Glossary

The terms we use, defined plainly and explained honestly.

Anyone can define these words. This page tries to do more: to tell you plainly what each one is, and then what it is actually worth knowing, from watching these things help and fail real organizations. Every term links here from wherever it appears on the site.

Alt text

A short written description of an image, carried in the page’s code rather than shown on screen. It is what a blind visitor hears in place of the image, and what a search engine reads to know what is in it.

Of the working sites we measured that use images, two-thirds leave at least one description blank. To a blind visitor that is the same as none. A page of photographs arrives as a row of silences, and a search engine has nothing to file. Describe every image that means something, the way you would describe the room to a friend on the phone.

Measured in The evangelical church web, 2026.

Canonical URL

The address you name as a page’s real home, stated in the page’s own code, for when the same content can be reached at several addresses. It asks a search engine to treat that one as the original.

The address in your search result is not always the one on your sign: an old name months after a rebrand, or the website builder’s own address. Most sites answer at several addresses without knowing it. Search engines pick one, usually well. Until they do, links pointing at one version do nothing for the other. Type your own address with www and without, then with http and https. All four should land on the address you print.

Client-side rendering

When a page arrives close to empty and is drawn in afterward by the browser, using JavaScript. A person sees the finished page; many crawlers and AI assistants see only the empty shell that arrived.

Google renders JavaScript, though on a delay it does not promise. Most other crawlers and AI assistants do not, because rendering costs money. On one large social platform we measured pages arriving at nearly a megabyte with fewer than fifty readable characters inside. Whatever a stranger has to know, who you are and where and when, belongs in the page as delivered, structured data included.

Measured in The evangelical church web, 2026.

DNS

The internet’s address book. It translates your domain name into the server your site lives on, and it decides where your email goes too.

Your domain, your hosting and your DNS are often three companies and three logins, and almost nobody has written down which is which. Whoever controls it can move your website and your email in ten minutes, which is what a hijacker does first, and we have seen it happen. Write down which company holds your domain, DNS and hosting, keep all three in the organization’s name, and make sure more than one person has the keys.

Domain name

The address people type to reach you, like yourname.com or yourname.org. It is rented, not owned, and renews on a date.

A domain feels permanent and is not. It is rented one renewal at a time, and the day the card on file expires or the person who set it up moves away, the address can lapse or fall into a stranger’s hands. We have watched a single missed renewal hand an organization decades of its own identity to spammers. Know who holds it, know when it renews, and make sure a second trusted person knows too.

Heading structure

The nesting of headings on a page, like an outline: one H1 saying what the page is, H2s for its sections, H3s inside those. Screen readers navigate by that structure, and search engines read it as one signal of what the page is about.

Every website builder puts Heading 1 in a dropdown beside the font sizes, so people pick a level because it looks right. Almost a third of the working sites we measured have no H1, and more than a quarter have several. Your headings alone should read like a table of contents for the page. Use exactly one H1.

Measured in The evangelical church web, 2026.

HTTPS

The secure version of a web connection. It encrypts what passes between a visitor and your site, and browsers stamp "Not secure" on sites that lack it.

HyperText Transfer Protocol Secure: the S is the whole of it. A browser that stamps "Not secure" on your site is telling the truth, and the fix has been free for years. HTTPS used to be for banks. Now it is the price of being believed: ninety-six percent of the working sites we measured have it. If your site lacks the S, this is the most important thing to fix.

Measured in The evangelical church web, 2026.

Language attribute (lang)

A declaration at the top of a page’s code saying which human language it is written in. Screen readers use it to choose a voice and a pronunciation; browsers use it for spellcheck, hyphenation, and whether to offer a translation.

Set it wrong and a screen reader reads Spanish aloud with English phonetics, which is close to unusable. It is one word in one place. But it sits where nobody building their fourth English site ever had to look, so pages written entirely in Korean or Spanish still declare themselves English. Search engines ignore it and read the words instead. Declare the language the page is actually written in.

Link rot

When links stop working because addresses change, sites move and accounts lapse. Nothing about the link looks broken until somebody clicks it.

Nobody owns the far end of a link. Whoever published it never learns it broke, and whoever moved has no reason to think anyone was pointing at them. So the page goes on promising something it cannot deliver, and the visitor who follows the link gives up on you. We followed 3,651 links in three directories; one in ten led nowhere. Check yours on a schedule, because nothing else will.

Measured in The evangelical church web, 2026.

Meta description

A short line you write about a page, offered to search engines to show beneath your title in search results. They write their own if you do not, and often even if you do.

Just over half the working sites we measured have one. It is not a ranking factor. What it buys is the one line of your search result you get to write, and a real invitation earns a click that a scraped sentence does not. Write it as a welcome, not a summary, and start with the pages a stranger lands on first.

Measured in The evangelical church web, 2026.

Mixed content

When a secure (HTTPS) page loads an image, font, or script over an insecure (HTTP) connection. Browsers block the risky kinds outright and stop treating the rest of the page as fully secure.

One insecure file is enough: an image or embed still pointing at an http address. Thirty percent of the working sites we measured carry at least one, which is the usual reason a single page is downgraded while the rest of the site is fine. If it is a picture pasted in from elsewhere, uploading your own copy fixes it. The rest is a developer’s job, and now you can say precisely what is wrong.

Measured in The evangelical church web, 2026.

noindex

A single instruction in a page’s code, or in the reply from its server, telling search engines not to list it. Nothing about the page looks any different, and the page itself goes on working perfectly.

A site with this on is invisible in search, however good it is. The traffic that never arrives leaves no trace. Every website builder offers the switch for use while a site is being built, but we found forty-two finished sites still carrying it. Check the day you launch, and once a year after.

Measured in The evangelical church web, 2026.

Open Graph

A few lines of code in a page naming the title, description and image to use when somebody shares your link. Most apps read them to turn a pasted address into a preview card with a headline and a picture.

A shared link is a recommendation, and the preview is the part of it you control with Open Graph tags. Plenty of sites never set it, so the platform improvises: a stray navigation word as the title, or a gray box. X prefers its own Twitter Cards but falls back to these, so one set covers both. Two honest limits: no ranking effect, and previews are cached, so today’s fix will not repair last week’s share. Send yourself a link. What arrives should be what you would have chosen.

Parked domain

A web address that still answers but has nothing of yours behind it: a hosting company’s placeholder, an advertising page, or a notice that the account has closed. Everything about it works. None of it is yours any more.

Every layer of the request succeeds, so no status code will tell you and no link checker will either. Only reading what is on the screen settles it. It is the failure that looks most like nothing being wrong, which is how these stay on directories for years. If it is your address, find out whether the hosting lapsed or the registration did. You can rebuild your site, but the domain can be difficult or impossible to recover.

robots.txt

A plain text file, always at yoursite.com/robots.txt, telling search engines and other crawlers what to read and what to skip. A mistake here can stop them reading your site at all.

It is a request, not a fence: well-behaved crawlers obey it, the rest ignore it. One line, Disallow: /, tells every crawler to skip the whole site. Nothing disappears at once; your listings go stale and nothing new is ever read. This is also where a site says which AI assistants may read it, and most people have never opened the file their builder shipped. Yours is worth a look, if only for that one line.

Security headers

Instructions a site sends along with each page that tell the browser how to keep visitors safe: no insecure connections, no other site wrapping yours, no scripts from strangers. Most are a single line, and they close off whole classes of attack.

This is the invisible work: nobody sees it, no one thanks you for it, and it matters only on the day something goes wrong. Most sites ship none of it, because it is no one’s job. Most are one line and safe to add yourself; the Content Security Policy is not. Set it wrong and it blocks your own fonts, forms and payments, and nobody knows but the person who could not pay you. Leave CSP to a professional.

Sitemap

A file listing the pages on a site, usually at yoursite.com/sitemap.xml, offered to search engines so they can find every page. Most website builders generate one without being asked.

Eighty-four percent of the working sites we measured already have one, because the platform made it. Type yoursite.com/sitemap.xml and see yours. It will not lift your rankings, whatever you are told: search engines find pages by following links, and on a small, well-linked site they will find them anyway. It earns its keep on brand-new pages and pages nothing else links to. Better to have one than not.

Measured in The evangelical church web, 2026.

Structured data

A small block of code in a page that states what you are, where you are and when you are open, in a form search engines read without guessing. Those facts can then appear in search results themselves, not just a link.

You will hear it called schema markup, after the vocabulary, or JSON-LD, after the format. Near enough the same job. It is not a ranking trick; what it buys is eligibility for the richer results a search engine can show. Half the working sites we measured publish some. Six in a thousand state what, where and when. Your platform decides which, not you, so ask yours what its markup actually says.

Measured in The evangelical church web, 2026.

Web hosting

The service that stores your website’s files, data and code, and serves pages to visitors. It is separate from the domain name: the domain is the address, the hosting is the building it points to.

Some hosting is bare floor space: the company keeps the building standing and everything inside it is yours, which is how a site can sit unpatched for a decade with nobody assigned to it. An all-in-one website builder is the other extreme, owning the whole stack and keeping it current, but you cannot take the keys with you. That is the tradeoff, and it is worth making on purpose. Ask either kind what happens the day the site is wiped. Own your domain, content and data. Rent the machines.