Microsoft Ads Publisher Reports: Audit Partner Traffic Before You Exclude It
A weak publisher row is a reason to investigate, not proof of click fraud or wasted budget. Microsoft's Website URL (publisher) report can show impressions, clicks, spend, and conversions by website URL across Microsoft Advertising and syndicated search partner networks. It can be sorted by website URL, account, campaign, and ad group. Use it to locate delivery and connect it to business outcomes before deciding whether a scoped exclusion is justified. Microsoft Advertising: Report types
The report does not establish incrementality, lead quality, margin, fraud, or why a conversion did not appear. Small samples, tracking gaps, delayed sales, and campaign-type limitations can all affect interpretation. Treat it as a diagnostic view, not an account verdict.
A repeatable publisher-quality audit
1. Set a comparable reporting window
Choose a period that includes normal delivery and the conversion lag in your sales process. Match account time zone and dates across the publisher report, campaign data, analytics, CRM, or order system. Annotate tracking outages, promotions, stock issues, destination changes, and major campaign edits. A mature 30-day cohort cannot fairly be compared with the last three days of campaign data.
2. Export the Website URL (publisher) report
Include date, account, campaign, ad group, publisher URL, network, impressions, clicks, spend, and conversions where available. Save the export. Microsoft documents those metrics and dimensions for PublisherUsagePerformanceReport. Check current schema and campaign-type limitations before interpreting an absent row as zero performance. Microsoft Advertising API: Report types
3. Separate delivery and scope
Group results by campaign and ad group before combining them. A row in one campaign does not justify excluding the site everywhere. Check the network label and the campaign's ad distribution settings; partner inventory and Microsoft-owned properties are not interchangeable labels for one placement. Record the narrowest scope that fits the evidence. Keep branded, nonbrand, product, and audience activity distinct where the account allows it.
4. Reconcile clicks with business outcomes
Follow the actual funnel:
| Layer | Compare | What it may reveal |
|---|---|---|
| Delivery | Impressions and clicks by publisher | Whether there is enough activity to review |
| Cost | Spend and CPC by campaign/ad group | Financial exposure in that scope |
| On-site | Landing sessions, form starts, checkout events | Whether visits reach and use the destination |
| Conversion | Valid leads, purchases, or primary events | Whether tracking reports a completed action |
| Business quality | Qualified leads, fulfilled orders, net revenue, contribution | Whether actions create business value |
Align attribution windows and event definitions. Confirm that the conversion goal is active and not missing because of consent, redirects, offline imports, or tagging problems. Reconcile platform conversions with CRM stages or settled orders where possible. A missing conversion can mean poor traffic, a measurement issue, or an outcome that has not matured; this report alone cannot choose among them. For ecommerce, include cancellations, returns, discounts, shipping, and the business's margin definition. For lead generation, assess qualification and contactability.
5. Calculate rates with denominators
Click-through rate = clicks ÷ impressions. Conversion rate = consistently defined conversions ÷ clicks. Cost per chosen conversion = spend ÷ that conversion count, when nonzero. Qualified-lead rate = qualified leads ÷ leads, with the same cohort and maturity rules.
Show counts beside percentages. A rate based on one click is unstable. If a row has no conversions, report its clicks, spend, and maturity; don't call its true conversion rate zero when the sample is small or the sales window is open. Set review thresholds from the account's economics, volume, and sales lag, rather than copying a universal “block after N clicks” rule.
6. Check measurement and destination problems
Check that the final URL resolves, the page matches the ad promise, and users can complete the action on mobile. Compare click records with analytics sessions while accounting for consent and reporting differences. Inspect tags, redirects, inventory, service coverage, and sales follow-up. If clicks do not appear as sessions, the gap could involve measurement, browser behavior, or invalid traffic. Preserve evidence and use Microsoft's invalid-click reporting and support process if needed; don't make a public fraud allegation from one metric.
Choose a decision and define a test
| Decision | Evidence | Next action |
|---|---|---|
| Keep and monitor | Mature outcomes are acceptable or evidence is insufficient | Keep delivery and review a comparable window later |
| Investigate | Clicks, sessions, conversions, or CRM records do not reconcile | Check measurement, landing page, attribution, and follow-up |
| Test a narrow exclusion | A credible concern is isolated to a publisher and scope | Exclude only at the narrowest eligible scope; watch for displaced delivery |
| Exclude | Mature, reconciled evidence repeatedly shows unacceptable quality or a policy/brand-safety reason | Record evidence, scope, owner, date, and review condition |
Low CTR alone is not enough to exclude a site when the goal is qualified conversions; high CTR alone is not proof of value. Microsoft documents campaign and ad-group website exclusions for syndicated search websites, with ad-group exclusions overriding campaign exclusions. Availability and limits vary, so verify the current account controls before acting. Microsoft Advertising: Entity hierarchy limits
Before testing, write the hypothesis, scope, dates, primary outcome, spend guardrail, conversion maturity window, and inconclusive rule. Keep other changes stable where practical. A before-and-after comparison can be confounded by demand, bids, competition, seasonality, and sales operations. If the account cannot isolate the exclusion, call it a monitored rollout rather than a causal experiment. Watch total campaign outcomes: a removed publisher's spend may shift elsewhere.
Keep a log of the date range, time zone, scope, publisher, counts, spend, conversion definition, business outcomes, tracking checks, decision, owner, and review date. Afterward compare the same definitions and total campaign results, not just the removed row. For adjacent checks, see Paid Search Query Review and Audit an Ad Agency Report.
Short-video script: Audit publisher traffic without guessing
Hook: “A low-converting publisher row is a signal to investigate, not proof of click fraud.” Show: Display a redacted or synthetic Website URL report with URL, campaign, clicks, spend, and conversions. Explain: “Match the time period. Check conversion maturity, landing sessions, tracking, and CRM outcomes.” Demonstrate: “Show counts beside rates. Separate campaigns. If the evidence is isolated and repeatable, test the smallest eligible exclusion and set a review date.” Close: “Don't block a network based on a scary headline. Keep the report and the decision rule together.” Production safeguard: Never expose customer identifiers or imply an unmeasured result.
Educational carousel: The six-step publisher audit
- Cover: Verify evidence before excluding a Microsoft Ads publisher.
- Export: Include dates, campaign, ad group, network, clicks, spend, and conversions.
- Scope: Separate campaigns and delivery types; one row is not an account-wide verdict.
- Maturity: Align conversion windows, CRM stages, and order settlement.
- Diagnose: Check tags, landing sessions, forms or checkout, and follow-up.
- Decide: Keep, investigate, test narrowly, or exclude with a reason.
- Measure: Compare qualified outcomes and total spend, not only the removed row.
- Save: Record evidence, owner, date, scope, and rollback condition. Caption: The report describes delivery; validated tracking and business records help explain value. Counts and maturity belong beside rates.
Field experiment thread: Test one exclusion responsibly
This is a proposed test, not a report of measured results. 1/ Choose one campaign and a publisher with a documented quality concern; name the business outcome at risk. 2/ Export baseline dates, time zone, campaign, ad group, network, impressions, clicks, spend, and conversions. 3/ Reconcile conversions with sessions and qualified CRM or order records. Mark gaps and immature outcomes unknown, not zero. 4/ Define the narrowest eligible exclusion, primary outcome, spend limit, maturity window, and rollback rule. 5/ Change only the exclusion where practical. Log it and monitor total delivery for displacement. 6/ Compare the same definitions after maturity. Keep, reverse, or extend by the prewritten rule; call confounded evidence inconclusive.
Tool-neutral prompt pack: Review a publisher report from evidence
Review this Microsoft Advertising Website URL (publisher) export and matching campaign and business records. Use only: [dates/time zone], [campaign/ad group], [publisher URL/network], [impressions/clicks/spend/currency], [conversion event/window], [landing sessions], [CRM or order outcomes], [tracking changes], and [other campaign edits].
Return what the report directly shows; rates with numerator and denominator; whether outcomes are mature; evidence for and against a quality concern; measurement gaps; the narrowest defensible action (keep, investigate, test, exclude); and a decision-log entry with owner and review date.
Do not label a publisher fraudulent from low conversion volume alone. Do not invent thresholds, missing data, lifts, or causality. Separate observation, inference, and recommendation. If evidence is insufficient, state what to check next.
Stop-doing swipe file: Replace network panic with evidence
- Stop: “Partners waste 30% of every budget.” Use: “Check this account's report, business outcomes, and cohort maturity.”
- Stop: “No conversions means fraud.” Use: “Verify sample size, lag, tracking, sessions, and lead or order quality.”
- Stop: “One poor website means exclude all partners.” Use: “Isolate the campaign and publisher; use the narrowest eligible exclusion.”
- Stop: “High CTR proves quality.” Use: “Compare clicks with qualified outcomes and the campaign objective.”
- Stop: “Removed spend equals savings.” Use: “Check total spend, displaced delivery, and qualified volume afterward.”
- Stop: “Before-and-after proves causality.” Use: “Account for other changes, seasonality, demand, and maturity.”
Hook–Value–Template: Make the decision auditable
Hook: “Before blocking a Microsoft Ads publisher, make the evidence reproducible.” Value: A defensible decision connects one publisher row to a campaign, reliable outcome definition, mature data, and a recorded rule. Template: “Review date: [date]. Window/time zone: [range]. Campaign/ad group: [scope]. Publisher/network: [row]. Impressions/clicks/spend: [counts/currency]. Conversion definition/window: [details]. Mature conversions: [count]. CRM/orders and quality: [verified outcome]. Tracking checks: [results]. Confounders: [list]. Decision: [keep/investigate/test/exclude]. Scope/hypothesis: [details]. Guardrail/rollback: [rule]. Owner/review date: [name/date]. Result after maturity: [measured result and limits].”
Microsoft Advertising documentation and review limits
- Microsoft Advertising API: Report types — publisher report purpose, metrics, and dimensions.
- Microsoft Advertising API: Report attributes and performance statistics — meaning of PublisherUrl.
- Microsoft Advertising API: Entity hierarchy limits — exclusion scope and hierarchy.
Fields, eligibility, interface labels, and controls can change. Verify current account documentation. This guide claims no universal waste percentage, fraud finding, or performance improvement.