Website revision rounds and how to use them well

Owners hear "two rounds of revisions" and read it as a limit on how much they are allowed to care. It is really a limit on how many times the whole team stops, reads, and reworks.

The comparison that makes it obvious: this is a punch list on a build out. The contractor does not want you calling every afternoon with one more item. They want one walkthrough, one list, one pass.

Why are rounds limited at all?

Because each round has a fixed cost regardless of its size.

A round means gathering feedback, interpreting it, deciding what conflicts with what, making the changes, reviewing them internally, and presenting again. That overhead is the same whether your list has four items or forty, which is why forty items in one list is cheap and four items across ten emails is expensive.

Unlimited revisions sound generous and produce worse work. Without a defined end, nobody commits, decisions get reopened, and the site drifts toward whatever the last person mentioned. Projects with no boundary do not finish, they get abandoned.

This is also the mechanism behind quotes that look different for the same job. A fixed price with defined rounds and a quote with open-ended revisions are pricing different risks, which is part of what a small business website actually costs and why comparing two numbers side by side rarely tells you much.

What does a scattered round actually cost?

How do I give feedback a designer can act on?

Be specific, give the reason, and do not prescribe the solution unless you are sure.

Weak: make the header pop. Strong: the phone number is hard to find on a phone, customers call more than they submit forms, can we make it the most visible thing up top.

The second version gives the designer a problem to solve. The first gives them a guess to make, and if they guess wrong you have burned a round.

Three more rules that shorten every project.

One list. Numbered, in one document or one email. Not comments spread across texts, a call, and two forwarded screenshots.

One decision maker. If your spouse, your manager, and your brother in law all send feedback, somebody has to reconcile it before it reaches the designer. That somebody is you, not them.

Separate accuracy from preference. "We do not offer that service anymore" is a correction and it is not optional. "I would prefer a different blue" is a preference. Mark which is which, and the important items stop getting lost among the small ones.

Feedback that names the problem gets solved. Feedback that names a fix gets implemented, and then reopened.

What about the button color argument?

It comes up in every project and there is a specific piece of evidence worth having ready, because somebody will cite the opposite.

The claim in circulation is that red call to action buttons beat green, and it traces to a single published case study by Joshua Porter. Kohavi, Deng, Longbotham and Xu name that exact result in their peer reviewed KDD 2014 paper, generalising from thousands of controlled experiments at Microsoft, Amazon, Booking.com and LinkedIn, as their example of a finding that does not travel. Their words: "Since we do not see a lot of sites with red call-to-action buttons, we believe this is not a general result that replicates well." The same rule warns that published case studies vary wildly in quality, including tests published with a p-value above 0.05, which is normally considered not significant at all.

So nobody knows what color your button should be, and at your traffic level nobody is going to find out. Which is liberating rather than depressing. Pick a color that is obviously a button, that contrasts with everything around it, and stop spending rounds on it.

There is one place where color stops being preference. The World Wide Web Consortium sets a contrast ratio of 4.5 to 1 for normal text and 3 to 1 for large text at level AA under success criterion 1.4.3, and WCAG 2.2 adds a minimum target size of 24 by 24 CSS pixels at AA under 2.5.8. A request to lighten body text to a pale grey, or to shrink a tap target to fit a layout, is not taste. It fails a published standard, and it is one of the few pieces of feedback a designer should push back on. Ask them to.

What counts as a revision and what is a new request?

A revision changes something that already exists. A new request adds something that did not.

Changing the headline is a revision. Changing the color of a button is a revision. Adding three service pages that were not in the scope is a new request, and it should be priced as one. That is not a vendor being difficult, it is the only way a fixed price can exist at all.

The grey area is usually the logo. Owners often decide mid-project that the mark is not working, and logo rework is a separate piece of work with its own rounds. It is also worth being honest about whether the logo is genuinely the problem, because what actually matters about a logo for a local business is a short list and most mid-project logo doubts are really doubts about something else.

The other grey area is length. Owners frequently ask for pages to be cut down because they look long in a mockup. Long pages are not automatically bad, and for some offers they perform better, so why long pages sometimes convert better than short ones is worth reading before you spend a round deleting content that was doing work.

What should I actually be reviewing?

Accuracy first, then clarity, then appearance. In that order, every time.

Accuracy. Services you still offer, cities you still serve, hours, prices if you publish them, staff names, licenses. Errors here are the ones that cost you customers. Two of them also carry a ranking cost worth knowing about. Whitespark's Local Search Ranking Factors report, published 6 November 2025 with 47 local search experts scoring 187 factors, puts business open at time of search fifth overall with 189 points and HTML name, address and phone details matching your Google Business Profile fifteenth with 153. That is expert opinion rather than a controlled test, and Whitespark says so, but wrong hours and a mismatched phone number are worth catching in a review round rather than in six months. Accuracy errors also reveal a site quietly describing an older version of the company, which is how a website ends up built for a business you no longer run.

Clarity. Would a stranger know what you sell, whether you serve them, and what to do next? Ask somebody outside the business to look, because you cannot see your own site fresh anymore.

Appearance. Last, and least. This is where most feedback goes and where the least revenue lives.

A good discipline is to review on a phone before a laptop, because that is where the traffic is. Another is to leave the mockup alone for a day before writing your list, since first-impression feedback is mostly nerves.

While the project is open, spend some of that reviewing energy outside the site. The questions customers ask repeatedly belong in your copy and also on your listing, and seeding your Google Business Profile questions properly is a small job that pairs naturally with writing the site content.

Where does this break down?

Two places.

When the brief was wrong. If round one comes back and your reaction is that the whole direction is off, that is not a revision problem. Stop, and have the conversation about what the site is for. Spending a round redirecting a design that was aimed at the wrong customer wastes the round and produces a compromise.

When the vendor uses rounds as a wall. A round limit is a scheduling tool, not a way to refuse corrections. Fixing something that was built wrong, or that contradicts what you supplied in writing, is not a revision. Say so plainly and expect it fixed.

What to do this week

Before you send feedback, sit with the design for twenty-four hours. Then open it on a phone.

Write one numbered list. For each item, name the problem and why it matters, and mark it as accuracy, clarity, or preference.

Put the accuracy items at the top. Those get fixed regardless.

Run the composition test from step five. If preference is most of your list, cut before you send.

Then send it once, from one person, and let them work. Interrupting a round in progress with three more emails is the most reliable way to make a project take an extra month.

Questions about feedback you are about to send?

Email me at eric@seod.com with the feedback list you have drafted for your designer, exactly as written. I will rewrite it into something they can act on, sorted into accuracy, clarity, and preference, and I will flag anything that is really a scope change so you are not surprised by an invoice. Send it before you send theirs.

More on managing a project in the web design and branding library.