up — it is very hard to displace. Getting there is where nearly all of them die.

What kills it. The cold start, disintermediation once the two sides have met, and thin economics: 10 percent of a $60 transaction is $6, which does not cover acquiring either side.

Who it suits. Experienced founders with funding, an existing audience on one side, or a way to manufacture liquidity in a narrow geography at the start. Almost nobody else.

The comparison, side by side

Treat those ranges as the middle of the distribution, not a promise. A service needing a licence or a fleet costs more; a SaaS that is one form and a database costs less. If your plan sits far outside the range, you should be able to say exactly why.

The five inputs that actually decide it

Answer these honestly and the choice narrows to one or two options.

1. How much capital can you lose without it hurting?

Not how much you have. How much you can lose and still be fine. Under $5,000 puts you in a service, a productized service or a lean ecommerce test. At $50,000 that is not rent money, SaaS becomes reasonable. Founders routinely answer with everything they could theoretically scrape together, then run out at exactly the point the business was starting to work. Work out what it actually costs before you pick, not after.

2. How soon do you need this to pay you?

If the answer is "within three months", you are choosing a service or a productized service. Nothing else here reliably pays anyone in a quarter. With eighteen months of runway from a job or savings, the software and marketplace end opens up. This constraint eliminates more options than any other.

3. What can you personally do that is hard to hire?

If it is a delivery skill — designing, fixing, advising, building — sell exactly that as a service. If it is writing software, SaaS is on the table in a way it is not for anyone else. If it is product knowledge or sourcing, look at ecommerce. If it is sales, most of these work and you decide on the other four.

4. What happens if it fails?

A failed service costs a few months and some pride. A failed SaaS costs a year and the build budget. A failed marketplace costs both plus the acquisition spend. Match the possible loss to what you can absorb.

5. What do you actually want?

Skipped constantly, and it changes the answer completely. A business that pays you $200,000 a year and needs you present is a fine outcome — that is a service or a retained practice. A business you can sell means recurring revenue, because buyers pay multiples for revenue that survives your departure. A business that runs without you means software or ecommerce with real systems. Decide which, before you pick a model that cannot become it.

If this is true, start here

Rules, not laws. If two apply, take the more conservative one.

1. You need income within 90 days. Start with a service. Sell it before you build anything.
2. You have a marketable skill and under $5,000. Start with a productized service: fixed price, fixed scope, one page selling it.
3. You already deliver the same work repeatedly for the same clients. Move to a retainer. You have done the hard part.
4. You can write the software yourself and have six months of runway. SaaS is genuinely on the table. Sell three annual contracts before you write meaningful code.
5. You cannot write software and are not sure the problem is real. Deliver the outcome by hand for ten paying customers first — a service, and the cheapest possible research into whether the software should exist.
6. You have $15,000 of risk capital and real product knowledge. Ecommerce is reasonable. Start with a narrow range, not a catalogue.
7. You have an audience on one side of a transaction. A marketplace becomes plausible, because you can seed the hard side. Without that, choose something else.
8. None of the above, and a strong idea. Validate it before choosing a model at all. Validation usually tells you which model fits.

The case for starting with a service

Here is the unwelcome part. Most first-time founders should start with a service or a productized service and grow into recurring revenue rather than starting with software.

Not because software is bad. Because the sequence matters. A service forces you to sell before you build. In week two you learn whether anyone will pay, what they will pay, what they want and what words they use for the problem — exactly what a SaaS founder normally buys with eleven months and a build budget, and buys badly, inferring it from behaviour instead of hearing it from someone handing over money.

Service revenue is also the cheapest capital there is. It does not dilute, does not have to be repaid, and arrives while you are learning. Plenty of durable software was paid for by a founder doing that same work by hand first, at which point the product was just the automation of a process already known to be valuable.

And it caps the downside. If the service does not sell, you have lost weeks and still have the money.

When this advice does not apply:

• You can build it yourself, quickly. If version one is weeks of your own work rather than a five-figure invoice, the maths changes. Build it — but still sell it before it exists.
• You already have the customers. Years inside an industry and twenty businesses that take your call means you already hold the discovery a service would have bought you.
• The product only works at scale. If manual delivery is genuinely impossible rather than merely tedious, the service stage does not exist for you.
• The window is narrow and someone else is moving. Real timing pressure is rarer than founders think. Be honest about whether the race exists.
• The service would build the wrong business. If hand delivery fills your year with clients who make the product impossible, cap it — a fixed number of clients, a fixed end date.

The failure mode to avoid is starting a service, enjoying the money and never building the thing. Set the trigger in advance: at a stated monthly revenue figure, or by a stated date, the product starts. Write it down.

Marketplaces are the hardest first business

Worth stating plainly, because a marketplace is the most attractive idea a founder can have and the most likely to end badly.

Every other model here has one customer acquisition problem. A marketplace has two, locked together: neither side shows up before the other is there. There is no clever way around that cold start — every fix that works is expensive or slow. Subsidize one side, own the supply yourself at first, or pick a geography so small that thirty sellers counts as liquidity, then repeat it city by city. Meanwhile you capture perhaps 10 to 20 percent of each transaction, so you need enormous volume before the revenue supports anyone, and the moment the two sides like each other they have every reason to leave and keep your fee.

Marketplaces are not bad. When they work they are among the most valuable businesses there are, precisely because they were hard to start. They are just the wrong place to learn, and a genuinely good one will still be a good idea in two years.

How these models grow into each other

The most common durable path runs in one direction: service → productized service → subscription → software.

You do the work by hand for individual clients. The same job keeps repeating, so you package it at a fixed price. Clients want it continuously, so you put them on a monthly plan. You are running the same three steps every month for everyone, so you automate them — and now you have a product, built with real knowledge of what it must do and customers to sell it to on day one.

Each step raises what the business is worth: project work sells for a small multiple of profit, if it sells at all, recurring contracts with documented delivery are worth substantially more, and software with retained customers more again.

Two warnings. Do not skip steps to reach the multiple faster — the value came from the operational knowledge each stage produced, not the label. And change one thing at a time: a product for your existing customers is a model change, but a product for different customers is a new company, and doing both at once usually costs you the revenue funding the move. Winning first customers is its own job.

If you want that pressure-tested against your own numbers, it is what the blueprint stage exists for — see how it works.

The short version

• The model is a bigger decision than the idea. One insight builds five different companies.
• Choose on five inputs: capital you can afford to lose, how soon you need income, what you can personally do, what failure costs, and what you want the business to become.
• Costs and timelines vary enormously — days and a few hundred dollars for a service, a year and five figures for SaaS or a marketplace.
• Most first-time founders should start with a service and evolve toward recurring revenue. The exceptions: you can build it yourself, you already have the customers, or it cannot be delivered by hand.
• Marketplaces are the hardest first business — two acquisition problems that block each other, thin economics, and constant pressure to be cut out.
• Whatever you pick, sell it before you build it. That applies to all six.