fix(martech): adding customer-golden click-tracking parity harness and improve block's parity by ramboz · Pull Request #872 · aemsites/intuit-erp

ramboz · 2026-08-27T01:11:48Z

What this is

A second, customer-authoritative click-tracking golden (built from 161 real prod
erp.intuit.com beacons the customer shared) plus an un-gameable oracle to validate our
EDS rebuild reproduces it — and to drive the remaining parity work to ≥95% across the board.

This is harness/tooling (under scripts/diff/) plus one block wiring change
(blocks/feature-grid/feature-grid.js). Draft — opening to share progress.

Why

Our existing clicktrack-golden.json was reverse-engineered by us. The customer set is
ground truth, covers 13 net-new pages, and carries the full ~60-field envelope — so it can
validate the whole beacon (not just the 11 DOM-derivable fields) and cross-check our golden.

What's here

2nd golden + transform — payloads-to-golden.mjs turns the customer drop into a golden
(integrity-locked: the oracle re-hashes it and refuses a hand-edited golden).

Un-gameable oracle — fixtures/field-policy.json (single source of truth: every field
is GATED value-match / frozen PRESENCE-only-with-reason / structural-exception; host +
index normalization) + oracle-lib.mjs (MIN-across-axes verdict — overall × event ×
component × field × coverage — so a weak component can't be averaged away).

Gates — customer-oracle.mjs (fast offline loop gate) and stage-parity.mjs +
stage-sweep.mjs (live full-envelope gate).

Supporting — contract-audit.mjs (reconciled the backend contract; adds pseudonym_id,
link_href, link_href_domain), golden-crosscheck.mjs (91.4% agreement vs our
reverse-engineered golden), parameterized parity-gate/gen-sheet/coverage-matrix.

One behavior change — feature-grid CTAs now emit the feature trail (trackAs),
guarded by test/feature-grid-tracking.test.js (real decorate + delegated runtime).

Results

Offline oracle: 88.9% across-the-board over all 153 elements; the only axis <95% is
product_banner — now identified as our highlight block (dark variant) on /events,
currently un-wired for tracking (the one remaining fix).
Live (real Chrome on stage): full ~59-field envelope reproduced; every gated field
matches prod (host-normalized stage→erp, env=prod, consent, site_section,
accordion_item_N index-tolerant); only ui_action open-vs-close differs (capture artifact).

Privacy

Raw payloads + both goldens + sheet stay local (gitignored under fixtures/local/) — they
carry session/visitor ids and campaign codes. The committed surface is counts/field-names
only (gitleaks-clean).

Notes for review

Tooling PR: no user-visible page change beyond feature-grid click-tracking. Feature
preview base: https://claude-golden-tracking-payloads-405475--intuit-erp--aemsites.aem.page/
(branch name may exceed the aem.page DNS-label cap; the change is validated via the scripts +
npm test, not a visual page diff).
Pre-existing test failures are unchanged by this PR.

🤖 Generated with Claude Code

Second, customer-authoritative golden (from 161 real prod beacons) plus an un-gameable oracle for validating our EDS rebuild against it: - payloads-to-golden: transform the customer drop into a golden (integrity-locked) - field-policy + oracle-lib: full-field taxonomy (GATED value-match / frozen PRESENCE-only / structural-exception), host + index normalization, MIN-across- axes verdict, golden hash lock - customer-oracle: fast offline gate; stage-parity + stage-sweep: live gates - contract-audit (adds pseudonym_id, link_href, link_href_domain), golden-crosscheck, coverage-matrix parameterized (--golden/--sheet/--out) - wire feature-grid CTAs to the `feature` trail + real-render test Customer payloads/goldens stay local (gitignored under fixtures/local/); the committed surface is counts/field-names only. Pre-existing test failures unchanged. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

aem-code-sync · 2026-08-27T01:11:51Z

Hello, I'm the AEM Code Sync Bot and I will run some actions to deploy your branch and validate page speed.
In case there are problems, just click a checkbox below to rerun the respective action.

Re-run all PSI checks
Re-run failed PSI checks
Re-sync branch

Commits

248a002 ✅ (latest)
5af82ab ✅
d323a81 ✅
040158a ✅

aem-code-sync · 2026-08-27T01:12:17Z

	Page	Scores	Audits
📱	/
🖥️	/

The highlight callout is prod's banner. Its trail is variant-dependent — the `dark` promo banner (e.g. /events "Register now"/"Schedule a demo") reports `rw_banner`; the default/light callout (e.g. blog "Learn more") reports `product_banner`. Wire blocks/highlight/highlight.js with a variant-aware trackAs + BLOCK.product_banner in the gate, guarded by test/highlight-tracking.test.js (real decorate + delegated runtime, both variants). Closes the product_banner component gap (offline oracle 88.9% -> product_banner 100%, all components now >=95%). Confirmed live: stage /events renders these CTAs in the highlight (dark) block. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

ramboz · 2026-08-27T01:50:38Z

Wired the last identified gap: prod's rw_banner banner = our highlight block. Confirmed live on stage /events (Register now / Schedule a demo render inside <div class="highlight dark block">).

The trail is variant-dependent — both goldens proved it:

highlight.dark (e.g. /events promo) → rw_banner
default/light highlight (e.g. blog "Learn more") → product_banner

Wired variant-aware trackAs + BLOCK.product_banner + a both-variants real-render test. product_banner 88.9% → 100%; all components now ≥95%.

Offline oracle is at 94.1%, now blocked only by the ui_access_point field aggregate — 4 cards carousel-arrow entries (synthetic-gate depth artifact; runtime is correct live) + 4 case-study-header social links (prod double-keys the share row). Both are pre-existing and live-authoritative, not introduced here.

Keep CLICK-TRACKING-COVERAGE-customer.md and CLICK-TRACKING-ORACLE.md local/uncommitted. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

Two click-tracking trail gaps surfaced by the customer golden: - cards: the prev/next arrows + dots live in .cards-controls (sibling of .cards-track), so they resolved to rw_cards_container instead of prod's rw_cards_container|carousel. Wire .cards-controls -> carousel. - case-study-header: the share row emitted a flat social_media; prod nests it under the article hero. Make .case-study-copy the qrc_article_hero root and nest .case-study-share -> qrc_article_hero|social_media (eyebrow/byline stay flat; ToC stays TableOfContents). Gate BLOCK trails + real-render tests (cards-tracking, case-study-header-tracking) updated to match. Customer-golden offline oracle now PASSES (100%; only the live-only link_href/ link_href_domain remain, emitted by the injected tracker at click time). Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

## What Wire click-tracking on the **blog-template article share row** (`.blog-share`), a follow-up to the merged parity harness (#872). ## Why Live-verifying the case-study social fix on stage revealed that the customer golden's `social` share entries (facebook/twitter/linkedin/youtube on regular blog articles, e.g. `/blog/construction/automation-in-construction`) are rendered by **blog-template's `.blog-share`**, *not* the `case-study-header` block — and `.blog-share` was **completely un-wired** (fired no beacon). The offline oracle had modeled these as `case-study-header` (a synthetic-gate false pass); the live check caught the real gap. ## Fix `.blog-share` now carries a **self-contained** trail so it reports prod's `qrc_article_hero|social_media` and survives the mobile↔desktop relocation of the widget: `qrc_article_hero` on the row, `social_media` on a new `display:contents` `.blog-share-links` span wrapping the links. **No visual change** (tracking attributes only). Guarded by `test/blog-share-tracking.test.js`. The earlier `case-study-header` fix remains correct for `/blog/case-study/*` pages. Note: prod is inconsistent here (double-keys these links as standalone `social_media` on some templates and `qrc_article_hero|social_media` on others) — we match the customer-authoritative nested trail. Test URL: https://claude-golden-tracking-payloads-405475--intuit-erp--aemsites.aem.page/events 🤖 Generated with [Claude Code](https://claude.com/claude-code)

aem-code-sync Bot deployed to claude/golden-tracking-payloads-405475 August 27, 2026 01:49 View deployment

ramboz changed the title ~~Customer-golden click-tracking parity harness~~ test(martech): adding customer-golden click-tracking parity harness Aug 27, 2026

Drop local-only harness docs from the PR

5af82ab

Keep CLICK-TRACKING-COVERAGE-customer.md and CLICK-TRACKING-ORACLE.md local/uncommitted. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

aem-code-sync Bot deployed to claude/golden-tracking-payloads-405475 August 27, 2026 14:28 View deployment

aem-code-sync Bot deployed to claude/golden-tracking-payloads-405475 August 27, 2026 15:24 View deployment

ramboz changed the title ~~test(martech): adding customer-golden click-tracking parity harness~~ fix(martech): adding customer-golden click-tracking parity harness and improve block's parity Aug 27, 2026

ramboz marked this pull request as ready for review August 27, 2026 15:34

ramboz merged commit f98fee4 into main Aug 27, 2026
2 checks passed

ramboz deleted the claude/golden-tracking-payloads-405475 branch August 27, 2026 15:35

ramboz mentioned this pull request Aug 27, 2026

fix(click-tracking): wire blog-template article share-row tracking #890

Merged

Navigation Menu

Uh oh!

fix(martech): adding customer-golden click-tracking parity harness and improve block's parity - #872

fix(martech): adding customer-golden click-tracking parity harness and improve block's parity#872
ramboz merged 4 commits into
mainfrom
claude/golden-tracking-payloads-405475

ramboz commented Aug 27, 2026

Uh oh!

aem-code-sync Bot commented Aug 27, 2026 •

edited

Loading

Uh oh!

aem-code-sync Bot commented Aug 27, 2026 •

edited

Loading

Uh oh!

ramboz commented Aug 27, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

Uh oh!

Conversation

ramboz commented Aug 27, 2026

What this is

Why

What's here

Results

Privacy

Notes for review

Uh oh!

aem-code-sync Bot commented Aug 27, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

aem-code-sync Bot commented Aug 27, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

ramboz commented Aug 27, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

aem-code-sync Bot commented Aug 27, 2026 •

edited

Loading

aem-code-sync Bot commented Aug 27, 2026 •

edited

Loading