“We rebuilt the website a few years ago. It still looks fine. So why are we getting fewer enquiries from it every year?”
An operator asked a version of that in one of the conversations that keep shaping this cluster, and the straight answer took longer than he wanted. The site was fine. The photos were current, the phone number worked, the copy said what the company does. Nothing about it was broken, and that is exactly why the decline felt so confusing. Something that is not broken is not supposed to produce less every year.
Civilizations have always built to outlast their builders. The pyramids were designed to survive every dynasty that followed them, and a medieval cathedral could take a hundred years to finish because nobody expected it to need finishing twice. A website inherited that same instinct by accident: build it once, get it right, leave it alone. That instinct built the world’s monuments, and it is exactly the wrong one for the fastest-changing habitat humans have ever built something in.
A truck works that way. A warehouse works that way. Bought once and maintained, it does roughly the same job in year six as in year one. Operators price their websites the same way because everything else they own behaves like a capital asset: a build cost, a lifespan of five to seven years, then a rebuild. Nothing about the industry default is lazy. It is a sensible rule applied to the one asset it does not fit.
A website is not a building. It is a position in a race that never stops. The day it launched, it was as competitive as it would ever be. Every day since, the ground under it has shifted and the competitors around it have improved, and it has done neither. The site did not stop working; it stopped competing. Those are different failures with different fixes, and almost every operator is budgeting for the wrong one.
The ground moves on a schedule you do not control
The first half of the answer has nothing to do with competitors: the environment a website lives in gets rearranged on a published calendar.
Google’s own documentation says it plainly: “Several times a year, Google makes significant, broad changes to our search algorithms and systems.” These core updates are broad by design, and each one reshuffles which pages get shown for which searches. A site launched in 2021 has now sat still through more than a dozen of them. Nothing penalized it. Google re-scored it, repeatedly, against criteria that keep changing, while its answers stayed the same.
Then the scoring changed in kind. Pew Research Center measured what happens when Google shows an AI summary above the results: users clicked a traditional link in 8% of visits, against 15% when no summary appeared. Roughly half the click-through, gone, for searches where the summary shows up. That shift arrived after most operators’ sites were built, and a site that was finished in 2021 has, by definition, adapted to none of it. What that means for being chosen at all is its own subject, and the piece on what happens when AI chooses the mover walks through it.
Here is the arithmetic that makes “set and forget” a decision rather than a default. If the environment changes several times a year and the site changes zero times a year, the distance between what the site is optimized for and what the environment rewards grows on a fixed schedule. No single quarter shows it. Five years show it unmistakably. The operator asking why enquiries fall every year is describing exactly that curve from the inside.
The other side ships every day
The competitors compound the problem, because the front doors an operator’s site competes against are not other operators’ brochure sites. They are marketplaces, lead platforms and national brands, and those are software companies wearing a logistics costume.
Software organizations measure how often they improve what customers touch. DORA’s State of DevOps research, the longest-running study of how software teams perform, clusters teams by deployment frequency: the top cluster ships changes on demand, often multiple times a day, while the lowest cluster ships somewhere between once a month and once every six months. Read that lowest tier again. The slowest software companies in the study improve their product more often than most operators’ websites change in a year. The platforms bidding against an operator’s site for the customer’s first click sit at the fast end, and every test they run on headlines, forms and pricing display compounds into next month’s version.
This is a cost-structure asymmetry, not a talent gap. A marketplace that improves its quote flow once spreads that improvement across every listing and every market it serves, so continuous development costs it a fraction of a cent per customer interaction. An operator who wants the same cadence has to fund it from one company’s marketing budget. The platform does not out-think anyone about websites. Its advantage is arithmetic: the same dollar of improvement gets divided by a number ten thousand times larger. That asymmetry is a big part of how demand ownership drifted to the middle of the market in the first place.
The drift did not start with websites and it does not end there either; the fuller account of who owns the customer traces where it leads.
The uncomfortable implication: a finished website is not neutral in this race. Against opponents who improve daily, standing still is moving backwards at their speed. The gap between a static site and a continuously developed one is not the gap between good and bad. A photograph is competing with a film, and the judging happens again every quarter.
Decay you cannot see on the site itself
None of this decay is visible by looking at the website, which is why “it still looks fine” is both true and beside the point.
We saw the result across the industry directly. In late 2025 we went through roughly one hundred removalist websites and measured behavior instead of appearance. About one visitor in a hundred began a quote request. Of the visitors who began one, only a small fraction finished. Nearly every site had a form, and looked complete, and was quietly producing almost nothing. The fuller account of that review is already on this site, along with what it did to how those businesses get valued as fulfilment companies. The specific page-level reasons a site underperforms are also their own article, on why an operator’s website was never built as infrastructure, and this piece will not retell them.
What matters here is the shape of the finding, not the defects. A website’s appearance and its output are two different assets ageing at two different speeds. The appearance dates slowly; a 2021 design still reads as professional. The output decays continuously, because output depends on the match between the site and an environment that moved. An operator who inspects the site sees the slow-ageing asset and concludes nothing is wrong. The enquiry count is reporting on the fast-ageing one.
What keeping up costs, priced honestly
The obvious retort is to staff the problem. Hire the people the platforms have, run the same loop of measuring, changing and shipping. Price that before recommending it.
The Bureau of Labor Statistics’ national wage data, May 2025 median annual wages, prices the roles. One software developer: $135,980. A web and digital interface designer: $104,000. A marketing manager to direct what gets built and measured: $166,790. A deliberately lean team of one manager, two developers and one designer prices out at $542,750 a year in salaries alone, before payroll taxes, benefits, software and advertising spend. Trim it to one developer and it is still over $400,000. The team composition is illustrative; the wages are not. For a mid-sized operator, that figure does not fit inside the marketing budget. In plenty of cases it exceeds the budget.
The operator faces a fork where neither path is the one on the brochure. Build the team, and carry a cost that only makes sense spread across far more demand than one company generates. Skip the team, and fall behind on the schedule described above, at a rate set by Google’s release calendar and the platforms’ deploy cadence. Most operators believe they chose a third option, the sensible one: build a good site, maintain it lightly, revisit in five years. That option existed when the competition was other static sites. It stopped existing when the competition became software.
The skeptical reader is already objecting that his agency retainer covers this, and the objection deserves a straight answer. A retainer that publishes posts and patches plugins is maintenance, and maintenance keeps the photograph sharp. The platforms are not maintaining; they are measuring behavior and changing the product weekly. Ask one question of whoever looks after the site: what did we change last quarter because of something we measured, and what did it do to enquiries? If the answer is a redesign story or a traffic report, the site is being maintained, not developed. There is a difference, and the enquiry curve knows it.
The website you finished and the enquiries you lost
So, the question this piece opened with: the website was rebuilt a few years ago and still looks fine, so why fewer enquiries from it every year? Because a website is not a finished asset; it is a position in an environment that changes several times a year while competitors improve their own positions daily. The site’s appearance ages slowly, so it looks fine. Its competitiveness ages continuously, so it produces less. Fewer enquiries each year is not evidence the site broke. It is what a fixed position yields in a moving race, and it will continue on that schedule whether or not anything on the site ever visibly fails.
The depreciation is real even though no ledger carries it. An operator who spent $40,000 on a rebuild in 2021 did not buy a five-year asset. He bought a snapshot of what competing looked like in 2021, and the industry that priced it for him as a capital project did him no favours.
The test costs nothing to run. Pull up the site and ask when it last changed in a way a customer would notice. Then count how many times Google has shipped a core update since, and remember that the platforms between you and your customer shipped something this week. Whatever those three numbers are, they move together in only one direction, and two of them are not yours to control. The one that is left is the whole question.
