Campaign diagnosis
Campaign issue log and renewal register
A campaign readout often ends with a confident recommendation and a scattered set of caveats. A register keeps the caveats alive long enough to decide what should be renewed, fixed, retested, or held.
Use this page after the readout QA pass and before the renewal meeting. It is designed for publishers, advertisers, agencies, and analysts who need one shared record for weak evidence lanes, owner actions, fix deadlines, retest requirements, and final supportable claim language.
When to open a register
Open a register whenever the next decision is stronger than the current evidence. A register is most useful when the team is tempted to solve every issue in the final slide instead of assigning each weak lane to a concrete next action.
| Signal in the readout | Why it needs a register | Best immediate action |
|---|---|---|
| Delivery was mostly in scope, but some placements, dates, or devices were pooled. | The renewal may need a narrower package definition. | Log the affected rows and decide whether they are excluded, repaired, or reported separately. |
| Creative, destination, or offer changes happened during the flight. | The result may mix media quality with changed message or page conditions. | Separate the change window and assign the creative or page owner. |
| Clicks or visits were strong, but qualified outcomes were weak or incomplete. | Traffic, form, routing, and follow-up evidence may point to different fixes. | Log the downstream break point and require status coverage before the next claim. |
| Attribution looks strong without a protected comparison. | The renewal may be reasonable while lift language remains unsupported. | Log the comparison gap and attach the next test requirement. |
| One slice looks like a winner with low delivery or uneven rotation. | The slice may be useful for diagnosis but too fragile for budget shift. | Log minimum-cell thresholds and retest before scaling. |
| The report recommends renewal, revision, and a lift test at once. | Different actions require different evidence and owners. | Split each action into its own row with a claim boundary. |
Minimum register fields
The register should be short enough to use in a meeting and specific enough to survive handoff. These fields keep the decision tied to evidence instead of opinion.
Issue ID
A stable label for the weak lane, such as delivery-scope-01, mobile-form-02, or comparison-gap-01.
Evidence lane
Delivery, exposure quality, traffic quality, creative, destination, lead status, comparison, or claim language.
Affected rows
Package, placement, creative, device, date, geography, destination, or outcome rows that are included in the issue.
Decision risk
The exact recommendation the issue could change: renewal, narrowing, creative revision, destination fix, budget shift, or lift test.
Owner and date
The person or team responsible for the fix, the evidence due date, and the meeting where the issue will be closed or carried forward.
Claim boundary
The strongest sentence allowed until the issue is fixed, excluded, or tested with a stronger comparison.
Register template
The values below show structure, not benchmarks. Replace the example IDs with the campaign's actual package, placement, creative, destination, and outcome records.
| ID | Evidence lane | Issue | Owner | Due | Retest rule | Decision status | Claim boundary |
|---|---|---|---|---|---|---|---|
| delivery-01 | Placement scope | Three lower-volume placements were pooled into the package total. | Ad operations | Before renewal note | Report scoped and pooled rows separately next flight. | Renew narrower package only. | Observed response is strongest in the verified placement set. |
| creative-02 | Creative and destination | Mobile creative drove clicks to a slower page variant than desktop. | Creative and web | Before relaunch | Retest after page speed and message match are corrected. | Fix before mobile scale. | Mobile response is limited by destination readiness. |
| lead-03 | Outcome quality | Lead status is missing for part of the second week. | Buyer analytics | Within status window | Do not compare lead quality until status coverage reaches the agreed threshold. | Hold lead-quality claim. | Submitted leads are reported; qualified lead quality is not complete. |
| comparison-04 | Baseline fit | Prior-flight comparison used different seasonality and a different offer. | Measurement | Next test plan | Reserve a matched baseline or holdout before lift wording. | Plan incrementality test. | The campaign produced directional response, not proven lift. |
| slice-05 | Small-cell readout | One creative-device cell ranks first but has limited delivery. | Campaign analyst | Readout revision | Keep the cell as a hypothesis until minimum delivery is met. | Do not shift budget yet. | The cell is a candidate for retest, not a proven winner. |
Severity bands
Severity should describe decision risk, not how embarrassing the issue feels. A small operational defect can be high severity if it changes the renewal claim.
| Severity | Use when | Allowed action | Do not do |
|---|---|---|---|
| Blocker | The issue prevents a decision-grade readout for the action being requested. | Hold renewal, rebrief, or exclude the affected scope until evidence is repaired. | Hide the issue in a caveat while recommending scale. |
| Decision limit | The campaign can support a narrower action but not the broad recommendation in the deck. | Renew a verified package, fix a weak lane, or plan a test with bounded language. | Let the strongest slice define the whole campaign. |
| Optimization | The issue changes how the next flight should run but not whether the current result was useful. | Assign a fix and make it a condition for the next report. | Treat the fix as proof that the previous result was causal. |
| Watch item | The issue is plausible but not yet large enough to change the recommendation. | Track the field and require a threshold before acting. | Use a weak warning to dismiss useful evidence. |
Closeout rules
Every row should end in one of five states. The closeout state should be visible in the renewal note so the final recommendation does not outrun the record.
| Closeout state | Use when | Final language |
|---|---|---|
| Fixed before decision | The missing field, trafficking record, destination issue, or status gap was repaired before the recommendation. | The row can support the decision, with the repair noted. |
| Excluded from decision | The issue could not be repaired, but the affected scope can be removed from the conclusion. | The conclusion applies only to the verified scope. |
| Condition for renewal | The result is useful, but the next flight requires a specific fix or field. | Renew only with the named condition and evidence due date. |
| Retest required | The result is a plausible signal but too confounded, small, or unbalanced for action. | Carry forward as a hypothesis for the next designed test. |
| Causal test required | The business decision needs lift, but the current report is descriptive or directional. | Use directional language and attach an incrementality test plan. |
Renewal note pattern
A useful renewal note names the clean evidence, the unresolved rows, and the next test without making the reader infer which caveats matter.
Recommended structureRenew the verified contextual package for the placements and devices with complete delivery, traffic-quality, and outcome-status records. Exclude the pooled placement rows from the renewal claim, fix the mobile destination before scale, and reserve a matched comparison before using lift language. The current readout supports a focused renewal and a retest plan, not a broad incremental-impact claim.
Meeting script
- Which unresolved row could change the renewal, creative, destination, budget, or test decision?
- Which issue is a blocker, which is a decision limit, and which is only an optimization note?
- Does every weak lane have an owner, due date, retest rule, and claim boundary?
- Which rows are excluded from the final recommendation, and which are conditions for the next flight?
- What exact sentence is allowed until the fix, status update, or comparison test is complete?
Pair with
Use this register after the campaign readout QA checklist and the creative and destination troubleshooting matrix. Pair it with the campaign reporting worksheet when the renewal meeting needs a one-page note, the private marketplace renewal scorecard when the decision needs scoring, the campaign renewal memo template when scored issues need final renew, revise, retest, or hold language, the campaign renewal evidence archive when closeout states must carry forward, the campaign renewal follow-up tracker when archived issues need owners, due dates, and next-launch gates, the campaign status-window closeout checklist when unresolved rows depend on delayed lead, matchback, attribution, or survey status, the campaign status-window closeout register when those delayed rows need IDs, exclusions, and next-review dates, the private marketplace readout export sample when rows need a consistent field structure, the package performance readout when the buyer-facing narrative needs tighter language, the campaign baseline comparison checklist when the comparison is weak, the outcome quality scorecard when lead or matchback status is incomplete, and the incrementality test plan template when the next decision needs causal evidence.