These are the transduction axes. Event integrity, identity resolution, attribution capture, and payload completeness. They are not separable in practice. A broken click ID degrades identity match because fewer user signals arrive. A thin payload caps identity match because there are fewer fields to hash. A missing deduplication key inflates event counts in a way that corrupts the value-per-event ratio the bidding engine is training on. The axes are described one at a time below, but they fail as a system.
The four states inherited from the consent gate apply in multiple ways at every transduction axis. Fully permeable. Modeled. Blocked. Miscalibrated. Not only do these states categorize the function or dysfunction of each axis, the consent state transduced through these axes is carried and compounded. State 4 is the structurally expensive case across all four axes, and it presents differently at each.
Axis 2. Event integrity
A purchase that the browser pixel reports, that the server-side container also reports, without a shared event ID between them, is counted as two purchases. The platform's reported CPA halves. Reported ROAS doubles. The bidding algorithm reads the inflated number as a signal to spend more, scales the budget, and continues training on the doubled count.
Meta's published data: accounts with a properly configured Conversions API setup deliver an average 17.8 percent lower cost per result than Pixel-only accounts. State 3 (blocked) here is the Pixel-only stack losing thirty to forty percent of conversions to Safari ITP and ad blockers, which is roughly knowable. State 4 is the stack that has CAPI deployed, dashboards that populate, and deduplication keys that are weak or absent.
Reported ROAS climbs. Spend follows. The miscalibration accumulates underneath. And if the consent state arriving at the server container is itself miscalibrated (browser respects the user's denial, server container never receives the update), the CAPI event that gets deduplicated is carrying full PII for a user who said no. The deduplication contract holds. The compliance contract does not.
Axis 3. Identity resolution
Meta's Event Match Quality score is the most explicit identity-fidelity benchmark any platform publishes. EMQ measures how confidently Meta can match an event to a real user using the identifiers the brand sent: hashed email, hashed phone, name, IP, fbc, fbp, external_id. Meta's documented "healthy" threshold is 6 out of 10; "Great" begins at 8. Most Shopify stores running Pixel-only land at 3 to 6 on Purchase; properly enriched server-side reaches 7 to 8.5 and above. A low EMQ does not stop conversions from being counted. It stops the optimization engine from learning who converted, which collapses audience quality downstream.
State 3 here is an EMQ of 4: the signal is arriving, the identifiers are not strong enough to match, and the brand can read it in Events Manager. State 4 is an EMQ of 7 produced by hashing that fires but normalizes incorrectly. Email addresses hashed before lowercasing. Phone numbers passed without country code. Names sent without trimming whitespace. The score reads as healthy. The platform reports matches. The matches land on the wrong users or no user at all, where the algorithm imputes one anyway. The bidding engine trains on the error. The widely-cited vendor figure of 18% CPA reduction from 8.6 to 9.3 EMQ is a single-source self-report, not an RCT; the honest read is high single digits to low teens from moving 5 to 8.
And the EMQ score itself is computed on the identifiers that made it through the consent gate. A miscalibrated consent state that suppresses external_id for users who never denied, or admits PII for users who did, produces an EMQ number that the platform reports as healthy and that the brand cannot legally rely on.
Axis 4. Attribution capture
A click ID (gclid, fbclid, msclkid, ttclid) that does not survive from ad click to conversion makes the conversion invisible to that platform's reporting and optimizer. The survival path is six handoffs long: ad click, redirect chain, landing page, consent banner, page navigation, checkout, confirmation. The click ID dies in redirect chains that drop query parameters, in SPA navigation that resets the URL, in consent banners that block GTM until interaction (by which point the user has already navigated past the click context), in untagged email. The practitioner consensus: direct traffic above 20 to 25 percent on a paid-media-heavy site is a defect signal.
"Our customers love us so much they come back direct." A site running 35% direct against a 20% market baseline is silently miscrediting roughly fifteen percentage points of conversions to organic loyalty. At $10M paid spend, those fifteen points are $1 to $2M of paid attribution credited to a channel that cannot be optimized. Meanwhile, the bidding engines see fewer conversions than they produced, lower bids, and the channels that actually drove the growth are starved of the budget that would have compounded their effect.
The consent banner is the single most common place this axis fails; the same banner that produces State 4 at the consent gate is, by the same mechanism, producing State 4 at attribution capture. One fault, two axes, inherited together.
Axis 5. Payload completeness
A Purchase event that fires without value and currency cannot be used by any value-based bidding strategy. A Purchase event without content_ids cannot drive dynamic product retargeting. A Purchase event missing customer-information parameters caps EMQ at Axis 3. The business cannot use what was not sent. The median credible recovery for a previously thin payload, across the Stape, Elevar, and Polar case study range (24, 32, 36, 41, 62%; vendor self-reports of self-selected accounts), is 20 to 30 percent.
State 3 (skinny payload) is recoverable: the brand can see the gap and fill it. State 4 is the payload that is complete, schema-valid, and confidently wrong. The content_ids are sent but map to the wrong SKUs because the product feed and the pixel reference different identifier schemes. The currency field carries "usd" instead of "USD" and Meta defaults to USD anyway, hiding the bug for dollar-denominated brands and surfacing it dramatically for everyone else. The value field is populated, but with gross order total instead of subtotal, or with shipping included when the platform expects it not to be. Each trains the bidding algorithm on a confidently wrong distribution.
Payload completeness is also the axis where consent-state inheritance is most legally consequential: a complete payload assembled for a user whose consent state was never correctly propagated is the payload that arrives at the receptor in full, with full PII, optimized for a transaction the brand had no lawful basis to record. The platform receives a perfectly-formed event. The brand has just shipped its largest exposure surface. This is the axis where the agentic commerce trajectory will amplify existing exposure most directly, because the brand's leverage to validate either the payload or the consent state at the source is collapsing at the same time.
Why these four fail together
The transduction axes are described as four separate sections above because each has its own mechanism, its own benchmark, and its own dedicated detector inside the Signal Fracture Audit. They fail as a system because each axis produces inputs the others consume, and because every one of them inherits its starting state from the consent gate.
The fifth dependency is the one that runs underneath all four. Consent state inherits through every axis in the transduction layer. A State 4 consent gate does not produce a State 1 anywhere downstream; it produces a State 4 sequence.
Deduplication works on a contaminated set. EMQ scores on identifiers the brand could not lawfully use. Click IDs are captured for sessions where consent was never resolved. Payloads ship in full for users who said no. Each axis can still be diagnosed on its own terms, and each can still be repaired in isolation; but the work performed at any axis downstream of a miscalibrated receptor is work performed on a corrupted substrate.
The integrated cost
A brand running State 4 across all four transduction axes (dedup broken, EMQ at 7 with wrong normalization, click IDs dying at the consent banner, payload technically complete but carrying wrong content_ids) is not running four independent 10 percent miscalibrations. It is running a system in which the bidding engine is trained on inflated event counts, attached to mis-identified users, sourced from misattributed channels, weighted by mis-coded values. The platform's optimization engine continues to spend at apparent efficiency. The dashboards continue to look healthy.
Three of these axes have published detectors. Meta surfaces EMQ in Events Manager. GA4 surfaces direct-traffic share. These platforms surface deduplication mismatches in the diagnostic tabs of their respective conversion managers. The detectors exist, but no single tool reads them as a system, and no platform diagnostic is designed to catch the case where the numbers look healthy on its own surface but the value being optimized is structurally wrong. And none of the platform-side detectors read the consent state at all. The numbers can be green at every axis while the receptor that gated them was producing State 4 from the first millisecond of the session.
This is what a Signal Fracture Audit is built to do. The audit reads the transduction layer end-to-end, with the consent state as the upstream input to every downstream check. Each axis is read against the four states. State 4 at any axis is a fault chain that propagates to every downstream consumer. State 4 at the gate is the fault chain that propagates to every axis.
What this article reserved
This article covered four axes (plus consent). It deliberately reserved a sixth. Value fidelity is the axis where the wrong number in the value parameter trains the bidding algorithm to acquire the cohort that produces the highest reported revenue rather than the highest contribution margin. The next article in this series takes it on directly.
The transduction layer is the stretch of infrastructure where most signal degradation actually happens, and where the brand has the most direct ability to repair it. Consent is the gate. Value is the receptor. The four axes between them are where the cost compounds, axis by axis, between cycles. Repairing any of them on a miscalibrated gate produces a cleaner pathway for a corrupted signal. The pathway has to be calibrated downstream, and the gate has to be calibrated upstream, or the receptor receives what the brand never meant to send.