# Achievement activation: measurement plan

Version: 2026-10-24. Campaign: `achievement-activation-20261021`.

Use this plan to decide which published resources help a prospective buyer choose a useful first scope. This document is public. Store provider exports, lead records, booking records and weekly reports in the private reporting folder outside this site's `site/` and `content/` trees.

## Three independent proof layers

1. **Publication:** the source audit counts working resources and links against commit `34a420cbd78d22ab11ced014f0b4eaf104204875`. Optional live checks compare each deployed response with the local source. This establishes what a reader can open.
2. **Website actions:** Vercel Web Analytics records page views, fixed public-resource clicks, form steps and the existing lead/call/booking-link events. An event-handler test establishes the code path. A provider readback establishes a recorded action.
3. **Business outcomes:** a saved lead, completed calendar booking, qualified opportunity and agreed scope each need their own operational record. Count each stage separately. A resource or booking-link click does not establish completion.

Search Console query/page clicks and impressions describe Google Search exposure. They are independent of on-site actions. Keep page and query exports separate from property totals; row limits and anonymized queries can prevent their sums from matching. Compare like-for-like property, country/device filters, search type, aggregation and closed date windows.

## Fixed campaign links

Use lowercase tags on external distribution links only. Preserve each page's existing clean canonical URL; internal links remain untagged. Never put a name, email, prospect domain, contact message or private identifier into a campaign tag. No campaign parameters are copied into forms, contact payloads or calendar URLs by this implementation.

| Placement | utm_source | utm_medium | utm_campaign | utm_content |
| --- | --- | --- | --- | --- |
| LinkedIn profile Featured: qualifications | linkedin | profile | achievement-activation-20261021 | profile-qualifications |
| LinkedIn profile Featured: practice hub | linkedin | profile | achievement-activation-20261021 | profile-practice |
| LinkedIn profile Featured: expert kit | linkedin | profile | achievement-activation-20261021 | profile-expertise |
| Article distribution | linkedin | organic-social | achievement-activation-20261021 | One of the six article IDs below |
| Qualifications PDF shared externally | linkedin | document | achievement-activation-20261021 | qualifications-pdf |
| Proposal resource shared externally | linkedin | document | achievement-activation-20261021 | One of the three proposal IDs below |
| Expert kit shared externally | linkedin | document | achievement-activation-20261021 | expert-speaker-kit |

Example: `https://suedeai.ai/seo/practice?utm_source=linkedin&utm_medium=profile&utm_campaign=achievement-activation-20261021&utm_content=profile-practice`.

PDF/Markdown resource requests may bypass the HTML analytics loader. A tagged direct file request does not guarantee a Vercel Web Analytics event. Prefer a tagged HTML landing page with a measured file link when a counted visit is needed.

Article IDs: `mcp-tool-discovery`, `mcp-error-contracts`, `canonical-noindex-checks`, `source-live-evidence`, `evaluation-fixtures`, `credential-relationship-checks`.

Proposal IDs: `ai-workflow-proposal`, `search-audit-proposal`, `answer-evaluation-proposal`.

## Event contract

Markup uses `data-activation-artifact` and `data-activation-action`. The script checks the identifier against an exact public path, validates the destination origin and rejects mismatched or arbitrary destinations. The registry is in `site/assets/analytics.js`; the source audit has a matching registry guarded by a test.

| Event | Trigger | Properties | Meaning |
| --- | --- | --- | --- |
| `activation_evidence_click` | An allowlisted proof, article, case or kit page link | `artifact`, `page` | Reader chose to open a public resource |
| `activation_demo_click` | Allowlisted `/demonstrations` link | `artifact`, `page` | Reader chose a demonstration |
| `activation_qualification_click` | Allowlisted `/qualifications` link | `artifact`, `page` | Reader chose the qualifications page |
| `activation_proposal_click` | `/practice#proposals` link | `artifact`, `page` | Reader chose the proposal section |
| `activation_download_click` | Allowlisted PDF or Markdown link | `artifact`, `page` | Reader attempted to open the file |
| `Assessment click` | Existing `#scope`, `/#scope` or `#assessment-request` link | `page` | Scope/contact intent; one legacy event per click |
| `Booking click` | Existing Google Calendar link | `page` | Outbound booking intent |
| `Phone tap` | Existing telephone link | `page` | Call intent |
| `Lead` | Existing successful `suede:lead` signal | `form`, `page` | Form success signal; confirm delivery in operational records |
| `form_view`, `form_start`, `form_submit_attempt`, `form_error` | Existing website-review form lifecycle | `form`, `page`, fixed `reason` on errors | Form diagnostics, counted independently of `Lead` |

Resource-click properties never include the destination URL, query string, link text, visitor identifier, form content or client name. `page` is the current pathname. The existing URL filter removes non-UTM query parameters before page addresses reach Vercel. No identity stitching or new analytics vendor is introduced.

Each click produces at most one resource/legacy click event. Repeated loading of the analytics file does not bind another listener. Deliberate repeated clicks remain separate actions. `?qa=1` prevents the analytics loader and all its listeners from running. Preview/manual QA uses that parameter; it cannot validate receipt by the analytics provider because tracking is excluded.

## Baseline and weekly report

Capture the baseline before distribution, using the last seven complete days in `America/New_York`. Run weekly on the same day. Retain the absolute start/end dates, provider timezone, extraction timestamp, property/project, filters, row limits and provider freshness. Label any current-day extraction as partial. Keep known internal visits visible in methodology if they were not excluded at collection time.

Report this sequence as separate aggregate columns: referring source/campaign → resource page views and artifact clicks → scope clicks → `Lead` signals → operationally confirmed leads → completed bookings → qualified scopes. These are adjacent stage counts; cookieless aggregates do not prove that the same person moved between stages. Do not manufacture a joined user funnel or add conversions across sources.

Use `unavailable` for inaccessible data and `not_yet_observed` only when a successful, fully specified query found no records. A verified zero must name the queried scope and date window. Do not fill missing values with zero. Rates require a compatible denominator and the same window; do not divide Search Console clicks by website leads to assert an individual conversion rate.

Public-safe schema for a **private** weekly report:

```json
{
  "schema_version": 1,
  "campaign": "achievement-activation-20261021",
  "extracted_at": null,
  "window": { "start": null, "end": null, "timezone": "America/New_York", "complete_days_only": true },
  "search_console": {
    "status": "unavailable",
    "property": null,
    "filters": null,
    "property_totals": null,
    "page_rows": null,
    "query_rows": null,
    "row_limit": null,
    "freshness": null
  },
  "website": {
    "status": "unavailable",
    "project": null,
    "filters": null,
    "source_campaign_rows": null,
    "resource_pageviews": null,
    "artifact_clicks": null,
    "scope_clicks": null,
    "lead_signals": null,
    "booking_link_clicks": null,
    "internal_traffic_notes": null
  },
  "operations": {
    "confirmed_leads": { "status": "unavailable", "count": null, "evidence": null },
    "completed_bookings": { "status": "unavailable", "count": null, "evidence": null },
    "qualified_scopes": { "status": "unavailable", "count": null, "evidence": null }
  },
  "interpretation": { "observations": [], "hypotheses": [], "next_action": null, "causal_claim": false }
}
```

## Inspect the public publication record

The case study names the baseline commit and records the original repository comparison date. Open the practical resource hub and check the linked proposal templates, six articles and speaker kit against the named units in that table. The source comparison itself was run inside Suede AI’s repository; the public resources let readers inspect the published deliverables.

For your own site, retain the same inputs: named units, source paths, input hashes, baseline revision and current revision. Compare the complete deployed responses with an authorized checkout before confirming publication. A non-200, blocked request or source mismatch is an incomplete live check. Provider settings, consent configuration and analytics retention are unchanged by this source implementation.