feat(schemas): completion_source qualifier — seller_attested vs vendor_attested completion_rate (closes #3861)#3877
Merged
Conversation
…endor-attested completion_rate Closes #3861 with Option C — qualifier slot, not metric split. completion_rate is dual-natured: seller witnesses via player events; third-party vendors independently attest via SDK callbacks / panel methodology / server-side beacon validation. The two paths can yield materially different rates (especially in SSAI). The qualifier slot established in #3576 is the right home. Schemas added: - enums/completion-source.json: ["seller_attested", "vendor_attested"] Schemas updated (qualifier slot now carries two keys — viewability_standard and completion_source): - core/package.json committed_metrics.qualifier - media-buy/package-request.json committed_metrics.qualifier - media-buy/get-media-buy-delivery-response.json aggregated_totals.metric_aggregates.qualifier (just shipped in #3848) - media-buy/get-media-buy-delivery-response.json by_package[].missing_metrics.qualifier Vendor identity is anchored on the matching performance_standard.vendor BrandRef, not duplicated on the metric row — same pattern as MRC viewability for DV/IAS. Reconciliation: the atomic-unit join (scope, metric_id, qualifier) extends naturally. completion_rate rows now carry completion_source the way viewable_rate rows carry viewability_standard. Doc updates: create_media_buy.mdx and get_media_buy_delivery.mdx list both qualifier keys with their conditional-required semantics and the symmetric missing_metrics flagging behavior. Backwards compatible (additive; closed-vocab so new keys ship explicitly in subsequent minors). Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Per product-expert review on PR #3877: the binary seller/vendor enum was correct as the primary trust-source axis, but the descriptions needed to address real-world measurement topology that doesn't cleanly split: - Walled gardens (YouTube, Spotify) where seller IS the measurement vendor — now explicitly collapses to seller_attested by trust-model logic (same party served and counted). - IAB-certified first-party podcast measurement (Podtrac on Podtrac's own platform, Triton on its own, Art19 on its own) — also seller_attested. - The same vendor on third-party inventory (Podtrac on a non-Podtrac publisher) — vendor_attested. - Panel measurement (Nielsen, Edison) — vendor_attested even though methodology differs from SDK callback (methodology granularity is a separate dimension; reserved for future qualifier keys if needed). Also clarifies the transparency-without-gate use case: buyer who wants vendor-attested completion for transparency without an accountability threshold MAY commit a vendor metric via vendor_metrics on the product rather than committed_metrics with this qualifier. The trust axis is named explicitly: "who is independent of the seller's revenue interest," not "who runs the SDK." Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
bokelley
added a commit
that referenced
this pull request
May 2, 2026
closes #3857) (#3884) * feat(schemas): unify outcome measurement into the row-shape vocabulary (closes #3857) Outcome metrics (incremental_sales_lift, brand_lift, foot_traffic, conversion_lift, brand_search_lift) collapse into the existing available-metric.json + qualifier-slot pattern. Attribution methodology and window become qualifier keys. outcome_measurement field on product is deprecated. Why: outcome_measurement was always the same conceptual category as the seller-reported attribution-derived metrics already in delivery-metrics (conversions, conversion_value, roas). The two surfaces existed because the protocol predated the unified row-shape vocabulary established by #3576 / #3848. With the qualifier slot proven generalizable (#3877), the collapse is clean. Schemas added: - enums/attribution-methodology.json: ["deterministic_purchase", "probabilistic", "panel_based", "modeled"] Schemas updated: - available-metric.json: adds 5 outcome metrics - delivery-metrics.json: adds 5 scalar properties - Qualifier slot expanded at all 5 sites (committed_metrics, missing_metrics, metric_aggregates, package-request, performance-feedback) with attribution_methodology (string enum) and attribution_window (structured duration — first object-valued qualifier key) - outcome-measurement.json: deprecated, migration table in description - product.json outcome_measurement field: deprecated No separate attributed_sales metric — that's conversion_value with qualifier.attribution_methodology: "deterministic_purchase". The unified pattern handles the deterministic/probabilistic/modeled split via qualifier, not parallel metric IDs. attribution_window in the qualifier slot is the first object-valued qualifier key. Established that qualifier values can be structured (not just string enums). Window isn't disambiguating "which version" the way viewability_standard does — it's parameterizing — but the join-on- (metric_id, qualifier) pattern handles same-metric-different-window correctly so placement works. Doc updates: commerce-media.mdx shows new pattern alongside legacy field; media-products.mdx, create_media_buy.mdx, get_media_buy_delivery.mdx reflect the four-key qualifier vocabulary. Migration: existing outcome_measurement payloads work for one minor. New implementations use reporting_capabilities.available_metrics + qualifier on commit. Removed at next major. Deferred: reporting_frequency / reporting_format migration to a reporting_capabilities extension (separate concern from per-metric). Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * fix(schemas): brand_lift dimension qualifier + object-valued canonicalization + Walmart migration Three fixes per protocol + product review on PR #3884: 1. (Product, real bug) brand_lift as scalar oversimplifies — Kantar, Upwave, Cint, DV all report awareness/consideration/favorability/ purchase_intent/ad_recall as separate measurements with their own sample sizes. Single scalar would force vendors to pick one or composite. Fix: add lift_dimension qualifier key (5th key in the slot). Same pattern as completion_source / viewability_standard. 2. (Protocol) Object-valued qualifier canonicalization — first object-valued qualifier (attribution_window) needs a canonicalization rule for the join. Added to qualifier description: heterogeneous value types; consumers MUST dispatch on key name; structured-value qualifiers join on canonical (key-sorted) deep equality. Schema description on attribution_window explicitly forbids shorthand strings ("14d") and calls out the object shape. 3. (Protocol) attributed_sales discoverability — buyer agents searching available_metrics for "attributed_sales" won't find conversion_value. Fix: outcome-measurement.json migration table now explicitly maps Walmart Connect's attributedSalesIn14Days → conversion_value + qualifier.attribution_methodology + qualifier.attribution_window with the canonical example. Same pattern called out for Amazon DSP, Kroger Precision, Criteo Retail Media. Plus: rate-style metric inheritance note in qualifier description (per protocol expert) — when a rate carries attribution_methodology qualifier, it applies to the underlying conversions/events. Plus: third-party retail measurement guidance (Circana, NielsenIQ, IRI for retail MMM; Kantar/Upwave for brand lift on retail-media inventory) in outcome-measurement migration description — performance_standards. vendor is the escape hatch for non-seller-vendor cases. Files updated: - enums/lift-dimension.json (new) - core/package.json: qualifier description tightened, lift_dimension added - core/delivery-metrics.json: brand_lift description updated - core/outcome-measurement.json: migration table expanded - core/performance-feedback.json: lift_dimension added to metric.qualifier - media-buy/package-request.json: lift_dimension added - media-buy/get-media-buy-delivery-response.json: lift_dimension added to metric_aggregates and missing_metrics qualifier blocks - get_media_buy_delivery.mdx: qualifier vocabulary updated to 5 keys - changeset: reflects all three fixes Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
This was referenced May 17, 2026
EvgenyAndroid
pushed a commit
to EvgenyAndroid/adcp
that referenced
this pull request
May 17, 2026
…umbrella update Round 3 of @bokelley's third-sibling framing on adcontextprotocol#4616 (issuecomment-4470049814): the epistemic-model table maps 1:1 onto delivery measurement, not just signals. The strawman's shape was already cross-axis, but the descriptions buried measurement as a secondary axis. Promoting it to first-class. Changes are description + examples only — no structural change to the primitive, the fidelity enum, or the conversion schema. system-reference.json: - Description rewritten to explicitly enumerate the four primary axes (identity / taxonomy / geographic / measurement). Calls out measurement currencies (Nielsen P18-49) AND methodology sources (panel / set_top_box / ACR / census / server_logs / SDK per the construction-methodology row in adcontextprotocol#4616) as distinct measurement- side use cases. - `system` examples list extended with `nielsen_p_18_49` and `measurement_source` to surface the measurement axis. - Two new top-level examples: a measurement currency (`nielsen_p_18_49:P18-49:2025`) and a measurement-source row (`measurement_source:set_top_box`) — the latter is the canonical adcontextprotocol#2041 use case. system-reference-fidelity.json: - Description expanded to name two consumer types: buyer agents activating signals, and measurement consumers accepting delivery reports. Adds "methodology sources" to the generalized axis list. system-reference-conversion.json: - Description expanded to name the two surfaces (signal deployments + delivery measurement). References adcontextprotocol#3877's `completion_source` qualifier as shipped prior art for the D3 seller-attested vs vendor-attested split. - New example: STB measurement projected to Nielsen P18-49 currency via Samba TV methodology — the measurement-side analog of the existing signal-side panel projection example. Changeset updated to reflect the broadened scope. Strawman remains shape-only — no row-level adoption, no structural change to the primitive. adcontextprotocol#2041 / adcontextprotocol#4472 / adcontextprotocol#4475 / identity-substrate RFCs still adopt independently against whatever shape D1 settles on. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
4 tasks
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Closes #3861 with Option C from the issue (qualifier slot, not metric split).
completion_rateis dual-natured: the seller witnesses completion via player events; third-party vendors independently attest via SDK callbacks, panel methodology, or server-side beacon validation. The two paths can yield materially different rates — especially in SSAI environments. Samemetric_id, two semantics — exactly the case the taxonomy doc addresses.The qualifier slot established by #3576 is the right home. Adding
completion_sourceas a second qualifier key (alongsideviewability_standard) proves the pattern is generalizable, not viewability-specific.Schema
Added:
enums/completion-source.json— closed enum["seller_attested", "vendor_attested"].Updated (qualifier now carries two keys —
viewability_standardandcompletion_source):core/package.jsoncommitted_metrics.qualifiermedia-buy/package-request.jsoncommitted_metrics.qualifier(buyer-side)media-buy/get-media-buy-delivery-response.jsonaggregated_totals.metric_aggregates.qualifier(just shipped in feat(schemas): metric_aggregates partition — row-symmetric delivery aggregation (closes #3848) #3876)media-buy/get-media-buy-delivery-response.jsonby_package[].missing_metrics.qualifierHow it works
Buyer commits to "85% completion_rate by Nielsen" via:
performance_standards:{ metric: "completion_rate", threshold: 0.85, vendor: { domain: "nielsen.com" } }committed_metrics:{ scope: "standard", metric_id: "completion_rate", qualifier: { completion_source: "vendor_attested" }, committed_at: "..." }Delivery report:
metric_aggregatesrow{ metric_id: "completion_rate", qualifier: { completion_source: "vendor_attested" }, value: 0.87, ... }matches the contract; clean.missing_metricsrow{ scope: "standard", metric_id: "completion_rate", qualifier: { completion_source: "vendor_attested" } }flags the missing vendor commitment.Vendor identity is anchored on
performance_standard.vendorBrandRef, not duplicated on the metric row — same pattern as MRC viewability for DV/IAS.Why Option C (vs A or B from the issue)
seller_completion_rate/verified_completion_rate) is breaking-ish; existing reports usecompletion_rate; tooling needs migration.Files
static/schemas/source/enums/completion-source.json(new)static/schemas/source/core/package.json— qualifier description + completion_source fieldstatic/schemas/source/media-buy/package-request.json— qualifier description + completion_source fieldstatic/schemas/source/media-buy/get-media-buy-delivery-response.json— qualifier blocks in metric_aggregates and missing_metricsdocs/media-buy/task-reference/create_media_buy.mdx— committed_metrics qualifier section now lists both keysdocs/media-buy/task-reference/get_media_buy_delivery.mdx— qualifier vocabulary + missing_metrics example updated.changeset/completion-source-qualifier.md— minorTest plan
Related
metric_aggregates(the row-symmetric delivery surface that benefits from this qualifier)🤖 Generated with Claude Code