Placeholder text left in production

Lorem ipsum, in any of its variants. The most recognisable and, oddly, the least harmful — a visitor knows instantly that the page is unfinished.

Developer markers left in copy: `TODO`, `FIXME`, `test test`, `xxxxx`. These usually sit in a field rather than a paragraph, so they surface in a meta description or an alt attribute rather than on the page.

Framework and CMS defaults: `Hello world!` as a first post, `Add a description here`, `Just another site`, the sample page every installation creates.

Commercial placeholders: `Coming soon`, `Content to follow`, on a page that has been live for two years.

Documentation domains: `example.com`, `example@example.com`, left in a contact block or a code sample that was meant to be edited.

These are eleven distinct strings and one problem. Reporting them separately produces eleven findings for one edit.

Meta descriptions and title tags, which nobody reads on the page and everybody reads in a search result. A result reading `Add a description here` is the most expensive version of this defect.

Image alt attributes, which are invisible unless an image fails or a screen reader reads it.

Footer and contact blocks, especially address and phone fields copied from a theme.

Pages deep in a section nobody navigates to — a legal page, an old campaign landing page, the second page of a multi-step form.

Structured data, where a placeholder in a JSON-LD field is invisible on screen and read by every machine that parses it.

The direct cost is credibility. A visitor who reaches a page reading `Lorem ipsum dolor sit amet` on a site asking for payment details draws a conclusion about the rest of the site.

The indexing cost is real too: a page whose substance is filler is thin by definition, and Google's helpful-content guidance describes exactly this — content created to fill a slot rather than to serve a reader.

And placeholders in metadata are the version that reaches people who never visit. A title or description carrying filler is shown in results, and the click is lost before anyone sees the page.

Search the site for the strings directly. A site search for `lorem ipsum`, `coming soon` or `example.com` finds the ones that are indexed; a database or repository search finds the ones that are not.

Check the metadata separately from the body. `document.title` and `document.querySelector('meta[name=description]')?.content` on a sample of pages catch the field-level cases the body search misses.

Include structured data. A search for the placeholder string in the rendered JSON-LD finds what nothing on screen shows.

Do it after every launch and every theme change, which is when defaults get reintroduced.

Make the fields required where the content is required. A CMS that accepts an empty description will keep producing pages with the theme's default in it.

Fail the build. A check for the handful of known placeholder strings in templates and content exports is a few lines and catches the class rather than the instance.

Delete the sample content a new installation creates, on the day it is created rather than after launch.

Never use `example.com` in a live template. Use a value that is obviously wrong to a human reader, so it cannot pass review by looking plausible.

`structure.html_quality` reads the served HTML of each crawled page and reports the known placeholder strings it finds, with the URL and the field they appeared in.

The scanner tags each variant with its own code — lorem ipsum, TODO, coming soon, and the rest — so a report can tell a developer marker from a commercial one, while the page explaining the problem stays a single page.

What it cannot catch is a placeholder nobody has seen before. The detection is a list of known strings, not a judgement about whether a sentence was meant to be replaced.