In 1996, the average lifespan of a webpage was estimated at 75 days.

The internet was young, chaotic, and nobody expected it to last. Websites were experiments. Links were suggestions. Permanence was not a design consideration because the whole thing felt temporary anyway.

Thirty years later, the average lifespan of a webpage has improved. It is now estimated at somewhere between 2 and 3 years before a URL either redirects, returns a 404, or disappears entirely. We have built an enormous, load-bearing global infrastructure for human knowledge, commerce, and culture on a foundation that deteriorates faster than a paperback left in the rain.

And we have designed it this way deliberately, without ever quite deciding to.

What Link Rot Actually Is.

Link rot is the process by which hyperlinks stop working over time. The page they pointed to has moved, been deleted, or the entire domain has lapsed. The link remains. The destination does not.

It sounds like a minor inconvenience. It is not.

A 2021 study by Harvard Law School's Perma.cc project found that more than 50% of URLs cited in Supreme Court opinions no longer work. Legal documents, referenced at the highest level of a judicial system, pointing to nothing. A study of academic papers found that links in scientific citations had a half-life of approximately 9.3 years. Medical research, policy papers, and institutional knowledge, dissolving quietly in the background whilst everyone assumes someone else is maintaining them.

For brands, the implications are less dramatic but equally corrosive. A campaign microsite from 2019 returning a server error. A press release linking to a product page that no longer exists. A portfolio piece pointing to a case study that was never migrated when the agency redesigned its website. An entire archive of brand communications quietly becoming unreliable.

None of this is catastrophic in isolation. Cumulatively, it is a brand credibility problem dressed up as a technical one.

The Dependency Problem.

Link rot is the visible symptom. The underlying condition is dependency.

Modern websites are not self-contained documents. They are assemblies of third-party components, each one a potential point of failure.

A typical brand website in 2026 might depend on:

• A headless CMS serving content via API

• A third-party font service rendering typography

• An external video hosting platform embedding media

• A cookie consent management platform

• An analytics and tracking layer

• A payment processing integration

• A marketing automation tool handling form submissions

• A CDN delivering assets from distributed servers

• Multiple social media embed integrations

Each of these is maintained by a separate company, on a separate commercial trajectory, with a separate set of priorities. Every single one of them is a dependency the brand does not control.

When Heroku discontinued its free tier in 2022, thousands of small applications and websites that relied on it went offline overnight. When Twitter restructured its API access the following year, embed integrations across hundreds of thousands of websites broke simultaneously. When a Google Fonts API update changed URL structure, sites that had not updated their references began serving fallback system fonts without anyone noticing for weeks.

These are not edge cases. They are the normal operating conditions of the modern web.

And most brand websites are built with the assumption that all of these dependencies will continue behaving exactly as they do today, indefinitely, for free, without interruption.

That assumption has never been safe. It is becoming less safe every year.

Why Agencies Bear Responsibility for This.

The easy answer is to blame clients. Short budgets, unrealistic timelines, no appetite for maintenance retainers. All of this is real. None of it fully explains the problem.

The deeper issue is that the web development industry has not built digital permanence into its value system. Agencies are incentivised to launch. New builds generate revenue. Maintenance does not. Redesigns every three to four years generate more revenue than a single well-built site maintained carefully over a decade.

The commercial model of most agencies actively works against building for longevity.

A physical architect who designed a building that required complete structural replacement every four years would not be considered innovative. They would be considered negligent. The web has not yet developed equivalent professional standards, and brands are paying the price in broken links, deprecated integrations, and digital estates that require constant, expensive emergency intervention.

"We would never accept a physical building that collapsed when the supplier of one component went out of business. We accept it online without question."

This needs to change. And the change has to start at the design and build stage, not the maintenance stage.

Designing for a 50-Year Digital Lifespan.

This is not a theoretical exercise. Institutions, governments, universities, and major cultural organisations genuinely need digital infrastructure that survives decades. The BBC has content from the early 2000s that still needs to be accessible. National archives are grappling with digital preservation at scale. Law firms need documents to remain referenceable across the lifetime of a case or a precedent.

The principles that apply at institutional scale apply to any brand that wants its digital presence to remain coherent over time.

Reduce external dependencies to the absolute minimum.

Every third-party dependency is a future liability. This does not mean building everything from scratch. It means being honest about which dependencies are load-bearing and which are conveniences.

Self-host fonts rather than serving them from external APIs. Store video assets on owned infrastructure rather than embedding from platforms whose terms of service change annually. Build forms that submit to owned databases rather than third-party marketing tools that may restructure their pricing model next year.

Each dependency removed is a future failure point eliminated.

Build on open standards rather than proprietary frameworks.

Frameworks come and go. JavaScript libraries that were industry standard five years ago are now unmaintained. Proprietary CMS platforms get acquired, pivoted, or discontinued.

HTML, CSS, and vanilla JavaScript have been stable for decades and will remain so. Websites built close to web standards are readable by browsers built twenty years from now. Websites built on framework-of-the-moment are dependent on that framework's community remaining active, funded, and interested.

The most durable digital objects are the ones that require the fewest assumptions about the future.

Design URL structures that do not need to change.

URL structures are often redesigned when websites are rebuilt. New navigation, new taxonomy, new CMS, new slug format. Every change orphans every link that ever pointed to the old structure.

Permanent, logical URL structures decided at the outset, and committed to across redesigns, are one of the simplest and most overlooked acts of digital preservation available. If the URL for a product page has been stable for ten years, every reference to it, in press coverage, in backlinks, in archived documents, continues to function.

Redirect maps for historical URLs should be considered a core deliverable, not an afterthought.

Create static archives of critical content.

Dynamic websites serve content assembled on request from databases and APIs. If either fails, the content disappears. Static files, plain HTML and assets stored as documents rather than assembled at runtime, survive indefinitely regardless of what happens to the infrastructure around them.

Annual static archives of critical brand content, key campaign pages, historical press materials, core product information, are an insurance policy against the inevitable evolution of the web stack beneath them. They cost very little to store. They cost an enormous amount to reconstruct from memory.

Document everything as if the team will change completely.

Because it will. The average tenure at a digital agency is under three years. The developers who built the site will not be there when it needs significant maintenance. The internal team who managed the launch will have moved on.

Undocumented websites rot faster because nobody knows what they are dependent on, what decisions were made at build, or what will break if a single component is updated. Documentation is not an administrative task. It is a preservation strategy.

The Archivability Argument.

There is a broader cultural argument sitting underneath all of this.

The web is now old enough to have a history worth preserving. Digital campaigns, brand communications, and cultural moments from the early internet are disappearing faster than anyone is capturing them. The Internet Archive does extraordinary work, but it cannot save everything, and it was never supposed to be the sole repository for human digital output.

Brands that build with archivability in mind are not just protecting their own interests. They are contributing to a more durable digital culture. Websites that work without JavaScript for core content, that use semantic HTML that screen readers and archival crawlers can parse, that store assets in standard formats rather than proprietary ones, are websites that future researchers, journalists, and historians will actually be able to read.

A brand's digital estate is a record of its existence. Most brands are building records that will be unreadable within a decade.

What Good Looks Like.

Agencies serious about digital longevity ask a different set of questions at the brief stage.

Not just: what does this need to do at launch?

But: what does this need to still be doing in ten years? What happens when this dependency is discontinued? How do we migrate content if this CMS is acquired? What is the plan when this framework reaches end of life?

These are not comfortable questions. They extend timelines and increase costs at the outset. They require clients to think about maintenance as a continuous investment rather than a one-off launch expense.

But the alternative is a brand digital estate that requires emergency reconstruction every three years, haemorrhages broken links across every communication it has ever sent, and presents a progressively less coherent face to anyone who arrives from anything other than the homepage.

The web is not getting more stable. Consolidation, acquisition, API restructuring, and framework churn are accelerating. The brands that build for permanence now are not being conservative.

They are being strategic about something the rest of the industry is choosing not to think about yet.

The 404 page is not a technical error.

It is the cost of building something without asking how long it needed to last.