Why unlimited revisions is a warning sign, not a benefit
Two quotes are open on your desk. One offers three rounds of revisions. The other offers unlimited revisions. The second one looks obviously better, and it is the one more likely to end with a half-finished website and a relationship nobody wants to be in by month five.
That is not an argument against unlimited revisions. We offer them. It is an argument against unlimited revisions sold as a headline benefit with nothing attached to it, because on its own the phrase does not describe generosity. It describes the absence of a finish line.
Everything in a project that runs well happens because someone decided when a thing was done. Remove that and you have not removed a restriction, you have removed the mechanism that gets you a website.
What unlimited revisions is actually pricing
Nobody gives away unbounded labour. So when an agency offers it, one of three things is true.
The first is that the risk has been priced in. Somebody sat down, estimated how badly this could go, and added a margin big enough to survive the worst version. That margin is now in your quote whether or not you ever use a single extra round. Clients who send tight, consolidated feedback are subsidising the ones who do not, and nobody in the conversation says so.
The second is that the risk has not been priced at all, which is more common with smaller studios and freelancers, and it holds until the hours run out. After that the work does not stop, it degrades. Replies get slower. Changes get made carelessly because making them carefully is no longer economic. If you have ever had a project go quiet at eighty percent complete, this is usually what happened. Nobody announces it.
The third is that unlimited rounds are bounded by something other than a count, and the studio has thought hard about what that something is. That is the only version worth buying, and it is the only version anyone should be selling.
Why does an agency offer unlimited revisions at all?
Because “three rounds” is a genuinely bad thing to say to a buyer, and most agencies know it.
A round count turns every piece of feedback into a transaction. Is this worth spending a round on? Should we hold this note back in case something worse comes up? Clients start hoarding comments, batching unrelated observations into one enormous email so it counts as a single round, and swallowing real objections because they are rationing. Every one of those behaviours produces a worse website.
The instinct behind unlimited is therefore correct. Feedback should not be metered. A client who spots something on the fourth look should say it, and a studio should want to hear it, because a small correction on day forty is cheaper for everyone than a rebuild in year two.
The failure is not the removal of the count. It is that nothing replaces it.
The thing that has to replace the count
A phase structure, and an approval that means something.
Our projects run in four phases: discussion, design and build, feedback and amendments, then handover. Web builds add a support window after that. This is not unusual and most competent studios do something similar. What matters is what an approval does at the end of each phase.
An approved phase is a foundation. Page templates are built on an approved direction. Content is written against an approved structure. The build is made from approved designs. When you approve something, you are not being polite, you are releasing the next phase to start. That is why reopening it later is a different act from refining something inside it.
The Association for Project Management describes change control as the process through which requests to change an approved baseline are captured, evaluated and then approved, rejected or deferred. Web projects rarely use that language and mostly should. The point is not bureaucracy. The point is that the approved baseline exists at all, so that everyone can see what a change is changing.
So the honest version of the offer reads like this. Unlimited rounds, because counting them makes the work worse. Feedback treated as collaboration rather than as a transaction. And a caveat stated at the start rather than discovered in month three: revisions to a phase that has already been approved will extend the timeline accordingly, because the work downstream has to be redone.
That last sentence is the whole difference between a policy and a slogan.
Consolidated feedback is worth more than another round
The other half of the arrangement is on the client side, and it is not a big ask.
Send feedback per phase, in one consolidated set, from one person who holds the pen internally. Not six emails over nine days. Not a Monday note that contradicts a Thursday note. Not one director’s view on Tuesday and a different director’s view the following week, because the second director is going to see it eventually and the only question is whether that happens before or after the work is built.
Incremental feedback is expensive in a way that is invisible from outside. Each round has a fixed cost that has nothing to do with the size of the change: reopening the file, reloading the context, making the edit, exporting, writing the reply, waiting. Ten notes in one message cost a fraction of ten notes in ten messages, and the consolidated version produces a better result because the notes get resolved against each other rather than in sequence. Two comments that quietly contradict each other are a five-minute conversation when they arrive together and a fortnight of drift when they arrive apart.
This is the trade. We do not count your rounds. You do not make us discover your position one email at a time.
What a project with no end looks like afterwards
We do not get to see the contracts behind other people’s websites. We do see the result.
Across 54 Singapore B2B sites we reviewed in August 2026, 19 of the 23 whose footers we examined carried an out of date copyright year, with stale years going back well over a decade. Three still carried live template placeholder content that had never been replaced.
Placeholder text on a live commercial website is not a revision problem, and it would be dishonest to claim we can trace it to one. But it is a picture of the same ending. Somebody stopped. The site did not reach a handover, it reached the point where continuing was no longer worth it to whoever was doing the work, and it has been sitting there ever since, quietly telling every visitor that nobody is home.
A project with a defined finish gets finished. A project without one gets abandoned at whatever stage it happened to be in when the energy ran out, and the client is left owning it.
How to read any agency’s revision policy
You do not need to take our structure. You need to check that the agency in front of you has one. Five questions do it, and you can ask all five in an email.
What ends a phase? There should be a specific answer. A written approval, a signed document, a reply to a named email. If approval is a vibe, the phase cannot close, and if the phase cannot close there is no revision policy, only an open tab.
What happens to the launch date if we reopen an approved phase? The correct answer is that it moves, and a good studio will say so cheerfully rather than defensively. Anyone who tells you an approved direction can be replaced in week six at no cost to the date is either not planning to do the work properly or is planning to bill you for it later.
Who consolidates our feedback, and in what form? If the agency has no opinion about this, they have never run a project with more than one stakeholder, or they are absorbing the cost silently, which brings us back to the price.
Where is the line between a revision and a new request? Ask for examples in both directions. A different headline is a revision. A fifth page that was never in the sitemap is a new request. Both are perfectly reasonable things to want. Only one of them should leave the fee and the timeline untouched, and you want to know which is which before you are attached to the idea.
What happens after launch, and for how long? Every web project we deliver includes 30 days of post launch support covering bug fixes, minor content edits, image swaps and questions. Larger or more complex builds extend that to 45 days. New features, new pages and redesigns sit outside that window, which we say up front because the alternative is saying it in an awkward email six weeks later.
An agency that answers those five clearly is safe to work with whether they count rounds or not. An agency whose entire answer is the word unlimited has told you nothing, and the gap will be filled at some point by whoever has less patience.
Where this shows up in the price
Structure is also what keeps the number honest.
Our web projects start from S$4,000 and what moves the figure is scope and how much of the content already exists in usable form, not a guess about how difficult you might turn out to be. We can quote that way because the phases carry the risk instead of the margin. For Hong Hock Global, a water infrastructure contractor whose work is awarded almost entirely on prequalification, the slow part was never revisions, it was assembling the certification and personnel evidence that had to be on the site at all. Phases exist so that the slow part is visible and scheduled rather than absorbed.
Our pricing bands are published rather than quoted on request, and we have written separately about what a B2B website actually costs in Singapore and what sits behind each band.
If you are comparing quotes right now, the useful move is not to pick the one with the most rounds. It is to ask each of them what ends a phase, and see which one has an answer ready.