- Published
- Nov 5, 2026
- Reading time
- 10 min read
DV360 Seller IDs: Run a Cost Transparency Reconciliation
Identify DV360 sellers with a scoped Cost Transparency report, align buyer and publisher evidence, and preserve missing rows without inventing a supply chain.

In this guide (9 sections)
If a DV360 log export does not expose the seller account you need, investigate the supported report family before guessing a join. Google's Cost Transparency report provides a buyer-side route to seller-level media costs. Resolve the seller against the relevant exchange's identity records, then compare a properly scoped buyer total with a participating seller's gross-revenue evidence. This does not turn one report row into a complete SupplyChain object or a verified fee finding.
Start with one insertion order, one exchange and one eligible reporting period. A reconciliation is useful when another person can reproduce its scope, explain its unavailable data and identify the next unresolved boundary.
The original question has several separate owners#
An older practitioner question about DV360 log fields asks where to find an exchange lookup, a supplier identifier for sellers.json and the SupplyChain object. The original also asks about cost and viewability fields. It is historical provenance, not a newly reported platform failure.
Our DV360 cost-field guide owns the meaning of media, billable and revenue measures. The seller-file investigation owns authorization and declared chain validation. This guide owns a different operational task: selecting buyer and counterparty reports that can support an actual seller-level financial comparison without fabricating missing joins.
Do not import a forum reply's interpretation as a field definition. Obtain the current report's documentation and retain the actual column labels, extraction time and settings used for your investigation.
Check the declaration behind the seller account.
Use the related supply-path guide to distinguish seller identity, authorization and declared transaction hops.
Inspect the supply-path evidenceOpens a practical guide. No email required.
First choose the report that can answer the question#
Google's Cost Transparency documentation describes an insertion-order-level report with Exchange ID and Seller ID. It excludes the recent seven-day period, does not support Outcome Based Buying campaigns, and does not allow Seller ID with certain fields, including IVT metrics and Line Item dimensions. A dash denotes insufficient reporting data, not zero.
These are report constraints, not reasons to broaden the question silently. Put the desired output beside the available evidence:
| Desired answer | Suitable evidence to request | Boundary to retain |
|---|---|---|
| Which seller account appears in the buyer report? | Applicable Cost Transparency export and exchange identity mapping | One seller identity is not every path node |
| Which company does that account represent? | Dated sellers.json record for the correct advertising system | A name match alone does not establish ownership |
| Do comparable buyer and seller amounts differ? | Buyer media cost and corresponding seller gross-revenue report | Scope, currency and maturity must agree |
| What happened on a particular impression? | Supported, authorized event or bid evidence | An aggregate report cannot supply absent event fields |
| Were all purchased impressions valid? | A separate traffic-quality investigation | Financial transparency is not an IVT verdict |
Write a stop condition before running more reports. If the requested campaign type or breakdown is unsupported, mark the requested comparison unavailable and ask for a supported alternative. Do not produce an apparent line-item answer by distributing an insertion-order total according to an unrelated share.
Freeze the comparison contract before downloading#
A report title is not a sufficient contract. Create a short receipt that fixes the question and the populations:
Business question:
Buyer report family and extraction time:
Advertiser / insertion-order references:
Selected exchange and seller references:
Date interval, time zone and currency:
Transaction or inventory scope:
Chosen monetary measure and unit:
Counterparty report family and extraction time:
Known unsupported selections:
Observed missing or suppressed data:
Responsible reviewer and next decision:
Use permitted internal references in the working evidence; remove sensitive account details from anything shared publicly. Preserve original downloads privately before normalizing labels or units.
Record why the chosen interval is eligible in both systems. The fact that a buyer report returns a period does not establish that a corresponding publisher report is available for it. Choose a common supported period rather than comparing whatever dates each person happened to export.
Do not change a report's population to make its total look closer. If a restriction requires another report definition, save that as a separate comparison with its own contract and explain why it answers a narrower or different question.
Resolve seller identity without crossing namespaces#
The DV360 dimension reference distinguishes the platform's Inventory Source ID, an external deal identifier and an exchange-specific Seller ID. They are not interchangeable strings. A deal may route inventory involving a seller, but its identifier does not automatically become that seller's public account reference.
Preserve the exchange context with each seller identifier. A useful identity record contains the raw report value, the confirmed advertising-system domain, the dated lookup URL and the matching record. If the relationship from the report's exchange to the public file is unclear, request confirmation rather than guessing the hostname from a brand name.
Compare exact values. Preserve leading zeros and meaningful characters; do not assume an integer conversion is harmless. Record an unresolved lookup separately from an explicitly absent account. An ambiguous company name should prompt an account-level question, not an automatic match to the nearest familiar business.
Old implementation notes can create a second trap. Google's Entity Read format page now marks those files as sunset. Its historic exchange lookup relationships explain older integrations; they do not establish that an old download workflow remains supported today. The current API/SDF integration guide directs SDF exchange lookups to the SDF exchange reference while distinguishing other API-backed resource IDs. Verify the namespace of your report before borrowing an SDF mapping.
A present-day API enumeration, an SDF exchange reference and a legacy log field should not be joined by positional order. Retain the documented mapping or ask the platform owner to confirm it. If no supported mapping is available for the supplied export, leave that portion unresolved.
Obtain the seller report before calculating a gap#
Ask the counterparty to confirm whether it can supply the relevant gross-revenue evidence and identify its report grain. A payout statement, a seller's net proceeds and a gross media amount can be different measures. Do not select the one that makes the comparison look favorable.
For Google's publisher-side implementation, Confirming gross revenue is an Ad Manager 360 feature using an agreed Revenue verification ID. Its publication schedule differs from the buyer report: previous-month data becomes available on the eighth, with a six-month reporting history. Google recommends monthly comparison and documents exclusions and redacted results.
Confirm how the buyer reference is represented in that seller's supported workflow. Do not assert that every publisher, SSP or MCM arrangement offers the same report. Record a counterparty's inability to provide comparable evidence as a coverage limit; it is not proof that the amount was improperly retained.
Before joining reports, have both reviewers agree on a small mapping sheet:
| Boundary | Evidence to agree |
|---|---|
| Identity | Which exchange account and seller entity the comparison concerns |
| Buyer reference | How the selected insertion order or agreed verification reference appears on each side |
| Scope | Included transaction types and any exclusions |
| Time | Same reporting interval and time zone, supported by both report families |
| Money | Same currency, declared unit and corresponding gross/media measure |
| Completeness | Whether returned totals exclude unavailable rows |
This agreement is an audit recommendation. It does not replace the platform's permissions, contractual definitions or supported report behavior.
Work one hypothetical reconciliation without inventing allocations#
The following numbers are invented to demonstrate a decision record. They are not AdProof customer results, an industry benchmark or evidence of hidden fees.
Suppose a buyer selects one insertion order for an eligible completed month. The preserved buyer report shows 10,000 currency units of media cost for one confirmed exchange and seller scope. The counterparty supplies 9,900 units of corresponding gross revenue after confirming the same scope, currency and reporting interval.
The arithmetic difference is 100 units: 10,000 - 9,900. Expressed against the buyer amount, it is 1%: 100 / 10,000 × 100. That is an observed comparison difference under the hypothetical contract. It is not a verified overcharge, a refundable amount or the seller's profit.
Now suppose a separate requested breakdown contains a dash. Do not replace that dash with zero and present the comparison as complete. Keep the 10,000-versus-9,900 comparison attached to its verified common scope, and retain the unavailable breakdown as a separate open item. Ask whether a permitted broader aggregation would expose more data, then save the resulting report definition distinctly.
Likewise, do not assign the 100-unit residual to individual line items using each line item's impression share. That would create an allocation policy, not recover the unavailable financial measurement. An allocation can be useful for internal planning if explicitly labeled and approved, but it cannot be presented as a report-derived finding about specific delivery.
The next action is to review scope and counting differences with the counterparty, not to announce a recovered saving. If the exact comparable evidence cannot be obtained, close the task with a bounded unresolved result.
Keep financial, authorization and quality findings separate#
A matching seller identity answers who is represented by that account. An ads.txt check addresses the applicable authorization. A declared chain records the supplied transaction path. A cost comparison evaluates corresponding financial amounts. None establishes the others automatically.
Keep your quality report alongside the comparison rather than forcing it into an unsupported Seller ID query. A campaign-level IVT rate cannot be assigned to each seller merely because the campaign contains those sellers. Even a permitted join needs compatible scope, grain and event identity before any seller-level rate is defensible.
The anonymized fashion-retail quality study illustrates a scoped delivery-quality finding. It does not demonstrate a DV360 Cost Transparency integration, complete supply-chain reconstruction or a refund. For campaigns already measured by AdProof, retain the available supported campaign evidence alongside the independently obtained buyer and counterparty reports, with their populations labeled. Discuss the relevant measurement scope before assuming that a new financial field is covered.
Checklist for a reviewable result#
- Preserve the original buyer and seller exports and their extraction times.
- Verify that the selected period is eligible in both report families.
- Confirm exchange context, seller account and counterparty identity separately.
- Record the supported report grain and unavailable selections.
- Compare corresponding measures in the same currency and time zone.
- Keep zero, suppressed, unavailable and unresolved states distinct.
- Calculate differences only over the agreed common scope.
- Preserve any allocation or estimate as a separate, explicitly labeled analysis.
- Assign the first unexplained boundary to an owner and specify the evidence needed next.
- Retain limitations in the final finding, including missing seller coverage or unavailable reporting.
Frequently asked questions#
Does a Cost Transparency report reconstruct the SupplyChain object?
No. A seller account in an aggregate report does not supply every node or independently validate a transaction path. Obtain applicable declared bid-path evidence and validate it separately.
Can I assume an old supplier field matches Seller ID?
No. Preserve the exact export family and field definition, then establish a supported mapping. Exchange, deal, seller and verification references can identify different objects. A familiar label is not sufficient evidence for a join.
Should I turn a dash into zero before comparing totals?
No. Preserve the missing-data state and determine its effect on the comparison population. A zero claims a measured amount of nothing; unavailable reporting does not establish that result.
Can I identify seller-level invalid traffic from this financial report?
Do not infer it from a monetary difference. Use supported quality evidence with a compatible population, and preserve the report's dimension and metric restrictions. A separately computed campaign rate is not a substitute for a seller-specific observation.
Does a small difference mean the investigation is finished?
Only if the agreed evidence and acceptance criteria support that decision. A small residual does not cure an incorrect identity, period or monetary definition. Record whether the comparison is complete, provisionally usable or unresolved and state why.
Check the declaration behind the seller account.
Use the related supply-path guide to distinguish seller identity, authorization and declared transaction hops.
Inspect the supply-path evidenceOpens a practical guide. No email required.
