Google Search Partners and Display Network: How to Audit Invalid Traffic
Audit Google Search Partners and Display separately, using the reports each campaign actually provides. Compare qualified outcomes and measurement coverage, not just cheap clicks or raw forms. Neither network membership nor a poor lead establishes invalid traffic by itself.
Quick answer
Start at network level, then inspect available placements, locations, devices and CRM outcomes. Search Partner Network full placement reporting now covers Search, Shopping and App campaigns, with site-level impressions and reporting thresholds. It does not by itself connect each placement to an individual qualified lead.
- Separate network performance from placement visibility.
- Compare matured outcomes at the level supported by the identifiers.
- Test an available scoped control without labelling the whole network fraudulent.
For marketing, analytics, media buying and security teams reviewing paid traffic quality. This guide is technical and operational information, not a legal finding about a specific source.
Search Partners is not the Display Network
Keep Google Search, Search Partners and Display in separate rows before comparing their contribution. A query-led visit and an audience or contextual placement have different acquisition contexts. Record campaign type, network eligibility, budget and optimization goal for the period.
Use the network segment offered by the relevant campaign report, including the Search Partners breakdown where available. Do not assume a Search network setting or exclusion behaves identically in Display, Shopping and Performance Max. If a field is not available for the selected campaign, report the limitation rather than inventing a split.
Which reports are available, and what is missing?
Google's Search Partner Network full placement report is accessed through Report editor and Content suitability. For Search, Shopping and App campaigns it identifies sites and impression counts, subject to inclusion thresholds. Missing rows are not evidence that the campaign never appeared elsewhere.
Google also documents Search Partner segmentation in Performance Max channel reporting. Keep that campaign-level performance view separate from placement visibility. A site impression does not disclose the visitor's identity, prove a billable click or establish which CRM record came from that site.
| Evidence source | Useful for | Do not infer |
|---|---|---|
| Network performance | Comparing available campaign metrics across network segments | A complete per-visitor or per-placement transaction ledger |
| SPN full placement report | Checking reported sites and site-level impressions | A placement-to-qualified-lead join without additional supported identifiers |
| Display placement reporting | Reviewing reported websites, apps and available placement metrics | That every impression is listed or that the report reconciles exactly with billing |
| First-party and CRM records | Assessing accepted enquiries and observed downstream outcomes | A network or placement attribution absent from the captured evidence |
Compare sources, locations and devices on equal terms
Choose comparable periods and conversion definitions. Segment by network first, then by location, device, landing page and time where the volume supports it. Preserve the underlying counts and tracking coverage beside every rate; small cells can produce misleading rankings.
Assess network quality against the intended offer and actual qualification process. A location mismatch can arise from targeting choices, inferred geography or a genuine traveller. A low contact rate can reflect misleading creative or delayed sales follow-up. Neither should be relabelled bot traffic without corroboration.
Join qualified outcomes without inventing placement attribution
Link an accepted submission to its stable CRM record and captured campaign context. Record how many leads can be associated with a source and how many remain unmatched. If the available identifier only supports campaign association, keep the conclusion at campaign level.
Use a documented follow-up window and the same eligibility rules for each cohort. Report pending, genuine but unsuitable and supported fabrication separately. Cost per assessed qualified lead and qualification coverage can inform a response; they are not substitute definitions for Google's invalid-click rate.
Document the identifiers that support the CRM associationUse consistent qualified-lead definitions across sources
Test proportionate source controls
When a recurring pattern warrants action, document the exact setting, campaign scope and available exclusion mechanism before changing it. A brand-suitability exclusion and a fraud finding answer different questions. Turning off a network can change audience mix and volume, not merely remove unwanted visits.
Preserve a pre-change export and a comparison where feasible. Review qualified outcomes, legitimate-user impact and unresolved records after the agreed observation window. A fall in raw forms alone does not show that the remaining traffic improved. Google determines platform invalid-traffic credits separately.
Reconcile platform credits separately from lead-quality findings
Traffic-source audit checklist
Create one record per finding so source-level decisions can be reviewed later. Do not blacklist an entire source based on one suspicious visit.
- Export campaign settings, network breakdowns and available placement reports with the same date scope.
- Record report thresholds, unavailable dimensions and attribution coverage.
- Compare location, device and landing cohorts using the same qualified-outcome rules.
- Inspect suspicious sequences and legitimate examples before choosing a response.
- Assign an owner, review date and rollback condition for every proposed control.
Related reading
Sources and further verification
Platform rules, standards and detection practices change. Check the current primary source and your own measurement scope before applying a control.