Skip to content

The Case for Bringing Back the 'Under Construction' Page

Beatrice Faulkner

Somewhere between the first hand-built homepage and the modern template, the web adopted an unwritten rule: nothing goes public until it looks finished. That rule quietly deleted the draft stage from public view, and with it a whole category of pages that were useful long before they were complete.

When Launch Polish Deletes the First Draft

The threshold for publishing has drifted upward. A coherent visual system, somewhere around five to eight populated navigation sections, copy edited to the last comma, a working search box — that is now the informal minimum before a personal project is considered presentable. Most beginnings die at that threshold, in a folder, unpublished.

The alternative is older and simpler. An Under Construction page, understood as a truthful status signal rather than a nostalgic sticker, lowers the entry cost to one useful page. It tells a visitor what exists, what does not, and that someone is still at the desk. The striped GIF was decoration; the declaration underneath it was infrastructure.

The scope of this argument matters. It applies to personal sites, creative experiments, link directories, essays, and archives that grow over years. It does not extend to payment flows, account systems, emergency information, or anything where a visitor's plans depend on a complete and reliable core journey. Those still ship whole.

What the Striped Bar Told Visitors in 1997

From roughly 1995 through 2003, hand-built personal and hobbyist pages carried a recognisable status cluster: one small construction image, close to one to three navigation items that visibly led nowhere yet, and a sentence naming the next planned page. "Photo gallery coming once I get the scanner working." That sentence did more interpretive work than the animation beside it.

Consider what a visitor could infer. An empty navigation label with a construction marker signalled intent. The same empty label without one signalled abandonment, or a mistake. Early browsers gave no error state for "this exists in the author's head," so authors invented one out of clip art.

The second effect was on the authors themselves. Visible incompleteness collapsed the distance between making something and publishing it. A single essay, a list of nearly twenty links, one experiment with animated GIFs — each could go live under a banner that said, in effect, more is intended. The convention was widespread enough across personal homepages of that window to be read fluently by strangers, which is what made it function.

The Striped-Marker TestRemove the construction graphic and ask what remains. If the page still offers a visitor a working link, a readable list, or a usable essay, it is a draft worth publishing. If nothing survives the removal, the marker is covering an empty room.

Room-by-Room Publishing Against the Fixed Launch

Two publishing states are worth comparing directly. The fixed launch withholds every page until the whole structure stands, then releases it in one motion. The live draft exposes one functioning resource and lets the rest arrive in public.

For the creator, the live draft answers a question the fixed launch cannot: does anyone actually use this? One working directory page, revisited by a handful of readers over a month, is stronger evidence of value than a fully designed site that nobody has yet seen. Momentum follows evidence. So does the willingness to abandon a project that turned out to interest only its author.

For the visitor, the payoff is orientation. A candid status block can name three things in two to four lines for a straightforward setup: what works now, which section is still thin, and what is planned next. That is enough for a reader to decide whether to bookmark, contribute, or leave. Where a project is genuinely active, revisiting that block every seven to fourteen days in everyday practice keeps promises from going stale without pretending to daily development.

What an Unfinished Page Still Owes the People Who Visit It

The objection is fair and should be stated at full strength: unfinished sites waste attention, hide broken journeys behind cheerful notices, and let "coming soon" become a permanent alibi for neglect. Anyone who has clicked in the neighborhood of four dead navigation items in a row has earned their scepticism.

The response is a distinction, not a defence. Missing annotations, unwritten category pages, and a thin about section are harmless incompleteness. Broken primary navigation, controls that keyboard users cannot reach, careless handling of submitted data, and descriptions that overstate what the site contains are failures, and the construction notice does not absolve any of them.

Before publishing, walk the primary path: home page to the useful resource and back again, checking keyboard focus, descriptive link text, and the page at narrow mobile width. Show an exact reviewed date only if that date will be maintained. If no review happens within fourteen to thirty days, delete the near-term promise and leave a neutral work-in-progress statement in its place.

The One-Room BoundaryUnfinished annotations and empty category pages may stay public. Broken primary navigation, inaccessible controls, unsafe data handling, and unsupported claims do not enter the house.

Publish the Smallest Complete Room First

The release sequence is ordered by what a visitor gets, in this order:

  1. One usable page — a list, an essay, a tool, something whole on its own.
  2. One status sentence saying plainly what is and is not finished.
  3. One working navigation path in and back out.
  4. One named next addition, plausible enough that it will actually appear.
  5. One feedback route, added only after deciding who reads it.
Image showing sequence

Calm Status WordingWrite it flat: "This directory is usable now; annotations and category pages are still being added." Skip countdown timers, launch dates nobody can guarantee, and the word imminent.

Keep the marker inside the page header or the navigation, where it informs without blocking. A splash screen that stands between a reader and a working link has inverted the entire purpose.

Sunday evening, a bedroom desk, a laptop balanced on a stack of library books. Between seven and nine, a student publishes twelve research-tool links, every destination opened and checked first. The descriptions are half-written and the category pages do not exist yet. Under the title sits one line: twelve links work, notes still coming. By Monday morning someone has already used three of them.