L1 · Board
What the coherence case counts
What this counts: contradictions stopped before release, how long the estate takes to carry a decided position, what the company has argued in one jurisdiction against another, and the standing rules that constrain every agent before it writes.
Live positions
8
Divergences
12
Caught pre-release
75%
Jurisdictions
9
Scope
The efficiency case and the coherence case
Protecting against a >2% share-price impact was scoped on the observation that comparable sectors have seen 4–15% negative impact following large-scale adverse media events. One avoided misstep at a company of this scale exceeds the entire efficiency case, and is unprovable in advance.
“What is countable: contradictions caught before release, how long the estate takes to carry a position once it is committed, and the exposure of everything currently live.”
Contradictions blocked
0
of 7 found · stopped at a gate
Found after publication
0
each one a defect with a post-mortem
Hours to align the estate
0h
7.0× faster than the 428h baseline
Stale citations live
0
of 669 artifacts · all block at G5
The ledger
Four views of the same question
Is the company saying one thing? Positions answer it against the Canon, the divergence register across markets, the jurisdiction diff on the public record, the risk register before anything is written.
Internal drift
-12 vs 11wIs what we are publishing still what the Canon says?
Last 4w · -2
Every live artifact against the Canon version it cites.
On breach · B4 auto-opens a job for the owning domain agent. Artifacts citing a superseded version are flagged to their owner and block at G5.
Market drift
-4 vs 11wAre two markets saying incompatible things about the same topic?
Last 4w · +1 · reversing
Market A against market B, topic by topic, on the same facts.
On breach · Classified as emphasis, adaptation or contradiction. Only contradiction blocks; the other two are logged and visible.
Perception drift
-7 vs 11wIs what we say what the world believes we said?
Last 4w · +3 · reversing
Published position against measured stakeholder belief in the signal lake.
On breach · Routed to the owning domain agent as a narrative job, not a correction. Perception distance is the slowest to close and the most expensive to ignore.
Twelve weekly readings · three distances
0 = identical · 45 = unrecognizable
We source beef from suppliers who can evidence deforestation-free origin, and we report progress against a verified baseline rather than an aspiration. Where a market has a stricter statutory standard we meet the statute; we do not represent the statutory floor as a global commitment. We do not claim a supply chain is verified where verification covers a sample.
Variance permitted
Local verification percentages, named supplier programs, and the order in which commitments are presented.
Variance forbidden
Any statement that a market-level standard is the global commitment, and any use of "verified" beyond the audited sample.
Time to alignment · G1 commit to the whole bound estate carrying it
Value perception and franchisee pricing discretion
POS-2026-0421 · committed 2026-08-14 09:40
Laggards · —
Menu labeling and nutrition disclosure
POS-2026-0388 · committed 2026-07-30 14:05
Laggards · Canada (gap logged), Italy (filing window open)
Automation and AI in the restaurant
POS-2026-0352 · committed 2026-07-08 11:20
Laggards · Korea (licensee channel, no direct route)
Franchise renewal standards and operator investment
POS-2026-0397 · committed 2026-06-11 08:15
Laggards · Japan (held at G1), 7 IDL licensee markets
Packaging and single-use materials
POS-2026-0403 · committed 2026-05-18 16:50
Laggards · India (corrected), Mexico (channel not reached)
Beef sourcing and deforestation-free supply
POS-2026-0412 · committed 2026-04-02 10:00
Laggards · 5 IDL licensee markets on quarterly cadence
Time to alignment is the one coherence measure that is not a proxy. It is the interval between a position being decided and the estate actually carrying it, and before the platform it was measured in weeks because nobody could see who had not moved.
The failure mode this discharges
FM3 · Canon drift
The Narrative Canon says one thing, the market-level drafts say another, and no single person reads both. Coherence degrades over quarters.
Detection
Rising distance between published artifacts and the Canon version they cite; rising rate of agents citing superseded Canon versions.
Control
B4 runs continuously against published output, not against drafts. Any artifact citing a superseded Canon version is flagged to its owner automatically and blocks at G5.
The instructive case is the one everybody recognizes. A sustainability report says one thing about beef sourcing. A European market says something slightly stronger, because the local regulation demands it. A US franchisee association says something weaker, to reassure operators on cost. All three are defensible in isolation and the set of three is not. B4 is the only mechanism in the architecture that sees all three simultaneously — and it is the mechanism that decides nothing, because whether a divergence is drift to be corrected or evolution to be adopted is a human call at G1.
The workflow underneath
Multi-jurisdiction policy campaign — 18 steps, 15 unattended
A single narrative-share report across 114 countries, ~80 licensee markets, corporate media, franchisee communication, crew communication, investor audiences and regulators — produced weekly rather than quarterly, with the contradictions between markets surfaced rather than averaged away.
G1 POSITION — the global policy position, with explicit approval of where jurisdictional variation is permitted and where it is not
SVP Govt Relations + GC
G3 RISK — regulatory, litigation and franchise-relations exposure of the filed positions taken together
GC + SVP Govt Relations
G5 RELEASE — filings authorized per jurisdiction against their deadlines
Market GR leads
Ceiling · A3. Collection, normalization, analysis, narrative-share computation and draft production run unattended. What the board is told is decided by a person.
Against the way it is bought now
Where the scale difference is largest
Multi-jurisdiction policy monitoring and position drafting is bought from public affairs agencies today, market by market, at day rates. The displacement is external spend, not headcount.
How it is bought
Public affairs agencies, market by market, at day rates.
One orchestrator reading one Policy Register across every jurisdiction.
The alternative to this workflow is buying the same analysis 114 times.
How long it takes
Weeks. Months when the consolidation is done by hand.
Half an hour to a current picture; eleven working days from docket detection to counsel sign-off.
The half-hour benchmark is an operator report from a comparable global function, at twenty markets.
How wide it reaches
The top twenty markets, because that is what the budget covers.
Every operating jurisdiction, because marginal cost per jurisdiction is near zero.
Twenty is the current frontier. 114 is the target, and it is the part no agency model can price.
What it catches
Inconsistencies between markets, found by a journalist or a regulator.
Inconsistencies between markets, found at step 10 before anything is filed.
No agency sees more than its own market. That is not a failure of diligence; it is the shape of the arrangement.
What it does not replace
The conversation with the regulator. The relationship with the committee staffer.
The conversation with the regulator. The relationship with the committee staffer.
The judgment about when to fight and when to concede is A0, permanently.