We crawled webask.co.uk for this update. Not a client's site — this one. The WordPress install that has occupied this domain since 2020, and which the page you are reading is in the middle of replacing.
That crawl is the most useful thing we can bring to this argument, because it is first-hand rather than hypothetical. It is also public — every URL below can be checked against the live site until we switch it off.
The short version
Over three years, in sterling, the cheaper build stays cheaper — and the gap gets wider, not narrower. The published UK bands price a build by supplier rather than by platform, so a three-year table in which the expensive option wins on cash has almost certainly inflated a column. The case for a custom build is real, but it is not an arithmetic case.
What six years of not maintaining a WordPress site looks like
The sitemap listed 13 URLs. Among them:
/sample-page/— WordPress's own placeholder page, last modified in August 2020, still live, still submitted to Google nearly six years later./privacy-policy-2/— the-2is the scar of a duplicate page created once and never tidied./uncategorized/— the default category, and the only one. No taxonomy was ever built.- Three of the six entries in the post sitemap carried a last-modified date of October 2020.
- The footer linked to "Our Clients" and "Our Partners". Neither target appears anywhere in the sitemap.
/wp-login.phpreturned 200 and/xmlrpc.phpreturned a 520 — two endpoints that sit at the same address on millions of sites and are probed continuously by credential-stuffing bots.
Read that list again and notice what is missing: a WordPress defect. Every item is what happens when a website has no owner. Put the same neglect on a custom build and the symptoms change while the cause does not — dependencies years out of date, a deploy pipeline nobody can run, a contact form that stopped sending email and told no one. The platform decides the shape of the decay, not whether it happens.
Worth settling first, because a great deal of "WordPress is insecure" marketing is an argument about staffing wearing a technical costume.
The three-year sum, in sterling
Here is where a three-year comparison can quietly stop being honest. The UK market bands are published and easy to check.
One thing to say plainly before the sum: those bands are cut by supplier, not by platform. No reliable UK figure exists for a "WordPress build" against a "custom build", because both sit across the whole range — a £3,500 agency site can be WordPress. So the honest comparison is a cheap build against an expensive one, whatever is underneath either. If you are deciding between a single campaign page and a whole site, start with landing page or website.
Published UK freelance quotes run £800–£3,000 in one 2026 market summary and £1,500–£3,000 in another; a four- or five-page build is put at £1,200–£2,000. Agency work runs £2,500–£10,000. So take £1,500, which sits within both freelance bands, against a £3,500 agency build, and run each forward three years at the market's own running-cost rate.
At the midpoint of that 15–20% band, the cheap route spends roughly £790 on upkeep across the three years (£675–£900 across the full band) and the expensive route roughly £1,840 (£1,575–£2,100). Midpoint totals: about £2,290 against about £5,340.
The build-price gap was £2,000. Three years later it is between £2,900 and £3,200 — wider at every point in the range.
Why the percentage cannot rescue the expensive option
Running cost is quoted as a proportion of build cost, so it scales with the build. At the same rate on both, a percentage applied to a bigger number stays a bigger number: it cannot close the gap, only widen it. Reversing that would take a materially lower maintenance rate on the expensive build, and no UK source publishes one. Any three-year table showing the reverse has either inflated the WordPress maintenance column, set the custom column to zero, or added a projected revenue line that is a forecast wearing a cost label.
Two caveats point the other way. The narrower £100–£300 a year quoted for a cheaper build covers domain and hosting only — the cost of existing, not of changing. And a site that cannot be edited, extended or moved gets bought twice, which puts the whole build price back on the table. Neither reverses the sum. Both belong in the spreadsheet.
So what actually decides it
Not the money. Three questions do, and none of them is really about the platform.
Is the maintenance line optional?
On WordPress it is not. Core, PHP, themes and plugin licences all move, whether or not anyone is being paid to follow them. That is the direct cost of the ecosystem's greatest strength: scale means a shared attack surface, which is why /wp-login.php sits at the same guessable address on our legacy install as on every other WordPress site. A custom build carries a smaller surface — but "smaller" is not "none", and an agency implying otherwise is selling the same neglect in a nicer wrapper.
So the question for any supplier is not "is it secure" but "who applies the update on the Tuesday it ships, and what happens if they don't".
Does anyone actually edit?
WordPress's genuine advantage is the editing dashboard and the size of the pool of people who can drive it. That is worth a great deal — if content changes. If it changes twice a year, you are carrying the cost and the update obligation of a publishing platform in order to run a brochure. Our own domain illustrates it: six years of dashboard access, and five of the twelve content URLs still carrying a last-modified date from 2020.
So answer for the publishing cadence you will really sustain, not the one in the plan. That moves this decision further than any benchmark will.
Is the constraint structural or cosmetic?
This is where a custom build earns its price properly, and it has nothing to do with speed scores: in UK field data, WordPress sites passed Core Web Vitals more often than Next.js sites. Some sites have a required shape rather than a preferred one. UK clinics are the clearest example: the rules on advertising prescription-only medicines to the public shape where certain pages may sit and how they are reached — an information-architecture specification, not a styling choice. Constraints like that are cheap to build in and expensive to retrofit, on any platform. The detail is on websites for UK aesthetic clinics.
If your constraints are cosmetic — the site works but looks a decade old — you are looking at a design project, not a replatform.
Switching later: what a redirect carries, and what it doesn't
You can move. We are moving. But it helps to know precisely what survives the trip.
A permanent redirect carries
- The large majority of link equity from each mapped URL
- Rankings — in reported migrations, after two to four weeks of fluctuation; some sources describe a one-to-three-month settling period instead. Nobody can promise the timing, us included
- Your domain and its history — the domain isn't changing, only the platform
- Any page you rewrite and republish, provided the old URL is mapped to it
It does not carry
- Quality. A redirect moves a URL, not a page worth reading
- URLs you never found — author and date archives, feeds, tag pages,
/blog/page/2/ - Plugin functionality. Forms, booking and popups are rebuilt, not migrated
- Years of not publishing. That arrives intact on the new stack
Three specifics we hit while preparing this cutover, each easy to get wrong. Crawl the live site and treat the export as the map, because WordPress exposes far more URLs than its sitemap admits. Send genuine junk to a 410 Gone rather than redirecting it — /sample-page/ never held content, so a redirect lies to the crawler and a 404 invites months of retries. And do not tidy those paths away in robots.txt: Googlebot must be able to fetch a URL to learn it is gone, so blocking it keeps it indexed indefinitely. Then hold every redirect for at least a year, which costs nothing.
What we are deliberately not showing you
There is no before-and-after graph in this post. The replacement site is not live yet, and the site it replaces carries close to no authority, so any chart drawn here would be decoration.
Apply the same scepticism elsewhere. When an agency shows you a WordPress-to-custom traffic curve, ask what else changed that quarter: the copy, the content budget, the link building, the person finally paying attention. The graph may well be real. The attribution is what is being sold.
Where this leaves you
If the site is a brochure that changes rarely, WordPress or a builder is both the correct answer and the cheaper one. If the site carries the load — leads arrive through it, paid traffic lands on it, or its structure is set by rules rather than taste — the more expensive build has to justify itself on that job, not on a spreadsheet.
Our web development service sets out the four routes and what each really buys. If you already have a site and are not certain it needs replacing, a free audit will tell you — and "keep what you have" is a legitimate conclusion for one to reach. Starting prices sit on the pricing page. If search is the reason you are considering a rebuild, read our SEO page first; that can be the cheaper fix.