跳到正文
DEV Community· Wbcom Designs 💡·· 4 小時前AI 評分47

面對數百個舊頁面建置器網站,代理公司應用 AI 重建、繼續維護,還是放棄?

An agency with hundreds of old page-builder sites: rebuild them with AI, keep maintaining, or let them go?

AI 導讀

面對數十至數百個舊頁面建置器網站,代理公司應按商業價值分流,決定維護、重建或封存。仍有價值且狀況良好的網站可繼續維護並限制編輯;重要但脆弱的網站可規劃重建,無人使用的則可考慮退役。AI 可加快重建,但無法決定誰擁有和編輯成果,這點應先釐清。

正文

Wbcom Designs 💡

There's a conversation we keep running into, and it rarely sounds the same twice.

An agency that has been building sites for well over a decade, with a long tail of client sites made on an older page builder, is moving new work to AI-written code on a hosted database and wondering what to do with everything else. Another agency shares its numbers after swapping its freelance WordPress stack for an AI app builder and is broadly happy, while the replies ask who maintains the result in year three. Designers on other builders describe clients who keep overwriting post templates and breaking layouts. Someone else swapped one builder for another and watched rankings dip while the structured data caught up. And in long threads about sites leaving WordPress, a recurring observation is that many of them aren't moving to another CMS at all. They're being rebuilt as something custom, often with AI help.

Underneath all of it is one portfolio question: if you're carrying dozens or hundreds of older builder sites, what do you actually do with them?

The unpopular opinion

Here's the position we hear argued, and it's worth taking seriously:

Most of those old sites don't need a rebuild. They need an owner.

Not a new stack, not a migration project. Someone who knows what's on each site, keeps it patched, and tells the client honestly when it's time for something new.

Plenty of people disagree. Let's give both sides a fair hearing.

The case for rebuilding

  • Builder weight is real. Older builder sites can carry heavy markup, shortcode leftovers and plugin chains that make every update a small gamble.
  • Clients break things. When editors can drag any element anywhere, layouts drift. Some agencies spend more time repairing client edits than doing new work.
  • AI has changed the maths. A rebuild that used to take weeks can now take days. If it's that cheap, why keep patching something fragile?
  • One stack is easier than five. An agency that standardises on one approach can document it, staff it and price it more predictably.
  • Some sites are past saving. Abandoned plugins, a builder version nobody supports, a theme that won't run on a current runtime. At some point, maintaining is just postponing.

The case for keeping and maintaining

  • Working sites have hidden value. Years of content, search history, forms that feed someone's sales process. A rebuild resets a lot of that, and the dips are real even when they recover.
  • Custom code needs maintaining too. AI-written code doesn't maintain itself. Dependencies age, hosted services change pricing, and the person who prompted it moves on. "Cheap to build" isn't "cheap to own".
  • Clients lose the ability to edit. Many small businesses chose a builder precisely so they could change a page without calling anyone. A custom rebuild can turn every text change into a ticket.
  • Ownership gets murkier. On a CMS, the client usually holds the hosting and the admin. On a custom app, the repository, database and deployment often sit in the agency's accounts unless someone deliberately sets them up otherwise.
  • The client didn't ask. A rebuild the client doesn't see value in is a hard sale, and an unbillable one is worse.

The grey areas

This is where most real decisions live:

  • Lock down, don't rebuild. A lot of "clients break things" pain can be solved by restricting what editors can touch: content-only editing, a limited block or module list, locked templates, role-based permissions. It's cheaper than a rebuild and often more welcome.
  • Which sites earn their keep? A portfolio usually splits into a few sites that matter a lot, many that are fine, and some nobody has touched in years. Treating all of them the same is the mistake.
  • Who pays for the decision? Triage takes time. Is reviewing a legacy site part of the care plan, a paid audit, or something the agency absorbs?
  • What happens if the agency goes away? A builder site on a mainstream CMS can be handed to almost any freelancer. A bespoke AI-built app can only be handed to someone willing to learn it.
  • Search and structure. Any rebuild or builder swap needs redirects, structured data and content parity checked before launch, not after rankings move.

Where we lean

We maintain a lot of other people's WordPress sites, so take this with that bias in mind. We lean towards triage by business value, not by technical age:

  1. Sites that matter and are healthy: keep, maintain, and lock down editing where clients keep breaking layouts.
  2. Sites that matter and are fragile: plan a rebuild, on whatever stack the client can actually own and edit, with the accounts in their name.
  3. Sites nobody uses: have an honest conversation about archiving or retiring them, rather than quietly patching them forever.

AI can make the rebuild in bucket two faster. It doesn't answer the question of who owns and edits the result, and that's the part we'd settle first.

Your turn

We'd love to hear from anyone carrying a legacy portfolio, or who's been the client in one:

  • If you have dozens of older builder sites, how did you decide which to rebuild, which to keep and which to retire?
  • For those who've rebuilt client sites with AI-written code: who maintains it now, and has that been cheaper than you expected, or more expensive?
  • Has locking down editing worked for clients who kept breaking layouts, or did they push back?
  • Clients: would you rather keep a site you can edit yourself, or have a faster custom one where changes go through your developer?

來源:DEV Community · dev.to