The question as it arrives, and the question underneath it
This reaches us as a technical question, but it is a commercial one: what job is this asset doing, and who pays for that job getting done? A website, a web application, a mobile app and a subscription platform are not rungs on a ladder that a maturing business eventually climbs. They are four answers to four different questions.
A website publishes and persuades. It explains the offer, builds credibility, and moves a visitor toward contact or purchase. Its success measure is qualified enquiries, not screen count.
A web application operates. The user has an account and saved state they return to. The first screen is a work queue or a record list, not a headline.
A mobile app is a distribution channel more than a distinct category of software. The business logic is usually the same. What changes is that it is installed on a device and governed by two app stores.
A SaaS platform is a business model before it is an architecture: one product serving many customers, with tenant isolation, self-serve signup and recurring billing.
Confusing these four is the most expensive mistake we see. One company asks for an app when it means a customer portal. Another asks for a platform when what it needs is a landing page and an enquiry form that actually works.
Start with the job: publish, transact, or operate
Before any technology discussion, write one sentence describing what a user does in a single visit. That sentence decides most of the question.
If it is read, compare, then get in touch, the job is publishing. A website is the answer, and anything larger delays launch without buying you anything.
If it is complete a transaction that ends when it ends — buy, book, settle an invoice, submit an application — the job is transacting. That sits on a website with a tightly designed transaction path, and it rarely justifies an installed app.
If it is come back regularly to manage an ongoing state that belongs to me — live orders, stock, a team, subscriptions — the job is operating, and that is where a web application genuinely begins.
Often the honest answer is two surfaces, not one: a publishing surface for people who do not know you yet, and an operating surface for people who are already customers. The common mistake is forcing them into a single product because they share a brand name. Treat them as two products with one visual identity and a clear handover path between them, and you avoid years of design and architecture compromises.
One supporting test settles the remaining doubt: is there user-specific state that must persist between two visits? If there is not, do not build accounts, roles and permissions. You will pay for that machinery every month after launch while nobody uses it.
Five questions that settle the shape
Answer these five honestly and you will not need a technical session to decide what to build.
Who is the user, and how often do they return? Someone who visits twice before buying will never install anything. Someone who opens the tool five times a day earns an experience built specifically for them.
Is there persistent state, and are there roles? If different users must see different slices of the same data depending on their role, you are in application territory regardless of how simple the screens look.
Do you genuinely need device capabilities? Not nice to have, but cannot function without: sustained barcode scanning, background location, Bluetooth peripherals, full offline work in the field. If that list is empty in your case, the web is sufficient.
Who pays, and for what? If an external customer is buying recurring access, you are building a product with everything that entails. If the beneficiary is an internal department, the measure is time recovered, not revenue.
Who runs it after launch? This question gets skipped every time, and it predicts whether the project is still alive in a year. A tool with no owner named in person becomes technical debt within months.
When the answers conflict — a daily user with no real need for device capabilities, or a paying customer with no second one in sight — let the last question break the tie. Choose the shape your current team can actually operate after launch, not the most ambitious one.
When a mobile app is genuinely required
We build mobile apps when there is a reason, and we say so plainly when there is not.
The legitimate reasons are short. Real dependence on device capabilities. Field use that needs offline work and later sync. Notifications as an operational channel rather than a marketing one. Or the app itself being the product the customer buys.
The most common illegitimate reason is wanting presence in the store. Apple publishes a minimum functionality requirement in its App Store Review Guidelines, under which an app offering no more than a repackaged website is a rejection risk. Store presence is not a fundable objective on its own.
The cost is not the initial build either. A mobile app means two platforms with two release cycles, a review step before every update, a shipped version you cannot pull back from users devices the way you revert a web deploy, and a tail of users on old releases you still have to support. Before all of that, you have to persuade someone to install it, which is the hardest step in the funnel.
Before jumping, look at the middle option: a mobile-first web experience, optionally upgraded to an installable progressive web app with offline support. Its limits are real too. Notification support and access to certain device features vary by platform and version, and both should be verified on your audience devices before you build a decision on them.
SaaS is a business model before it is an architecture
The screens a user sees in an internal tool and in a SaaS product look deceptively alike. The difference is entirely in what surrounds them.
A product sold by subscription needs tenant isolation that does not depend on developer discipline, self-serve signup that works without you, plans and usage limits, recurring billing with a considered response to failed payments, roles inside each customer account, a status page and support, and a release policy that does not break one customer while serving another.
Before calling your project a platform, answer one question: is there a second customer who would pay for the same product without material changes? If every customer needs a different version, you are selling custom development wrapped in a subscription. That is a different business with different costs, margins and pricing logic.
If the answer is no, building an internal tool is a perfectly sound decision. Just do not price it or plan it as a product.
One necessary note: any platform handling payments, personal data or recurring contractual commitments carries regulatory and accounting obligations. Those belong with a qualified adviser before launch, not after.
Two-year cost of ownership — the line nobody estimates
Most budgets that reach us price the build and ignore what follows, then get surprised in year two.
A website is the lightest: content updates, security patching, performance work and search visibility. Marketing usually runs it without a permanent developer.
A web application adds a layer: separate staging and production, backups that have actually been restored rather than merely configured, monitoring and alerting, internal user support, and continuous change because the underlying process keeps changing. It needs a product owner, even part-time.
A mobile app adds two platforms, store review, near-mandatory work around each annual operating system release, and compatibility with older devices.
A SaaS platform adds billing operations, customer onboarding, an availability commitment to someone who is paying, and support with stated response times.
Add one line that is routinely overlooked in the Saudi market: running two languages. Every new screen, system message and notification needs Arabic and English written for each language rather than translated between them. That is a standing editorial commitment after launch, not a task that ends with it.
The rule we repeat in every estimation session: allocate a recurring annual operating cost from day one, and name who pays it and who owns the decisions about it. A project with no operating line in its budget is not fully funded.
Sequencing: what to build first without locking yourself in
The right answer is rarely one of them. It is usually this first, that if.
Start with the smallest thing that proves demand. A clear website with a well-designed enquiry path tells you within weeks whether anyone wants what you are offering, before you spend a full platform budget on an assumption.
Separate business logic from the interface from the first line of code. When the rules live behind an independent API, adding a mobile app later is another client, not a rebuild. That single architectural decision is the difference between an upgrade and a second full project.
Do not improvise authentication and permissions. Changing a role model after real data has arrived is far harder than getting it roughly right early.
Design the data model with a second customer in mind, even if none exists today, without complicating version one for a distant possibility. A tenant key on core tables is much cheaper than a migration later.
Finally, the signals that you picked the wrong shape: a brochure site that has quietly accumulated account and permission behaviour; a mobile app your audience opens once every two months; a platform that has had one customer for a year. Each of those is a prompt to revisit the decision, not to add features.
Key takeaways
- Write one sentence describing what a user does in a single visit — publish, transact or operate answers most of the question.
- Do not build accounts and permissions unless there is user-specific state that must survive between two visits.
- Tie any mobile app decision to a named reason: a device capability, offline field work, or operational notifications. Store presence is not a reason.
- Before calling it a platform, confirm there is a second customer who would pay for the same product without material changes.
- Put a recurring annual operating line in the budget with a named owner, and keep business logic behind an independent API.
Frequently asked questions
Our budget is limited. Website or app first?
Start with whatever proves demand for the least spend. Usually that is a clear website with a well-designed enquiry path and measurement that shows where enquiries come from. If the app exists to run a process your team is struggling with today, the priority flips: the app buys back measurable hours, and the website can stay simple until the process settles.
Is a progressive web app a real substitute for a native mobile app?
It is sufficient in many cases, particularly when the need is a fast installable experience with some offline capability. It is not sufficient when you depend on specific device features or on reliable notifications as an operational channel, since support for those varies by platform and version. The practical approach is to list the capabilities you actually require, then verify them on your audience devices before committing.
When should an internal tool become a SaaS product?
When a second customer is willing to pay for the same product without material changes, and when the tool is usable without you training anyone. The conversion is not a change of screens — it adds a whole layer: tenant isolation, self-serve signup, billing, support and a release policy. Estimate it as a separate project, not an update.
Can we start with a website and grow into a platform without rebuilding?
Yes, on two conditions taken early: business logic lives behind an API independent of the interface, and the data model can accommodate more than one customer. Without those two, the move is still possible but in practice it is a rebuild with a new budget.
Sources
- App Store Review Guidelines — Apple · https://developer.apple.com/app-store/review/guidelines/
- Progressive web apps — MDN Web Docs · https://developer.mozilla.org/en-US/docs/Web/Progressive_web_apps