--- name: platform-decision description: The site is block-based WordPress on Kadence, not Vercel/Next.js — decided because of who can change the site without a developer; the short reply sent to the client after 2026-07-30 concedes Vercel is not slower and not worse for SEO/AI search; the Kadence lock-in disclosure exists only in the unsent long draft. No third-party prices. metadata: type: project --- # Platform: block-based WordPress on Kadence, not Vercel/Next.js Source: the platform-decision reply sent to the client, digested in `notes/client-correspondence.md` under "The platform question (Vercel or WordPress)". Two drafts exist (long and short); the transcript of the original planning session shows the short one was prepared at Rian's request specifically to send, after the long one, and the sources indicate the short one is what went out — which version was actually sent is to be confirmed by Rian. Both drafts are kept outside the project because the long one quotes a third-party price. The reply was written after the proposal went out on 2026-07-30; the exact send date is not in the sources (the drafts carry no date and the email thread does not mention Vercel) — Rian confirms. Closed by the client's acceptance of the WordPress proposal on 2026-08-13. Staging on this server is already a Kadence template clone (Kadence 1.5.2 parent + `kadence-child` 1.0.0; Kadence Blocks, Kadence Blocks Pro, Kadence Pro active). ## The deciding reason: who can change the site Vercel is a hosting and deployment platform for custom-built front ends (usually Next.js); it does not store content or give the team anywhere to edit it, so the real comparison is a custom Next.js application versus block-based WordPress. On a custom build, adding a landing page or restructuring a section routes back through a developer every time. On block-based WordPress the client's team and AISV do it themselves. AISV's plan runs on a weekly publishing cadence with execution resting largely on one person (their own deck says so), and their Phase 01 describes drafting posts directly into WordPress via OpenForge. That difference compounds; it decided it. ## Said to the client (short reply) — never contradict these in front of the client - Vercel/Next.js is **not slower**. - It is **not worse for SEO or AI search**. - It solves problems the client does not have yet. - The place a Next.js app genuinely fits the roadmap is the interactive Smart Assessment AISV wants to build (the AISV deck's Creative Concept 03, "Smart BHA"), which is an application, not a page, and can be added later without redoing the site. - PlusROI can work with Vercel, but it would be a different project with a different shape, requiring a requote rather than an adjustment to the current proposal. - The client should confirm the platform question with AISV as well. - Closing ask: if someone has put a specific Vercel proposal in front of them, send it over. ## In the unsent long draft only Not said to the client unless Rian confirms the long version went out. Still true, so do not argue the opposite either: - A well-built Next.js site has a higher performance ceiling than WordPress. - Headless WordPress is not immature. - A static Next.js site has a smaller attack surface; WordPress needs ongoing core and plugin updates. - Vercel is not expensive (the draft quoted a third-party plan price; the figure is omitted here by the project's pricing rule). - Schema and SEO capability are platform-neutral; Next.js and WordPress both server-render. - The three-option framing: A block-based WordPress, B headless WordPress + Next.js on Vercel, C Next.js on Vercel + a different CMS. Option B keeps the editor but makes layout changes and new page types developer work, and the client's licensed Search & Filter Pro and Gravity Forms do not carry over cleanly. ## The lock-in (not yet disclosed to the client) Stated in the long draft but not in the short reply sent: Kadence Blocks is not neutral either. Pages built with it carry Kadence-specific markup, and if Kadence were ever dropped those layouts would need rebuilding. Posts would not, because they are clean standard content. That is meaningfully better than the current Beaver Builder situation (every page locked in proprietary builder data) but it is not zero. Until Rian confirms otherwise, treat the client as NOT having been told this. Disclose it at the mockup review call; Rian confirms when and how. ## How to apply - Build page layouts as unsynced Kadence patterns whose inner blocks are declared editable (`/srv/projects/standards/wordpress.md`), so the client and AISV can change them unassisted — that is the promise the decision rests on. - Keep posts as standard blocks / classic HTML, never wrapped in Kadence layout blocks, so the lock-in stays confined to pages. - Keep the child theme minimal; no custom PHP templates for anything expressible as a pattern. - If the client raises Vercel again, reuse the long draft's three-option framing (A block WordPress, B headless WordPress + Next.js, C Next.js + another CMS) and ask where the question came from — the reply sent closed by asking for any specific Vercel proposal they had been shown.