Change it once

Matt Mullenweg, who co-founded WordPress, posted a short note recently about a long article by Jono Alderson. He was making the case for keeping WordPress. The part worth saying in plainer words is this:

AI can regenerate every page on your site. What it doesn't regenerate is the system that connects them. An author's bio, a product's availability, a service URL: each appears in profiles, listings, structured data, sitemaps, feeds, and paginated archives. Without a shared record and dependency tracking, every edit leaves stale copies behind. Static deployment makes it worse, because edits never reach HTML already built.

Matt Mullenweg

Jono wrote the long version for people who build websites. Here's the same idea as a phone number.

One phone number

A clinic changes its phone number. That number sits in the header, in the footer and on the contact page. It also sits in the structured data, a small block in the page that tells search engines the name, the address, the hours and how to call. A customer never sees that block. Google does.

On WordPress, done properly, the number lives in one field. The header, the footer, the contact page and that block all read the field. You change it, you press Update, and the next visitor gets the new number. You don't go looking for stray copies.

Plenty of themes don't do it properly. They paste the number into a few templates and leave it there. You still have copies, stored in the theme files. The fix is one field, and templates that read it. That's ordinary work, and it's the sort of thing I end up doing on sites I look after.

A person, then a product

A practitioner's bio has the same shape. The photo and the qualifications show on the team page, and a shorter line shows under the articles she wrote. On WordPress she is one entry. The articles point at that entry. They don't keep a private biography that goes stale in March. She finishes a course, you edit the entry, and both places catch up, because there was nothing else to update.

A shop does this with stock. A jacket sells out at 11am. The product page, the category list, the structured data and the feed you send to Google all need to say so before someone orders it. WooCommerce stores the stock on the product, and the lists read it. Nobody should have to walk the site ticking boxes by hand.

Renaming a service is where it spreads. Teeth whitening becomes cosmetic dentistry, and the web address changes with the name. If the menu item is the page itself, WordPress picks up the new address when it draws the menu. A link someone pasted by hand keeps the old address, so that one you do edit. The sitemap is built from the pages that exist now, so the new address is the one it lists.

The old address is the bit people forget. WordPress does not add a redirect when you change the slug, so the old URL 404s until a redirect plugin, or the SEO plugin you already run, writes one. I add those. The menu, the category lists, the sitemap and the articles are already asking the page for its current address.

A folder of finished pages

Some businesses are being offered a folder of pages. Describe what you want, get the HTML back, upload the files. For a brochure that changes twice a year I work that way myself, and I wrote up when a CMS is more machinery than a small site needs.

The gap shows on the second edit. Each page was its own document. The number in the footer matched the contact page on the day the files were made, and after that they had no connection. You correct the contact page. The footer still rings the old number. The structured data can too, so Google goes on showing the old one after the page a person reads has been fixed.

Uploaded files add a delay even when you did keep a shared note of the opening hours. The HTML on the server was built last Tuesday, and it still contains last Tuesday's hours, until someone rebuilds the pages that used them and uploads the new files. An ordinary WordPress site reads the current entry when a visitor asks for the page. A cache can hold the old one. A bloated caching plugin will, and I usually remove those.

The pages nobody opens

The leftovers that cause the real trouble are the ones a business owner never looks at. Page two of a person's articles. The old address after a rename. The feed a subscriber still has in an app. The sitemap line for a page you unpublished this morning, which should drop out at the same time. A rebuild that waits until midnight leaves the old URL listed for the rest of the day.

Jono follows an author who changes their surname, through every corner that has to come along. Worth reading if you want that level of detail. For running a business, here's the test I use. Change one fact. Then look at the footer, another page, the sitemap, and whatever Google is showing for the number or the hours. If two of them disagree, the site is keeping copies.

WordPress has had more than twenty years to grow the boring parts of this. Authors, menus, categories, feeds, sitemaps, stock levels, archives that shrink when a post comes out. A lot of it is easy to configure badly. I spend plenty of the week on sites where a couple of plugins disagree about the sitemap. The idea under that mess is still the right one. The site holds a fact once, and the other views ask for it.

A handful of pages that barely change don't need any of this. If the same phone number, the same person, the same product or the same opening hours have to stay correct in several places, and the person editing them is not a developer, this is the job WordPress is already doing. If your site does it, keep the site. I'll keep it running.

Ciprian Popescu runs WP Dublin and builds WordPress plugins at getButterfly. He has been helping Irish businesses get found online since 2002. More about me.