Seasonality adjustments tell Smart Bidding about a short future conversion-rate change, such as a major promotion. Data exclusions tell it to ignore affected conversion data after a tracking outage or bad measurement period. They solve opposite problems: real customer behavior versus unreliable data.
VISUAL LESSON
What you will learn
- 01Distinguish real behavior from bad measurement.
- 02Configure the narrow supported control.
- 03Verify bidding and reporting effects responsibly.

ILLUSTRATIVE WORKED EXAMPLE
Triage an illustrative anomaly set
PRACTICAL INTERFACE MAP
Move from anomaly diagnosis to controlled change
Compare promotion calendar, site and tag releases, conversion logs, CRM outcomes, reporting lag, consent state, and campaign history.
Set the supported campaigns, conversion actions, devices, click-date window, expected rate change, and duration shown in the current flow.
Check control status, bidding behavior, reporting, corrected measurement, downstream quality, expiry, and a named owner.
STEP-BY-STEP LESSON
Anomaly → behavior or measurement → narrow control → expiry review
THE LEAD ATLAS METHOD
Lead Atlas Data can research business contacts for a paid campaign's precise categories, locations, and market while the advertiser protects bidding inputs with evidence-based controls.See how custom list research works ↗Diagnose the anomaly
Create a timeline of campaign settings, promotion and price, inventory, site and form changes, tag releases, consent, conversion logs, CRM outcomes, imports, attribution, reporting delay, bids, budget, policy, and external events.
Decide whether customers truly converted at a different rate, measurement failed, or ordinary volatility and other campaign changes explain the movement. Do not use a bidding control to hide weak performance.
Choose the matching control
Use a seasonality adjustment for a known future, short event expected to create a significant conversion-rate change. Google says Smart Bidding already handles regular seasonality and describes the control as best for brief events, ideally about one to seven days.
Use a data exclusion when faulty conversion data would mislead bidding, such as a tracking outage or incorrect values. Exclusions apply to clicks in the selected date range and do not remove the data from normal reporting.
Set narrow scope and dates
Select only the supported campaigns and conversion actions affected, then set dates, timezones, devices, and any expected conversion-rate adjustment using documented evidence. Account for click-to-conversion delay before finalizing the window.
Avoid broad account-wide or long-running controls without evidence. Record the unaffected campaigns and conversions, approver, creation time, expected expiry, and conditions for edit or removal.
Repair measurement separately
For data incidents, fix the tag, consent flow, import, deduplication, value, event, or CRM source; validate with test conversions and reconciliation. Do not treat data exclusion as a substitute for correcting the pipeline.
For promotions, verify inventory, pricing, landing page, creative, operations, and post-event return to normal. Seasonality guidance cannot compensate for an unavailable offer or broken customer path.
Review effects and preserve evidence
Monitor control status, bids, budget, conversion delay, raw reporting, valid conversions, qualified outcomes, values, downstream quality, and system diagnostics. Compare with an annotated baseline and avoid causal claims from an uncontrolled before-after view.
Deliverable: anomaly timeline, root-cause decision, chosen control and rationale, exact scope and click-date window, conversion actions, delay analysis, measurement repair, test evidence, expiry plan, monitoring report, rollback owner, and postmortem.
THE TAKEAWAY
Use seasonality adjustments only for a forecastable short conversion-rate shift, data exclusions only for faulty measurement, and keep date, scope, click timing, and rollback evidence narrow.OFFICIAL REFERENCES