If Google Ads conversion rates suddenly jump or collapse, do not reach for both Smart Bidding tools at once. Use a seasonality adjustment when you expect a short, planned change in conversion rate—such as a flash sale. Use a data exclusion when conversion reporting itself is wrong because of tagging, website outages, or upload failures. This article gives you a diagnosis ladder, a fit scorecard, and a worked example so you can correct the bidding signal without teaching the algorithm the wrong lesson.
Key takeaways
-
Seasonality adjustments tell Smart Bidding that conversion rates will temporarily rise or fall for a planned event. Google recommends them for major short changes, ideally 1–7 days, and notes they may work less well beyond about 14 days.
-
Data exclusions tell Smart Bidding to ignore clicks tied to bad conversion data. They change what bidding learns, not what your conversion reports display.
-
Google states that Smart Bidding already manages ordinary seasonal patterns, so a seasonality adjustment is not the default response to every holiday or weekend spike.
-
If the root problem is broken measurement, a seasonality adjustment can amplify bad bids. If the root problem is a real short promotion, a data exclusion can hide useful demand.
What each tool actually changes
Both tools sit next to Smart Bidding, but they answer different questions.
Seasonality adjustments schedule an expected conversion-rate change for a future window. Google’s example is a three-day sale where conversion rates rise by about 50%: you can schedule that increase so bids rise during the event and then return afterward without needing a reverse adjustment.
Data exclusions isolate conversion-data problems. Google lists common triggers such as tagging issues, website outages, and data-import problems. If a tag was removed, reported conversions can drop even when demand did not. Excluding the affected clicks reduces the chance that Smart Bidding treats the outage as a true collapse in performance.
Availability also differs. Seasonality adjustments currently cover Search, Shopping, Demand Gen, and Display campaigns on Target ROAS or Target CPA, plus Performance Max and App campaigns (App in beta) across supported bid strategies. Data exclusions cover Search, Display, Shopping, and Performance Max campaigns that use conversion- or conversion-value-based Smart Bidding. Hotel campaigns are outside the data-exclusion path Google documents for these tools.
The Bid Signal Correction Ladder
Use this ladder before you create either control. Climb only as far as the evidence supports.
L0 — Confirm the conversion system is intact
Check the Google tag or Google Tag Manager container, thank-you or purchase events, offline conversion uploads, and consent firing. Compare Google Ads conversion volume with your CRM or ecommerce orders for the same window. If the business saw normal sales while Ads reported near-zero conversions, treat this as a measurement incident first.
L1 — Separate planned events from data failures
Ask one blunt question: did your team intentionally change the offer, checkout friction, inventory, or pricing in a way that should change conversion rate? Or did reporting break while the website experience stayed the same? Planned offer shocks and measurement failures can look identical in the conversion-rate chart. They require opposite tools.
L2 — Prefer seasonality for short, major, expected CVR swings
Create a seasonality adjustment when all of these are true: the change is expected, the estimated conversion-rate delta is material, the window is short (Google’s ideal range is 1–7 days), and Smart Bidding alone is unlikely to react fast enough. Google’s holiday Smart Bidding guidance also points to seasonality adjustments for short promotions such as a three-day flash sale with a significant conversion-rate lift.
L3 — Prefer data exclusions for conversion outages
Create a data exclusion when conversion counts or values are wrong because of tagging, site downtime, or upload failures. Exclude the click dates that could have produced the bad conversions, including conversion delay. Google’s example is an upload issue from October 15–18 with a five-day delay: exclude October 10–18 so earlier clicks with delayed conversions are covered. Aim to cover at least 90% of the impacted clicks.
L4 — Stabilize targets and budgets after the correction
Neither tool is a complete recovery plan. After a data exclusion, Google recommends reviewing CPA or ROAS targets and budgets while performance restabilizes. After a seasonality event ends, reassess targets and remove expired promotions or Merchant Center sale annotations so post-event bidding is not stuck in promotion mode. Give the system time across conversion cycles before judging the new equilibrium—calendar days alone can mislead.
Correction Fit Scorecard
Score each row. The higher total points toward the recommended first action.
|
Signal |
+2 toward seasonality |
+2 toward data exclusion |
|
Cause of the CVR change |
Known promotion, price cut, or limited-time checkout offer |
Tag missing, site outage, wrong event, or offline upload failure |
|
Duration |
About 1–7 days; avoid long windows beyond ~14 days |
Bounded outage; Google advises against frequent or prolonged exclusions |
|
Business vs Ads data |
Orders and Ads conversions move together |
Orders continue while Ads conversions collapse or spike incorrectly |
|
Timing |
You can schedule the adjustment before the event |
You can apply the exclusion quickly once the issue is found, or ahead of planned maintenance |
|
What you want bidding to learn |
Temporary higher or lower CVR is real and should influence bids now |
Reported conversions in that window are unsafe for training |
Interpretation: 6–10 points on one side is a clear first move. A near-tie usually means you still lack proof of cause—finish L0 and L1 before touching either tool.
Worked example: flash sale vs broken purchase tag
Consider a hypothetical Shopify brand running Search and Performance Max on Target ROAS. Conversion rate usually sits near 2.0%. On Thursday morning, Ads shows 0.4%.
Scenario A — planned 72-hour sale. The team launched a sitewide 20% offer, inventory is healthy, and Shopify orders rose with Ads conversions. Scorecard leans seasonality. Schedule a seasonality adjustment for the three sale days with an estimated conversion-rate increase grounded in last year’s comparable event or a conservative internal forecast. Keep the window short. After the sale, leave the adjustment off and reassess ROAS targets rather than leaving promotional pressure in place.
Scenario B — purchase tag stopped firing. Shopify still records normal orders, but Google Ads purchase conversions fall almost to zero. Scorecard leans data exclusion. Apply a data exclusion covering the outage plus conversion delay, fix the tag, and do not remove the exclusion later just because reporting looks ugly. Google advises against undoing exclusions and generally discourages backfilling for bidding purposes; if you backfill for reporting only after an exclusion, wait at least five days and keep the exclusion in place.
The same chart can justify opposite actions. The scorecard’s job is to stop the team from treating every conversion-rate swing as a promotion or every promotion as a tracking outage.
Implementation checklist
-
Document the hypothesis in one sentence: “planned CVR change” or “bad conversion data.”
-
Complete the Correction Fit Scorecard with evidence from Ads, analytics, CRM or store orders, and the change log.
-
If seasonality wins, create the adjustment for the event window, estimate the conversion-rate change carefully, and limit scope to the campaigns that will feel the offer.
-
If exclusion wins, set start and end times that include conversion delay, choose account or campaign scope, and apply device filters only when the outage was device-specific.
-
After either change, watch spend and efficiency for a few days. Google notes temporary bid or spend drops can still occur after exclusions, and longer outages can take one to two conversion cycles to settle.
-
Record what you did so the next person does not stack a second correction on top of an unresolved root cause.
Common failure patterns
-
Using seasonality for ordinary seasonality. Google already expects Smart Bidding to handle many seasonal patterns. Reserve adjustments for major, short shocks you can estimate.
-
Stretching seasonality across weeks. Long adjustments can work poorly; Google specifically flags extended periods such as more than 14 days.
-
Excluding only the days conversions looked low. Exclusions apply to clicks. If you ignore conversion delay, Smart Bidding may still train on contaminated earlier clicks.
-
Treating exclusions as a reporting cleanup. Reporting still shows the excluded conversions. The tool protects bidding learning, not your dashboard aesthetics.
-
Removing an exclusion after the fact. Google advises against undoing exclusions because that can reintroduce unstable learning.
-
Using exclusions for stale, long-running measurement debt. Google’s Shopping and Performance Max Smart Bidding guidance advises against relying on exclusions for issues older than about 14 days for Search and Display, or 60 days for Shopping, and against long-running use.
Limitations and exceptions
These tools do not replace trustworthy conversion values, clear goals, or adequate budget. If Target ROAS is chronically budget-limited, or if primary conversions are the wrong business event, correcting seasonality or outages will not fix the strategy. For multi-account agencies, seasonality adjustments can be created from a manager account and applied across linked client accounts, while data exclusions can also be managed at manager or sub-manager level with times based on the account time zone. New client accounts added after a manager-level seasonality event will not automatically inherit it.
Also remember the complementary controls. Google’s guidance for steering AI-powered Search ads places conversion value rules, data exclusions, and seasonality adjustments in different jobs: value rules reshape how much a conversion is worth, exclusions remove bad training windows, and seasonality adjustments announce short expected conversion-rate changes. Mixing those jobs is a common source of over-correction.
Recommended next steps
-
Pull the last 14 days of conversion rate, conversion volume, and store or CRM orders side by side.
-
Run the Bid Signal Correction Ladder through L1 before creating any adjustment or exclusion.
-
If you need seasonality, schedule it tightly around the promotion and estimate the conversion-rate change from comparable historical events when possible.
-
If you need a data exclusion, include conversion delay, apply it quickly, fix the root measurement issue, and leave the exclusion in place.
-
Revisit targets and budgets after the correction, then wait through enough conversion cycles before scaling or rewriting campaigns.
If your team is deciding how long to wait before judging post-correction performance, use the conversion-cycle approach in How Long Should You Wait Before Judging a Google Ads Campaign? If budget pressure remains after the signal is clean, review what to fix first when Target CPA or ROAS is budget-limited.
When the diagnosis spans measurement, bidding, and creative testing across Google and Meta, Oasbit’s digital advertising services can help you install the operating rhythm so Smart Bidding corrections stay rare, documented, and tied to real business events.
If you want a focused review of whether your next conversion-rate swing needs a seasonality adjustment, a data exclusion, or neither, book a growth strategy session.
