Source: 26-CANONICAL-METRICS-AND-WORDING.md

What this source is

The style guide for executive wording. This file defines the canonical phrasing for every recurring numerical claim so that slides, docs, and NotebookLM-facing material stay consistent. Per its own governance section: “Any document added to NotebookLM should conform to this file before packaging.”

This source resolves multiple wording ambiguities across the wiki and names the 6 production connectors for the first time (NetSuite, Business Central, Salesforce, HubSpot, ShipStation, Oracle). It also establishes explicit “avoid” lists for older phrasings that should no longer be used.

Read directly from the canonical Preston-Test repo.

Key claims

1. Canonical test wording (3-sentence sequence):
   1. “100% suite pass rate (419 suites).”
   2. “100% of executed tests passed (9,476 tests across unit, integration, and E2E).”
   3. “Full breakdown: 9,286 unit (23 skipped), 170 integration (11 skipped), 20 E2E portal.” This is the canonical breakdown: 9,286 unit + 170 integration + 20 E2E = 9,476 passing. 23 unit skipped + 11 integration skipped = 34 total skipped. 9,476 + 34 = 9,510 total. → production-proof
2. Canonical coverage: Statements 64.48%, Branches 52.34%, Functions 67.15%, Lines 64.59%. Short form: “65% line coverage across 45,757 lines of production TypeScript.” → 45,757 lines is a NEW number — first formally-ingested TypeScript LOC count. → production-proof and preston-test-repo
3. Canonical module wording: “16-module demo library (12 core + 4 extension/platform).” Confirms 22-module-library from another source. Also specifies: “If listing only core modules, explicitly say ‘12 core modules’.” → module-library
4. Canonical connector wording (NEW — 6 connectors NAMED): “6 production connectors demonstrated (NetSuite, Business Central, Salesforce, HubSpot, ShipStation, Oracle).” This is the first formally-ingested enumeration of the 6 connectors. Before this, only NetSuite and Business Central were named (the “dual ERP” pair); the other 4 were unnamed. Now: Salesforce (CRM), HubSpot (CRM + marketing), ShipStation (shipping/logistics), Oracle (possibly ERP or database). → suitecentral-2-overview
5. “Avoid” list for connectors: avoid “4 connectors verified” — that’s outdated phrasing. The current canonical number is 6.
6. Canonical AI provider wording (NEW — 4 providers with per-mapping costs): “4 production-ready AI providers (OpenAI, Claude, OpenRouter, LMStudio).” With cost context: OpenAI GPT-4o at $0.02/mapping,Claude3.5Sonnetat$0.003/mapping, OpenRouter (multi-model, free tier available), LMStudio (local, free). First formally-ingested per-mapping pricing. → production-proof
7. “Avoid” list for providers: avoid “3 providers” or omitting OpenRouter from the list.
8. Avoid absolute timing claims without refresh date: explicitly calls out “Uncontested 6-12 month window” as phrasing to avoid. This directly contradicts ai-governance-layer-video 04:05-04:14 which claims “The market window is 6 to 12 months. The technology is verified, the market is open, and Squire is positioned to capture this value.” The canonical metrics file wants that kind of phrasing avoided unless the source is date-locked. → suitecentral-2-overview (flag this tension)
9. Avoid “No competitor has this” unless source-locked and dated. Favor “Comparable in complexity dimensions” or “Positioned against” or “Based on publicly available sources (date-stamped)”.
10. Competitive pricing wording for Celigo (March 2026 refresh): “Celigo ~$10K-$50K+/yr, roughly $1K-$4K/mo equivalent, scope-based/per-endpoint pricing.” Avoid older “$3K-$10K/integration” and “$4-8K/mo” unless fresh sources support them. → first formally-ingested competitor pricing for Celigo.
11. Celigo dual-ERP framing: describe as “Generic / not dual-ERP-specific” — avoid positive-match framing that implies Celigo is purpose-built for dual-ERP, but also avoid a flat “No.”
12. Reviewer-facing proof routing: prefer reviewer-friendly proof pages over raw API/JSON payloads in executive-facing materials. Raw JSON endpoints should be labeled as technical proof only.
13. Date/decision wording: avoid hardcoded past dates in reusable documents. Preferred: “Decision requested within 10 business days” and “Quarterly refresh required for external claims.”
14. Governance requirement: “Any document added to NotebookLM should conform to this file before packaging.” This source is the style-guide gate for NotebookLM content.

Pages updated by this ingest

Created (1 new page):

• canonical-metrics — a concept page that captures the canonical wording as a reference, especially useful for future ingests that need to check whether their phrasing is current.

Updated (3 existing pages):

• production-proof — added canonical test sequence (9,244 + 146 + 20 = 9,410), all four coverage metrics, 45,757 production TypeScript LOC
• suitecentral-2-overview — named 6 production connectors (NetSuite, Business Central, Salesforce, HubSpot, ShipStation, Oracle), added AI provider per-mapping costs, flagged the “market window” phrasing tension with the hook video
• preston-test-repo — added the 45,757-line production TypeScript total

Notable quotes

“Any document added to NotebookLM should conform to this file before packaging.” — Governance section

“6 production connectors demonstrated (NetSuite, Business Central, Salesforce, HubSpot, ShipStation, Oracle).” — Canonical connector wording

“Avoid unless source-locked and dated: ‘Uncontested 6-12 month window’” — External comparison avoid list

Cross-references / contradictions found

• Market window phrasing tension (mild, important): this source explicitly says to avoid “Uncontested 6-12 month window” wording unless source-locked and dated. But ai-governance-layer-video at 04:05 says “The market window is 6 to 12 months.” The hook video is using phrasing the canonical style guide says to avoid. This is NOT a factual contradiction (the claim might still be true) — it’s a wording governance violation. The canonical guide wants either (a) the claim removed, (b) the claim re-stated with a date anchor, or (c) the claim kept but with “based on sources as of [date]” qualification. Flagged in suitecentral-2-overview.
• “6 connectors” resolves open question: the Talking Points source said “Six production connectors” without naming them. This source names them. → suitecentral-2-overview can now replace the “specifics not yet ingested” note with the actual list.
• “9 AI providers” with costs resolves another gap: previously, the AI provider stack was mentioned generically. Now we have provider names, their models (via 04-technical-proof), and per-mapping costs (via this source). Full stack documented.
• Oracle as a connector (not just a competitor): Oracle appears both as a competitor (ai-governance-layer-video critiques Oracle’s native AI; oracle-comparison compares to Oracle NSIP) AND as the 6th connector in this list. These are not contradictory — SuiteCentral 2.0 connects TO Oracle (as a data source) while differentiating AGAINST Oracle NSIP (as a competitor AI integration layer). Same vendor, different products.
• 45,757 lines of production TypeScript: first absolute LOC count for the codebase. 15-start-here-async-standalone said “2,282 tracked files / ~854K text LOC” — the 854K includes everything (config, tests, docs, generated files). 45,757 is the production TypeScript subset.

Notes

• This source is authored as a style guide, not a pitch. It’s meta — it’s about how to talk about SuiteCentral 2.0. That makes it unusually important for the wiki: it tells future ingests which phrasings are canonical and which should be avoided.
• The source explicitly mentions “March 2026 refresh” on the competitive pricing section — suggesting the canonical wording is versioned and refreshed periodically. Any future ingests touching competitive claims should check this file for the latest guidance.
• The “Avoid” lists are especially valuable — they tell us NOT to use “3 providers,” “4 connectors verified,” “No competitor has this,” “Uncontested 6-12 month window,” “$3K-$10K/integration,” “$4-8K/mo” without fresh dated sources. Saves future ingests from drifting back to older phrasings.
• The “6 connectors” list is production-DEMONSTRATED — not just coded. Distinction matters: there may be more connectors in various states of development, but only 6 are proven and demo-ready.