In the early years, building a website cost money. Not because the underlying technology was especially complex, but because the people who understood it were scarce and the tools were primitive and the range of things that could go wrong was extensive and not always visible until something had already gone wrong.
A site that worked—that loaded correctly across different browsers, displayed consistently on different systems, navigated without breaking, completed its transactions without losing data—required someone who understood the medium well enough to anticipate its failure modes and build around them.
I built sites in that period: efficient, structurally correct and technically sound. I was particular about the things that didn’t show—the code behind the page, the logic of the navigation, the behaviour of the form that sent the contact request to the right place rather than into nothing. I was paid well for this, which reflected the market at the time. The skill was valued because its absence was visible. A site that didn’t work was obviously broken. A client could see that it was broken. The broken thing was the evidence of insufficient skill, and the market priced accordingly.
Then the tools arrived.
The tools democratised production. Template systems, drag-and-drop builders, low-cost platforms that allowed anyone to assemble something that looked like a website without understanding what was underneath it. This was not, in itself, a problem. Democratisation of capability is generally a good thing. More people could build more things, and many of those things served genuine purposes for the people who built them and the people who used them.
What changed alongside the tools was the price. When anyone can produce something, the specialist premium disappears. Supply increases faster than demand, and price becomes the dominant factor in selection. The question stops being does this work properly and becomes does this cost less. These are different questions. They produce different results.
The low-cost provider does not build a worse site out of malice or indifference. That provider builds a site to the standard the market now requires, which is the standard of visible presence. The site exists. It loads. It has pages. The pages have content. The navigation produces something when you click on it. This is what the market now defines as done, and the market is correct in the sense that this definition is now what most buyers accept and most sellers deliver.
A site that works properly costs more than a site that merely exists. Efficiently coded, correctly structured, functional across all browsers and systems, complete with privacy policy and contact details and correctly formatted metadata and logical navigation and forms that actually deliver their content to someone—that costs more to produce than the site that merely exists. The additional cost produces qualities that are largely invisible to the buyer. Invisible qualities are qualities the market does not pay for.
I want to be precise about what invisible means in this context, because it is doing the central work.
The qualities I am describing are not invisible in the sense of being unimportant. They are invisible in the sense that the buyer cannot perceive their presence or their absence without specialist knowledge of the kind that, once the market democratised, most buyers no longer needed to acquire. In the early period, the absence of those qualities showed immediately—the broken layout, the failed transaction, the site that loaded on one browser and displayed nonsense on another. The failure was the evidence. The failure was visible.
The tools changed this. Template-built sites tend to be structurally consistent within their template, which means the obvious failures are less common. The layout holds. The navigation mostly works. The contact form sometimes delivers its content and sometimes sends it into nothing, but the buyer cannot easily tell which, because the form appears to function from the front. What doesn’t work is not visible from the surface.
The missing privacy policy is not visible until someone asks for it. The missing contact information is not visible until someone needs it. The bloated code that loads the page in four seconds instead of one is not visible unless someone measures it, and most buyers do not measure it. The metadata that tells search engines what the site is about and who it belongs to is not visible on the page at all. The broken link on the third page of a navigation that no one has clicked since the site launched is not visible until someone clicks it.
These absences accumulate behind a surface that looks complete. The site exists. The system takes existence as evidence of function. The evidence is wrong, but the system has no mechanism for detecting that it is wrong, because the mechanism would require the perception of things that are not visible on the surface.
Organisations do not reject quality. This is the important clarification. The companies and individuals who buy low-cost websites are not making a deliberate choice to accept inferior work. They are making a choice based on the information available to them, which is the information the surface provides. The surface shows a site. The site looks like other sites. It has the same general structure, the same visual conventions, the same apparent functionality. The buyer compares it to the standard they can perceive, and it meets that standard.
What they cannot perceive is the difference between a site that exists and a site that works. Not because they are incurious, but because perceiving that difference requires the kind of knowledge that the democratisation of production made it unnecessary to acquire. When specialists were scarce, buyers needed to evaluate specialist work. They developed, or hired, the capacity to do so. When the tools arrived and anyone could produce a site, the need for that evaluative capacity disappeared. The capacity atrophied. The standard the remaining capacity can measure is the standard of visible presence.
Quality becomes invisible when the system stops measuring it. And the system stops measuring it when the price signal no longer rewards it.
Companies once trained staff to perform specific roles well. Training made performance visible because the difference between trained and untrained staff showed in outcomes—efficiency, accuracy, reliability.
Training costs money. Its benefits are distributed over time and not always directly measurable. Many organisations reduced or removed it, preferring to hire for pre-existing ability and accept whatever variation in performance followed.
The result is consistent with the pattern. The staff member exists in the role. That is taken as evidence that the role is being performed.
The sites I see now, produced by low-cost providers at the standard the market accepts, share a set of characteristics that are consistent enough to constitute a pattern. They are visually adequate. They display on a screen and contain information about the organisation that commissioned them. They have a home page and an about page and a contact page, though the contact page often contains a form that is not regularly checked and no telephone number or physical address.
They do not have a privacy policy, or have one that was generated automatically and has not been read by anyone in the organisation. Their code loads resources they are not using, remnants of template elements that were not needed and not removed. Their navigation works on the most-used paths and breaks on the less-used ones. Their metadata was populated automatically with the template’s default values and was never corrected.
None of this triggers failure in any visible sense. The site loads. The home page displays. Someone searching for the organisation finds it. The organisation considers the website done, because the website meets the standard of done, which is existence.
The site is done in the way that a building is done without plumbing. The absence of plumbing is not visible from the street.
I am describing a shift in what the system can see, and therefore in what it values. This connects to something I have been observing across a range of other contexts.
A company requests evaluation of a service that has not yet been delivered, because its system recognises the sending of a reply as the completion of a service. Presence replaces performance. A request for feedback substitutes for an outcome.
A system asks whether anything else can be helped with before the first thing has been helped with, because its system recognises the sending of a response as the resolution of an enquiry. Response replaces resolution.
A transaction ends and then extends, through a review request that treats the purchase as incomplete until the buyer has produced content for the seller’s marketing. The transaction’s boundary is invisible to the system because the system measures its own activity, not the boundary of the agreement.
In each case, the same structure: the system measures what it can see, and what it can see is its own activity. The activity is taken as the outcome. The outcome is taken as the standard. The standard degrades to the level of the visible.
A website exists. That is taken as evidence that it works.
There is a loss embedded in this shift that goes beyond the immediate quality of individual websites. The knowledge required to build something that works—the understanding of the underlying structure, the anticipation of failure modes, the habits of testing across systems and browsers, the attention to the invisible elements that determine whether the form sends its content to a person or to nothing—that knowledge accumulated over time, in practice, through the experience of building things and discovering how they broke and learning to prevent the breakage.
When the market stops rewarding that knowledge, people stop acquiring it. When people stop acquiring it, it stops being transmitted. The institutional memory of what it takes to make something work properly does not disappear immediately. It fades. The people who carry it age out of the market. The people entering the market learn to use the tools, which do not require the underlying knowledge, and produce sites that meet the standard the market has settled on, which is existence.
No single site fails catastrophically. One poorly built website is not a crisis. But the accumulated result of a market that systematically selects for visible presence over functional quality is a large number of things that appear to work and do not, maintained by people who cannot tell the difference, serving organisations that have lost the capacity to ask the question.
Nothing appears to be broken. That is taken as success.
The site loads. That is sufficient.
The system sees a website. It does not see what is missing.
