To build a SaaS product without wasting money, find customers who will pay before you build, then build the smallest version that does one job well for them, and add the supporting parts every subscription product needs: accounts, billing, security and support. SaaS stands for software as a service: software customers use online and pay for on a recurring basis.
Step one: sell it before you finish it
The most expensive mistake is a polished product nobody wanted. The cheapest protection is to find out first.
- Talk to ten people who have the problem. Not friends. Ask how they handle it today, what it costs them, what they have already tried and why they stopped.
- Look at what they use now. A spreadsheet, a paper form and a stack of emails all show a real need. A shrug shows a weak one.
- Ask for a commitment. A signed letter of intent, a deposit, or a place on a paid pilot is worth far more than “that sounds great”. Where you can, take a payment, however small.
- Charge from the start. A free pilot tells you people will accept free. Even a modest paid pilot tells you whether they value it.
If nobody will commit, you have saved the whole build. If several will, you have your first customers, a list of real requirements and the beginnings of the price.
Step two: run it by hand first
Before writing code, deliver the result manually. If your idea is a tool that checks tenancy paperwork, check ten customers’ paperwork yourself with a spreadsheet and a form. You learn what customers actually send, which steps are slow, and which parts really need software. This is sometimes called a concierge version, and it turns assumptions into facts.
Step three: how to build a SaaS first version
The first version should do one job for one type of customer from start to finish. Everything else waits.
To decide the scope, write the customer’s journey in one line and cut everything that is not on it. Take the example of a job-costing tool for small landscaping firms (invented for this article). Version one might do only this: enter a job, add materials and hours, see the margin, send a quote. Reports across jobs, staff scheduling, equipment tracking and a mobile app for crews all go on the later list.
Write the scope down in a software requirements document so you and any developer quote from the same page. For how long a first version takes and what drives its price, see our guide to MVP development cost and timeline.
The parts of a SaaS nobody lists
The feature is only part of what you are building. A subscription product also needs the following, and each takes real effort.
| Part | What it involves |
|---|---|
| Accounts and logins | Sign-up, password reset, user roles, inviting team members, closing an account |
| Separate customer data | Every customer must see only their own data. This must be designed in from the start, because it is hard to add later |
| Billing | Plans, trials, upgrades, failed payments, invoices, tax rules that vary by country. A payment provider handles the money, but you still design the rules |
| Email and notifications | Welcome messages, receipts, alerts, and the deliverability work that gets them into inboxes |
| Admin area | A way for you to see customers, help them, and fix problems without editing the database |
| Security and backups | Encryption, access controls, backups you have tested restoring, and a plan for an incident |
| Terms and privacy | Terms of service and a privacy policy that match the countries you sell in. Have a lawyer review them |
| Support | A way for customers to reach you and for you to track what they ask |
These parts are the reason a “simple” SaaS costs more than the feature alone suggests. They are also why using standard services for billing and sign-in, instead of building them, is often the right call. Our payments service covers connecting the billing side.
What moves SaaS development cost
Instead of quoting a price, here is what changes it.
- The number of user types and the rules about who can see and do what.
- The number of other systems it connects to, and how well they are documented.
- How complex the billing is: one plan, several plans, usage-based charges, discounts, multiple currencies.
- How much data it stores and processes, and how fast it must respond.
- Whether you need web only, or also apps for phones.
- Compliance needs, such as health, financial or children’s data.
- How much of the design is original, and how much comes from standard components.
For the general principles on pricing custom work, see our guide to what custom software costs.
No-code first, custom later
No-code tools can get a first version in front of customers sooner and cheaper than writing code. They have limits: the per-user cost can rise as you grow, you depend on the vendor, and some things cannot be done at all. A common route is to prove demand with no-code, then rebuild the core in custom code once customers are paying and the limits are known. Our guide to Bubble versus custom code goes through the decision.
Worked example: pricing a job-costing tool
Continue the landscaping example. The figures here are invented for illustration, so replace them with your own. Suppose you interview twelve firms and find that most spend an evening each week working out which jobs made money. A tool that does it automatically saves them that evening.
Now test the price. Offer five of those firms a paid pilot at a monthly fee that is clearly less than the value of the hours saved. If three accept, you have evidence of both demand and price. If none do, ask why: the problem may be smaller than it sounded, the price too high, or the buyer someone else in the firm.
Next, count what it takes to serve them: hosting, the payment provider’s fees, support time and the developer’s maintenance. Remember to count the running costs that follow launch. If the monthly fee does not comfortably cover them, the product needs a higher price or a narrower scope before you build.
Who should do the building
- You, if you can code. Cheapest in cash, but your time is the cost, and you must also run sales and support.
- A freelancer or a studio. Faster to start. Choose by how they ask questions, what they show you each fortnight, and whether you will own the code.
- A technical co-founder. Aligned incentives, but you are sharing the company, so choose with care.
- An AI app builder. Useful for prototypes you show customers. Before relying on one for a paid product, check who maintains the result and how it handles security and data separation.
Price and sell in stages
- Before building: sell a pilot to a handful of customers at a founding price.
- During the build: show working software every couple of weeks to those customers and change course based on what they say.
- At launch: convert the pilots to paid plans, then widen to similar customers.
- After launch: review what customers actually use. Cut features nobody touches and raise the price if the value is clear.
Keep a running list of what customers ask for, with who asked and how often. After a dozen conversations, the requests that repeat are your next release, and the ones that come up once are noise. This list is also the strongest evidence you can show a developer when you ask them to price the second version.
Signs to continue, change or stop
Continue when customers use it weekly, pay on time and ask for more of the same. Change direction when they use a part you thought was minor, or when they ask for a different product. Stop when pilots lapse, nobody will pay and the feedback is polite rather than urgent. Stopping early is a result, not a failure: it costs far less than finishing.
CentriWeb builds custom software, including web apps and portals, and looks after what it builds. If you have a first version scoped and want it priced, a 30 minute call is the place to start.