Building
What actually drives the cost of a project
Quotes for the same brief routinely differ by a factor of five. The variance is almost never about the number of pages, and the drivers are knowable in advance.
6 min read
Ask five studios to quote the same project and the numbers will not resemble each other. That is not evidence that four of them are wrong. It usually means five different projects were understood from the same brief.
Cost is not driven by the things clients expect. Page count barely matters. Visual complexity matters less than it looks. What follows is what actually moves the number.
The real drivers
- How many states the product has
- A page that only displays is cheap. A page that behaves differently for a logged-in user, an empty account, a failed payment, and an admin is five pieces of work wearing one name. States, not screens, are the unit of cost.
- Whether anything has to talk to something else
- Payments, availability, inventory, an existing system nobody documented. Integration cost is dominated by the other side's quality, which is why honest quotes for it come with conditions attached.
- Who is allowed to do what
- One kind of user is simple. Owners, staff, and customers with different permissions is a different project, and it is usually mentioned late because it does not feel like a feature.
- How much content actually exists
- A build waiting on text and images is a build being paid for while it waits. Migrating years of existing content is frequently a larger job than the build itself.
- How many languages
- Not double the work, but not a translation either — separate content, separate search structure, and a maintenance burden that continues after launch.
- How certain the requirements are
- The most expensive projects are the ones that changed direction three times. Uncertainty is a real cost and pretending otherwise is how budgets get exceeded.
Nobody has ever gone over budget because a site had too many pages. They go over because the requirements moved.
What barely matters
- The number of pages, once the patterns are designed. The tenth page of a type costs almost nothing.
- How elaborate the visual design is. Considered restraint is not cheaper to design, but it is far cheaper to build and maintain than decoration.
- The technology, within reason. Sensible choices differ from each other far less than a well-scoped project differs from a vague one.
How to get a quote you can rely on
- 01
Describe the outcome, not the solution
"Guests should be able to book without calling us" is scopeable. "We need a booking system" invites six different assumptions about what that means.
- 02
Say what already exists
Current site, systems in use, who maintains them, what has to keep working. Discovering this mid-build is the most common source of change requests.
- 03
Name what you are not doing
An explicit exclusion is worth more than a paragraph of requirements. It is also the fastest way to spot that two parties understood the brief differently.
- 04
Get the number in writing before design starts
A quote that arrives after work has begun is not a quote. If the scope changes later, that should be a visible decision with a price attached, not a surprise at the end.
Describe what you are trying to do and you get a real number, not a range.
Start a conversationFollow-up
Questions this raises.
- Why do quotes for the same brief differ so much?
- Usually because five different projects were understood from one brief. Studios make different assumptions about states, integrations, permissions, and who supplies the content — and those assumptions, not the page count, are what the number is built on.
- Fixed price or hourly?
- Fixed price where the scope is genuinely settled, because it puts the risk of estimating on the studio rather than the client. Hourly is honest where the work is exploratory and nobody can yet say what it involves. The failure mode is a fixed price on an unsettled scope, which becomes a change-request argument.
- What happens if the scope changes mid-project?
- It should be a visible decision with a price attached, not a surprise at the end. Changes are normal; what causes damage is absorbing them silently until the budget runs out and the conversation happens too late to have options.
- Is a much cheaper quote a warning sign?
- Not automatically, but it is worth understanding. A far lower number usually means a smaller project was understood, or that discovery, testing, and content have been left out of the scope and will arrive later as extras.