How to tell whether your website needs a redesign — or just needs fixing
Almost everyone asking this question has already decided the answer is a redesign, because the site looks tired and that is the visible thing. In four situations examined here, a new site was the fix for the problem exactly once — and one of the sites that needed fixing instead was four weeks old.
The site looking dated is the symptom you can see, not the one that costs you.
A website has two audiences and they fail differently. The owner looks at it every week and sees a design that has aged. The customer never sees it at all, because they searched for something and were shown someone else. Those are different problems with different prices, and only one of them is fixed by a new design.
This matters commercially because the two options are not close in cost. A rewrite of four page titles takes an afternoon and can be measured inside a month. A redesign takes weeks, and if the constraint was somewhere else, it produces a better looking site and the same result — at which point the money is spent and the question is still open.
A redesign is the right answer often enough to be believable and rarely enough to be expensive.
Four checks, in order. All four can be run this week.
None requires hiring anyone. Three need Search Console, which is free, and the fourth needs you to look at what is installed. They run in order because a later answer cannot be trusted while an earlier one is broken — the same sequencing that the Growth System applies to the whole business.
- Is anything reaching the site at all? Open Search Console and read impressions, not rankings. If the site is shown for almost nothing, the problem is upstream of design — nobody is refusing your pages, because nobody is being offered them. A redesign of pages nobody sees changes nothing that can be measured.
- Is it shown a lot and clicked rarely? This is the opposite fault and it looks nothing like it. Google is willing to put you in front of people and they are choosing something else — which is usually the title, the description, or a page answering a question the searcher did not ask. It is almost always an edit, and it is the cheapest fix in this article.
- Are several of your own pages competing for the same search? Search the query and see how many of your own URLs appear, or read the query-and-page view in Search Console. Multiple pages chasing one term is self-inflicted, common, and resolved by giving each page one job — retitling and relinking, not rebuilding.
- Can you tell what happens after someone arrives? If there is no analytics, no event on the contact button and no way to count enquiries, then no redesign can be judged. You will spend the money, the site will look better, and you will have no way of knowing whether it worked. Install measurement first, always.
The second check is the one people skip and it is the cheapest thing on this page. A page earning impressions and no clicks is a page Google is already willing to show you for. Nothing has to be built to fix it. Something has to be rewritten — and until it is, a redesign is being considered to solve a problem that a title tag caused.
Four real situations. The answer was different every time.
These are not illustrations. Each one is an assessment or a reporting period Aroc ran between June and August 2026, and each is sourced at the foot of the page.
Why one business is named and one is not. Evergreen Spa gave permission on a recorded call on 3 August 2026, and every Evergreen figure below is already published at /work/evergreen-spa where anyone can check it. The July assessment gave no such permission, so the findings are reported and the identity is not.
- Case one · vehicle service contracts, Missouri · July 2026
A redesign would have been the most expensive possible wrong answer.
The only web address this company had ever published — it appears on every post of their own blog — still resolved. It served a template consulting site for a different company, with a different street address in the same city and a New York area-code phone number. Not a 404. Not a parked page. Another business, taking the call.
Nothing about that is a design problem. Commissioning a new website would have produced a handsome site at an address the company no longer controlled, while the old address kept answering the phone for someone else. The first item on the list is recovering or replacing the domain, and no amount of design substitutes for it. What that costs was not established either — it depends on whether the old address can be bought back — and the assessment says so rather than guessing.
Verdict: neither, yet. Settle the domain first, then ask the question again. Confirmed by direct fetch, 31 July 2026. Whether the domain lapsed or was sold was not established — WHOIS history could not be obtained.
- Case two · Evergreen Spa, Rawai, Phuket · June–July 2026
This one genuinely needed the new site.
A good business that could not be found. Strong reviews, returning customers, and roughly half of them Thai residents rather than tourists — the harder audience to win. The constraint was the earliest one there is: nobody could discover it. Not a conversion problem, not a copy problem. A findability problem, and the site was the instrument.
One month after the new site went live, Evergreen appeared in 23 distinct searches, having appeared in none, and was shown 552 times. That is what a redesign is for, and it is the narrow case in which it is the right purchase: when the honest answer to “what is stopping this business” is that the internet does not know it exists.
Verdict: redesign. Google Search Console, 28 June – 27 July 2026, published in full at /work/evergreen-spa with the client's recorded permission.
- Case three · the same Evergreen site, the same first month
A four-week-old site that already needed fixing, not rebuilding.
The same period that produced those numbers also produced the faults. The site went live at the start of July; by the end of the reporting window those 552 appearances had produced 8 visits, and the report led with that rather than burying it. Four things were recorded as not working. Two of them are the ones this article is about, and neither is solved by design.
The wellness page was shown 191 times and clicked zero times, because Google was matching it to people searching wellness resort in rawai phuket — people looking for somewhere to sleep. Evergreen is a day spa. That single search accounted for 256 of the 552 appearances. The fix is words on a page: say “day spa”, say there are no rooms.
Separately, three of the site’s own pages competed for rawai spa — an exact-offer search shown 51 times a month. Google resolved it by ranking none of them well: all three sat between position 42 and 55. The remedy is to give each page one job and point the internal links at whichever should win. Editing, not rebuilding.
Verdict: fix. Same Search Console period, same published page. The remedy for both faults is copy, titles and internal links on a site four weeks old. The other two faults recorded that month — desktop performance, and enquiries that could not be counted because tracking was never installed — are real and are not copy edits; they are named here so the two chosen above are not mistaken for the whole list.
- Case four · Aroc’s own website · August 2026
We rebuilt every page and it did not fix what was broken.
Aroc rebuilt its own site — twelve pages, every word new — and shipped it on 4 August 2026. It was done without a keyword or intent map and without anyone owning search. Copy was written as positioning, judged on how it read, and published.
A map of what each page should target was then written retrospectively, which is the wrong order and the reason the rule now exists. Its verdict on the new site was that it is well written and targets almost nothing measurable. Of its twelve rows, four record Search Console evidence, three are pages that should never target anything — legal, navigational and a noindexed thank-you page — and the remaining five carry no evidence at all. Those five are the three product pages, the method page and the one industry page: every page built to sell something. It reads better than the one before it and is, on the only question this article is about, in the same position.
The redesign was not wasted — the positioning it carries is the actual product. But it was not a visibility fix, and calling it one would have been the comfortable version.
Verdict: the redesign was the right call for a different reason, and answered the wrong question. Recorded in ARO-S-002 §0 and ARO-W-003 §1, both written before this article.
When a redesign genuinely is the answer.
Four situations, and they have nothing in common with how the site looks.
- The business cannot be found at all, and the site is the instrument for changing that. Evergreen above is this case, and the honest way to recognise it is that the number to be improved starts at or near zero rather than at a disappointing figure.
- The site cannot be changed. No access, an abandoned platform, or every small edit requiring a developer and a fortnight. If fixing is not actually available, comparing its cost is theatre.
- It fails on a phone, which is where most people will meet it. This is the one case where the visible problem and the real problem coincide.
- The business it describes is no longer the one you run. Then the words are wrong rather than the rankings, and no amount of editing round the edges catches up.
Outside those four, the cheaper move usually wins — not because it is cheaper, but because it can be measured within weeks. A change you can evaluate quickly is worth more than a larger change you will rationalise afterwards.
If you do redesign, the question most quotes leave out.
A redesign can lose you search visibility, and the mechanism is ordinary rather than mysterious: pages that had value are moved or removed, no redirect is put in place, and the value has nowhere to go.
Aroc did this to itself deliberately and wrote down what it expected. At the cutover on 4 August 2026, two pages were retired that between them carried 156 of the 182 impressions Search Console attributed to individual pages — 85.7% of the site’s page-level visibility. Both were redirected. The baseline was frozen on 3 August, before anything moved, and the fall in impressions was predicted in writing beforehand rather than explained afterwards.
Why that says 182 and not 157.Search Console’s own totals disagree with each other by design: the site was shown 157 times counted by date, while the per-page rows for the same period sum to 182, because a single appearance can be attributed differently across dimensions and rare rows are anonymised. Dividing a page-dimension number by a date-dimension total produces a tidier figure — 99% — that corresponds to nothing. Aroc published that 99% in its own baseline document and it was wrong; it was found while fact-checking this article and corrected on 6 August 2026. The comparable figure is 156 of 182, or 85.7%, and it is the one above.
Know what each page is worth before you move it. That sentence is the whole of the redirect discipline.
So the question worth asking any quote is not what the redesign costs. It is which pages currently carry search value, where each will point afterwards, and what the numbers are today so the change can be judged. A quote that has no answer has not looked.
The honest answer to “so which is it for me?”
Nobody can tell you from the outside without looking, and anyone who does is guessing in a direction that suits them. Run the four checks yourself — they are free, and if the answer comes back obvious, act on it and keep the money.
Aroc Assess is the paid version of that exercise: one constraint named, the evidence for it, what it costs, and what to do first — including, where honest, that major work is not justified yet. It is built entirely from what is visible from outside, with no account access, and that limit is written into the document rather than discovered later.
If the answer turns out to be a rebuild, Build is where that work lives. If it turns out to be the slower, cheaper, compounding kind of fix, Grow is where it lives instead. The point of diagnosing first is that those are genuinely different answers and the diagnosis is allowed to reach either.
Sources for every figure above
- Case one — the domain serving another company — Aroc assessment run 003, 31 July 2026, confirmed by direct fetch. Internal document; the business is unnamed because no permission was given. WHOIS history could not be obtained, no claim is made about how the domain changed hands, and no cost is given for recovering it because none was established.
- 23 searches · 552 appearances · 8 visits · 191 impressions and zero clicks · 256 of 552 · three pages at position 42–55 · “rawai spa” shown 51 times a month — Google Search Console, 28 June to 27 July 2026, re-verified 3 August 2026. Published with recorded client permission at /work/evergreen-spa, where each of these figures already appears.
- Twelve pages rebuilt without an intent map · “targets almost nothing measurable” · four evidenced, three exempt, five with no evidence— Aroc internal records ARO-S-002 §0 and ARO-W-003 §1, both written before this article. The quoted verdict is the map’s own; the four / three / five split is counted from its table rather than from its summary line, which states a different figure that does not reconcile with the table beneath it. That discrepancy is logged for correction and is not repeated here.
- 156 of 182 page-level impressions on two retired pages — Google Search Console, domain property, captured 3 August 2026 before the redirects went live. The two pages were
/phuket-seo(108 impressions) and/phuket(48). The six page rows sum to 182; the site total counted by date is 157. Both figures are correct and they are not interchangeable. The capture was requested for 6 July – 3 August but the daily rows it returned stop at 1 August — Search Console’s reporting lag — so the period is 27 days to 1 August, not 28 to 3 August. That gap was found on the day this article published and the capture script is being changed to record the dates it actually received. Both pages now redirect. - Search volume for “website redesign” — Google Ads Keyword Planner, United States national, pulled 3 August 2026: 2,900 per month, low competition. Referenced in the intent map rather than on this page, because a volume figure is evidence for choosing what to write, not a claim to make to a reader.
Nothing above is estimated. Where a number could not be verified it is absent rather than approximated. Every Evergreen figure used anywhere on this page — the seven listed above, and the two qualitative ones in the text — is quoted from a page already published rather than from the internal month-two work order, so that everything a reader is asked to believe can be checked in one place. The single percentage on this page is written as 85.7% rather than 86%, because an article about a rounding that flattered its author should not round in its own favour.