Website redesign or optimisation — how to settle it before you spend

Two different projects, two different bills, and one decision usually taken on instinct — most often after somebody in the firm says the site "looks dated".

A redesign changes the foundations — architecture, engine, the way content reaches the page, often the visual identity too. Optimisation squeezes the most out of what already stands. Confusing them costs the difference between a four-figure and a five-figure bill, and worse than overpaying is the common outcome: a rebuild where a fix would have done, losing along the way the positions the site spent years earning.

Below are the four questions that settle it, a description of both routes, and one risk the quotes stay quiet about.

Four questions, in this order

The order is not accidental — each question makes sense only if the previous one came out well. The first "no" ends the diagnosis and points to a redesign; only four answers of "the site is fine" lead to optimisation.

1. Does the site work properly on a phone? Not in a desktop preview — on your own phone, in ordinary conditions: on a tram, on a weak signal, one-handed. If the menu does not open, the form cannot be filled in or the text needs pinching to read, this is not a matter of fixes. You do not add responsiveness to a site that has none — that is rebuilding the layer everything else stands on.

2. Can your team add a page without a developer? An ordinary page with text and a photograph, in an hour, without emailing the contractor and waiting three days. If not, every future change costs money and waits. A site where nothing can be changed stops being updated — and ages faster than its appearance, because it describes an offer that no longer exists.

3. Does adding a new feature require a workaround? A payment gateway, an integration with a stock system, a second language, a client area. If the answer is "possible, but it has to be worked around" or "cheaper to build from scratch", that is technical debt — the most overlooked reason for a redesign, because the site can look good and have decent traffic while it is true. The cost appears in no invoice; it shows up as what the firm does not do, because "our site cannot".

4. Is there search traffic but no enquiries? If people arrive and leave nothing, that is work for optimisation — the cheapest on this list, because you have the material. If there is no traffic at all, neither route is right: the problem is visibility and that is where to start. The best-designed site with no visitors converts nobody.

Where to check the answer to each

The questions are simple, but the answers tend to be given from memory — and memory is often wrong here, because the owner sees the site daily and has stopped noticing what does not work.

Question 1 takes your own phone and five minutes: home page, one service page, the contact form. Can an enquiry be sent without zooming, does the menu open, do images push content off screen? If somebody in the firm has an older phone, ask them — the most problems show on hardware nobody on the team uses.

Question 2: ask the person meant to do it day to day to add an empty page with a paragraph and a picture. Do not ask whether it "can be done" — ask them to do it in front of you. The difference between "in theory yes" and "in practice something breaks every time" surfaces in minutes.

Question 3: recall the last three requests that went to the contractor and how long they took. If the answer to each was "possible, but it would be a bigger rebuild", you already have your answer — it just arrived in instalments.

Question 4 needs Search Console, the performance tab, three months. Two numbers: impressions and clicks. Thousands of impressions and a handful of clicks mean the titles and descriptions are the problem, not the site. No impressions means the site is not in the game.

Symptoms that genuinely point to a redesign

Beyond the four questions are situations where the decision is obvious, though it often gets deferred for years:

• Unsupported technology. A system no longer receiving security updates is a risk, not a saving. Nothing here to optimise — every month of delay raises the chance somebody else takes the decision for you, after a break-in.
• A change in what the firm is. New name, new customer group, new sales model. A site describing the previous firm cannot be edited into one describing the current one — the problem is not the words but a structure built around different questions.
• Architecture built around an offer that no longer exists. The menu reflects what the firm did five years ago, and new services get bolted on as pages with no place in the navigation. A workaround works while there is one; by the third the site stops making sense.
• Dependence on a single contractor. If nobody but the author can get into the site — no documentation, no access, built on something unusual — that is a business risk, not a technical one. All it takes is for that person to change career.

What is not on the list and should not be: "I do not like it". That is a reason for a conversation about design, not for a redesign — and the most expensive way to solve a problem that is usually the owner's, not the customer's. The customer is not comparing your site with what you saw at a conference, but with three others from the same trade, open in adjacent tabs.

Symptoms that point to optimisation

• Traffic is there, enquiries are not. The commonest case and the cheapest to fix. Usually what the site does not say — price, lead time, who it is for — or a form asking seven fields instead of two.
• The site loads slowly. Technical work, not design work. According to the Google/SOASTA study from 2017, the probability of leaving rises by 32% going from 1 to 3 seconds and by 123% at 10. The study has years on it and covered mobile traffic, but the direction is stable. The figure for your own address is in the website speed test.
• One page, or a few, dropped out of the results. A local problem, solved locally. Rebuilding everything because of one page is like renovating a house because a tap drips.
• The site does not answer the questions that come up in meetings. If your salesperson explains the same thing every time, a section is missing — not a new design.

What optimisation actually covers

The word is used as an umbrella for everything, so break it into specific kinds of work. Each fixes something different and each can be quoted separately:

Content and structure. Rewriting sections to answer questions in the order the customer asks them. Adding a price range, or what moves it. Shortening the form. Putting contact where the reader is ready to leave it, not only at the end.

Performance. Compression and correct image formats, removing scripts nobody uses, fixing the load order. Measurable the same day and needing nobody's permission.

Technical SEO. Titles and descriptions, heading structure, internal linking, the sitemap, removing pages competing for the same phrase. Most often this is the work that unblocks traffic — not new content.

Measurement. Counting enquiries, calls and form submissions correctly. Without it every later change is judged by impression rather than result — and it is the first thing to do, before anything changes.

To go through it systematically, we have a self-audit checklist for a website — no sign-up.

How a redesign runs in practice

Worth knowing what you are agreeing to: half the work sits with the client, and it is that half that stretches the timeline.

Stage 1 — the decisions. Why the site is to exist, who it speaks to, how it is ordered. A conversation, not a task for the contractor; set out separately in five decisions before the project. Skipping it does not save time — it moves it to the end, where changes are most expensive.

Stage 2 — content. The commonest cause of delay in the project. Copy, photographs, service descriptions, reference details. If they are produced at your end, start writing the day the contract is signed, not the day the contractor asks.

Stage 3 — design and build. Sits with the contractor and usually runs most predictably.

Stage 4 — moving the addresses. More below: the only stage whose omission costs after launch rather than during it.

Stage 5 — the first month after go-live. Observation, fixes, and checking that what was meant to improve did. A project ending on launch day ends at the worst possible moment.

The risk the quotes stay quiet about: your addresses

This is the part worth reading even if you skip the rest.

Optimisation is evolution — the site stays where it was and the risk is small. A redesign is a revolution, and carries one cost invisible in the quote: the web addresses. A new system usually means a new address structure. Every old address that stops working is a search position built over years and deleted on launch day.

Google describes the procedure outright: before the migration, map old addresses to new ones, take the most important from the sitemap, set the redirects and monitor both sets after go-live. It sounds like a formality and is the first thing to fall out of the schedule when the deadline slips.

We have seen what it looks like when it is skipped — in our own corpus. Tidying the blog, we found eleven addresses that for sixteen months had answered with status 200 and the content of a 404 page. The worst possible answer: for a user the page does not exist, for a crawler it looks healthy, so there is no reason to replace it or report a problem. Nobody noticed, because statistics do not show addresses nobody visits any more.

The lesson carries to every migration: checking that the existing redirects work says nothing about the addresses that never got one. Ours worked, forty out of forty — and the list was still incomplete. Ask from the other end: take every address the search engine knows, subtract the ones that respond and the ones that redirect, and look at what is left.

Three things worth having in the contract, or at least in an email, before a redesign starts:

1. Who prepares the address mapping and from what — the sitemap and the Search Console list, not memory.
2. When the redirects will be checked — before go-live, not "if something turns out to be broken".
3. Who watches the traffic for the first month and at what drop the alarm goes off.

Not technical requirements: three sentences that separate a redesign from a redesign that ended in losing what the site had built.

What it costs

Honestly: we have no market data on redesigns and will not guess. Redesign figures circulating on the internet usually have no measurement behind them, and copied from article to article they look like fact within a year.

What can be said, anchored on published price lists for new builds — never our own rates, and always with the market named:

• A new build gives you the floor to measure against. German agency price lists put a landing page at around €1,500 and a company site of up to a dozen or so pages at around €4,300. A Dutch agency publishing bands from its own delivered projects gives €3,600–16,500 for low-code sites and €31,000–73,500 for custom builds — the only primary European figure we have for the top tier. The UK is a separate price level and should not be blended in.
• The spread comes from the supplier, not the technology. The same Dutch source gives the ladder: freelancer €500–2,000, a small studio's custom work from €10,000, a mid-sized firm from €30,000, a large specialist from €100,000 up. A market guide with no stated sample, so read it as the shape of the market rather than measured prices — but the shape is the point: same work, different operating models.
• Geography moves the price more than buyers assume. The French freelance day-rate barometer prints separate columns for Paris, large cities and the regions, with a gap of roughly a fifth at senior level.
• A redesign is a new site plus moving what exists. Content migration, address mapping and checking nothing dropped out is work a new site does not have — and it accounts for the difference.

Optimisation is billed differently, as work on an existing site: in hours, usually monthly packages or one-off jobs with a defined scope. And rates are moving — median day rates for JavaScript roles in France fell by between 10.6% and 16.6% year on year, and the German freelancer survey recorded its first fall in the overall average since it began, so a rate card from two years ago is not a benchmark. Which is why "redesign versus optimisation" has no single-number answer: two different billing units. An indicative amount for your own scope is in the website cost calculator.

The hybrid approach — when it makes sense and when it is an excuse

It makes sense when a redesign is settled but not for now: you fix what can be fixed without touching the foundations — speed, content, forms — and plan the rebuild for when there is budget and a reason. That is managing the order of work, not avoiding the decision. One condition: a date in the calendar and an indicator that says the moment has come.

It is an excuse when the answer to question 1, 2 or 3 is "no" and you still add fixes every quarter to a site whose foundation needs replacing. After two years those fixes usually add up to more than the redesign — with the difference that nothing remains of them, because it all went into a system that has to be replaced anyway.

Before anybody issues a quote

Four things to do this week, none of them costing anything. They go straight into the request for proposal and halve the conversation with a contractor:

1. Go through the four questions at the top and write the answers down. That is the diagnosis.
2. Pull the numbers out of Search Console — impressions and clicks over three months. Without them nobody, including you, can judge afterwards whether it worked.
3. Write one sentence saying why the site is to exist. With a verb and a number. If it cannot be written, neither route yet has a goal.
4. Check who has the access — domain, hosting, system, reporting. The commonest cause of delay at the start and the worst moment to find out.

To check whether you are measuring the right things before changing anything, see website KPIs.

What next

• It came out a redesign → website quote, no obligation, with ranges instead of "it depends"
• It came out optimisation → technical SEO, measurement and work on the existing site
• Still not sure → we go through the four questions together and say plainly whether the site needs a service or a renovation — including when the answer is "nothing for now"