The Content Plan — how the 15-action plan is composed and personalised

Plain-English briefing on My Plan: what the customer sees, how the engine decides what goes on it, and why no two clinics get the same plan. Written 24 July 2026. Last verified: 3 September 2026 (Agent 11 — author-input survival (#2845/#2846) and email-link redirect (#2847) folded in).

What the customer sees

Every clinic gets My Plan: a single page listing 15 prioritised actions, each one a concrete thing that will improve their AI visibility. Actions are grouped into named sections (the internal "tiers"), each with plain-language explanation of why it matters. Most actions carry a Create Content button that kicks off the generation pipeline directly — the plan isn't advice, it's an executable queue.

The four tiers (what kinds of action exist)

Tier 1 — Your People. Complete clinician profiles: photos, credentials, registration numbers, professional bios, niches. This tier exists because AI trust is anchored in identifiable experts — a clinic's practitioners ARE its visibility foundation. The plan itself is gated on profile depth: content tiers unlock once enough of the team reaches profile completeness, which is deliberate — publishing content credited to empty profiles would waste the content.

Tier 2 — Foundation Health. Structural essentials: create the missing hub (pillar) page for an active topic cluster, or refresh an existing page scoring below the quality threshold. These are the load-bearing pages everything else links back to.

Tier 3 — Content Plan. The core: create the supporting pages (spokes) for the clinic's active topic clusters — condition explainers, treatment pages, service pages, blog posts, each answering one specific patient question — plus content targeting the clinic's declared priority topics. This is where topical authority is built.

Tier 4 — Refresh & Maintain. Existing published pages that are ageing, slipping, or have been overtaken — where refreshing beats starting again because the page already has earned authority. Refresh and improve actions are just as executable as create actions: the "Write it for me" button works on them too.

How the engine decides what makes the 15

This is the part that separates the platform from a content-ideas generator.

The engine works with content candidates — every possible action it could recommend for this clinic. Far more candidates exist than plan slots, so the engine's job is choosing the 15 that matter most right now.

Candidates come from evidence. Every candidate originates in something measured: a topic cluster activated by a practitioner's declared niche, a page the assessment scored below threshold, a topic the clinic declared as a priority, a gap in a cluster's coverage.

Candidates compete on merit. A handful of slots are reserved for the highest-value structural items, and a clinic's confirmed practice priority topics get reserved slots too (up to three): a confirmed priority not already covered by the plan is guaranteed its own plan item, carrying a priority badge. The rest is open competition. Every candidate gets a computed score: a base relevance score multiplied by a set of transparent multipliers — cluster boost (is this part of an active, strengthening topic cluster?), niche match (does it match a practitioner's declared expertise, and how strongly?), practitioner factor, declared-priority boost, and for refreshes an authority factor (an existing page with earned authority is worth more to refresh than a new page is to create, all else equal). The 15 highest-scoring candidates win, regardless of tier.

The scoring is transparent. On the admin side, every plan row shows its full derivation: base score, each multiplier, the combined multiplier (capped so no single factor dominates), and the final score — with an expandable panel showing the arithmetic. Nothing is "the algorithm said so."

And it doesn't stand still. The candidate pool is re-evaluated as reality changes — pages get published, scores move, clusters strengthen, priorities shift — so the plan the clinic works from always reflects where they are now, not where they were at signup. How and when the plan refreshes is covered below.

Why no two clinics get the same plan

The plan is personalised by construction, not by templating:

• The clinic's people drive the topics. A practitioner declaring "Vestibular Rehabilitation" as their niche activates that topic cluster for the clinic — and only clinics with that expertise get that cluster. The plan literally cannot recommend content the clinic has no one qualified to author.
• The clinic's declared priorities shape ranking. Up to three practice priority topics get research treatment and a scoring boost.
• The clinic's existing site shapes everything. What already exists, what scores well, what's thin, what's missing — all from the assessment's page inventory. A clinic with 54 blog posts and zero condition pages gets a very different plan from a clinic with the opposite shape. (Legal and boilerplate pages — privacy policies, terms and the like — are deliberately excluded from the customer-facing page lists; they're not content.)
• The customer shapes it too. Any suggestion can be marked "Not relevant" — the system asks briefly why (a short list of reasons, each with an optional "anything you'd like to add?" box; whatever's written is kept and visible to the support team), and that feedback genuinely teaches the plan: some reasons also clear similar suggestions from the same topic group, and the list refills with the next-best items. The dialog itself now explains what rejection does (removed and not suggested for 90 days; no instant replacement; the plan rebuilds when it runs low; a topic may deliberately resurface after the 90 days if practice factors point to it again; clearer reasons tune the algorithm — Tor-approved copy, live verbatim, #2703). Rejection suppression is 90 days per page type, and the rejection record itself is never deleted. Since 27 Aug 2026 (#2703): plans REFILL to the full 15 items at every rebuild (rejected and completed topics excluded, next-best eligible topics fill the gaps — plans no longer shrink over time), and plans top themselves up automatically — when completing or rejecting leaves 5 or fewer actionable items the system quietly rebuilds in the background (at most once per 24h per practice; customer keeps their current plan meanwhile and may briefly see "Adding fresh suggestions…"). Support answer: keep working top-down; when your plan runs low it tops itself up. The plan page also carries a short note explaining ordering: preference rankings feed in, but they're one of more than thirty factors the plan weighs (real search volumes, what's already on the site, what local patients search for), so preferences shape the order without overriding the evidence.
• The clinic's country changes the content itself (see the research briefing) — and country-specific compliance and funding context flows through to what's planned and how it's written.

Keeping the plan honest over time

The plan stays current without churning. The platform continuously tracks what's changing — pages published, scores moving, clusters strengthening, profiles deepening — and when enough has genuinely changed, it invites the customer to refresh: a Regenerate Plan button arms, with a plain-language explanation of what changed ("you've completed three actions, your Headaches cluster crossed a health band"). One click and the plan recomposes around the clinic's new reality, with visible progress while it works — regeneration completes (or, if something goes wrong, says so plainly) within about fifteen minutes; it never leaves the customer staring at a spinner. If a refresh fails or stalls, an honest "regeneration failed — try again" notice appears with a retry button, and the plan shown below is clearly labelled as the last successful version — never silently presented as current. The balance is deliberate: the plan is always up to date with what matters, but it doesn't reshuffle on every page load — a plan that constantly rearranges itself teaches the customer to ignore it, while a plan that says "three things changed, worth regenerating" earns trust and keeps momentum.

Progress is visible. Plan cards reflect live generation state (researching / writing / ready), published items leave the "what's next" queue (the plan is a to-do list, not a trophy cabinet — published work lives in Content Studio), and cluster health headers show topical progress.

Author input survives plan changes. Answers a clinician submits on a plan item are saved against that item and SURVIVE plan refreshes and regeneration. The plan card shows a green "Author insight received · Ready to write" badge with a read-only "View submitted responses" view, and writing starts from the saved answers via Write it for me. The "Author input received" notification email now shows the real article title and deep-links straight to the focused My Plan item, and the link works through magic-link sign-in (#2845/#2846, live 30 Aug 2026). Relatedly, a customer who lands on an internal Co-Kinetic screen (for example from an older notification email's link to the site root) is now redirected to their own My Plan instead of seeing an admin page; the sign-in flow is unchanged for logged-out visitors (#2847, live 31 Aug 2026).

Common questions this briefing answers

• "Is this just a list of blog ideas?" No — every action originates in measured evidence, is scored for marginal impact on this clinic's visibility, and is executable directly from the plan.
• "Why does my plan want profile completion before content?" Because AI search discounts anonymous content. Practitioner identity is the foundation; content credited to complete, verifiable profiles is worth multiples of the same content credited to no one.
• "Does my plan stay up to date?" Yes — the platform tracks everything that changes and invites you to refresh the plan the moment enough has genuinely moved, telling you exactly what changed. It updates when it matters, without reshuffling for the sake of it.
• "Why these 15 and not others?" Every action on the plan explains in plain language why it's there — the cluster it strengthens, the expertise it builds on, the gap it closes (a "Why this action?" expander appears where there's more detail than the card itself shows). Behind that, every candidate is scored on measured merit, and the full scoring derivation is maintained for every plan — so if a customer ever wants to know exactly why one action beat another, the answer exists and can be shown, not hand-waved.
• "I ranked my services by preference — why doesn't the plan follow that order?" Preferences feed in, but they're one of more than thirty factors weighed alongside real search demand and the state of the existing site. No single factor controls the order; the plan's job is to put the highest-impact next piece at the top. A short note on the plan page explains this.
• "Why did my condition topic get created as a blog post?" It won't any more — condition-shaped plan topics are typed as condition pages, and any existing wrongly-typed item corrects itself on the next plan refresh.
• "Do the answers I submit survive plan refreshes?" Yes. Submitted author input is saved on the plan item and survives refreshes and regeneration. The item shows a green "Author insight received · Ready to write" badge, you can reopen your answers read-only via "View submitted responses", and writing starts from those saved answers.
• "How do I get location-specific service pages for the map pack?" They're a different page type from anything currently in the plan and are releasing in September 2026 — nothing to prepare; they'll appear as plan options when they land.