Google Ads Auto-Apply Recommendations Checklist

Use this Google Ads auto-apply recommendations checklist before enabling a category, after a major account change, and during recurring governance. The essential sequence is: confirm account scope, validate conversion and business guardrails, select a narrow allowlist, verify the saved setting in History, inspect each application, and keep a documented rollback path.

Google’s official auto-apply documentation says the feature works at the account level, recommendations are applied when relevant, and the available recommendation types can change. Use this list with the current Google Ads screen and each recommendation’s details.

1. Account scope and policy

[ ] The customer account ID and account name are confirmed.
[ ] The account owner, approver, reviewer, and rollback operator are named.
[ ] Every campaign sharing the account-level setting is listed, including new launches, seasonal campaigns, experiments, and regulated offers.
[ ] The account’s primary business objective and acceptable risk range are written down.
[ ] Current budgets, bidding strategies, targets, and major constraints are recorded.
[ ] Allowed, manual-only, and prohibited recommendation categories are documented.
[ ] The review cadence, success metric, and stop condition are explicit.
[ ] Recent Manage, History, Account Settings, and Change history views have been checked for existing auto-apply subscriptions.

Auto-apply is account-level, so a category cannot be treated as a harmless campaign-specific experiment unless the account structure and policy support that interpretation. If campaigns have incompatible objectives, narrow the allowlist or resolve the account boundary before proceeding.

For a written policy example, see the auto-apply best-practices guide. Keep the policy outside a screenshot: the durable record should explain why the category is enabled and what would make it unacceptable.

2. Measurement and business guardrails

[ ] Primary and secondary conversion actions are intentional.
[ ] Form, call, purchase, lead-stage, and offline-import events are not duplicated.
[ ] Values, currencies, timezones, and conversion windows are consistent.
[ ] Consent behavior and data-retention requirements are understood.
[ ] The team can distinguish a tracking change from a performance change.
[ ] The baseline includes spend, impressions, clicks, conversions, value, CPA or ROAS, and a quality measure such as qualified leads or revenue.
[ ] Conversion lag and normal seasonality are recorded.
[ ] Budget ceilings, target ranges, brand rules, geographic boundaries, excluded products, and lead-quality requirements are explicit.

Do not enable an automation category that optimizes toward an event the business does not actually value. A recommendation can improve the reported conversion rate while the account produces fewer qualified leads if the optimization signal is wrong. Check the signal before diagnosing the recommendation.

Use the conversion tracking settings guide for the measurement inventory and keep the auto-apply decision separate. The goal is to know whether a later change affected demand, delivery, or the data used to evaluate delivery.

3. Recommendation selection and risk

[ ] The current Manage view, not an old screenshot, is being used.
[ ] Each selected recommendation has been read in full.
[ ] The affected object and likely side effect are documented.
[ ] Ad and asset changes meet brand, legal, product, and landing-page rules.
[ ] Keyword and targeting changes meet intent, location, audience, and negative-keyword rules.
[ ] Bid or target changes have a stable conversion signal, approved range, and named owner.
[ ] Measurement changes have a technical test and business approver.
[ ] Bundles have been expanded and reviewed rather than accepted blindly.
[ ] The allowlist is narrower than the set of categories the account merely finds interesting.

Use a simple decision test: Is the action narrow? Is it observable? Is it reversible? Does it align with the account objective? Can someone review it at the promised cadence? If any answer is no, keep the recommendation manual or resolve the missing control first.

Google’s auto-apply overview lists eligible families, but the current account determines what is offered. Recommendation availability and descriptions can change; as of January 26, 2026, Google says “Add responsive search ads” was removed from Auto-apply. Record dated research notes when a policy depends on a current list.

4. Enable and verify the setting

[ ] The Recommendations page is open for the intended customer account.
[ ] Auto-apply settings or the Account Settings auto-apply view is open.
[ ] Each chosen category is compared with the written allowlist.
[ ] Unapproved bundles and categories are left off.
[ ] The account ID, account name, and selected state are captured before saving.
[ ] The correct Save control is used for each selected bundle or category.
[ ] The saved state is checked after the interface confirms the change.
[ ] Manage, History, or Account Settings shows the expected subscription.
[ ] Change history identifies who enabled the setting and when.

Do not treat a successful Save message as proof that a recommendation has run. Google says recommendations are applied when they are relevant, so the first verification is the enabled subscription and its scope. The first automatic application must be reviewed as a separate checkpoint.

If the navigation label is different, follow Google’s current management instructions and record the actual path used. Interface changes are a reason to update the runbook, not a reason to select every visible control.

5. First-application review

[ ] The recommendation type and affected object are recorded.
[ ] Before and after values are compared in Change history.
[ ] The change stayed within the account policy and approved business scope.
[ ] No unapproved budget, bid, targeting, ad, landing-page, or conversion change occurred alongside it.
[ ] Search terms, asset quality, delivery, conversion events, and lead or revenue quality are checked according to the recommendation family.
[ ] Manual edits and external events during the review window are annotated.
[ ] The result is marked retain, limit, investigate, or disable.

Use immediate safety checks for policy, brand, destination, and financial errors. Allow enough time for conversion lag and normal variation before making a performance claim. Google’s apply-or-dismiss guidance recommends waiting before evaluating performance and checking Change history.

6. Monitoring and recurring governance

[ ] The Recommendations queue is checked on a defined cadence.
[ ] History is reviewed for enabled categories, last application, and recent application counts.
[ ] Change history is checked for exact object edits and the user who made the enablement or manual change.
[ ] Spend, traffic, conversion quality, revenue or margin, and the relevant campaign metric are segmented by affected campaign.
[ ] Budget changes, promotions, landing-page releases, tracking changes, and seasonality are annotated.
[ ] Weekly email summaries, if used, are treated as alerts rather than the authoritative record.
[ ] The category owner and next review date remain current.
[ ] The policy is rechecked after changes to conversion tracking, budgets, products, consent, account ownership, or campaign structure.

Do not use optimization score as the only monitoring signal. A category can raise the score while reducing qualified demand, and an applied recommendation can look worse temporarily because another change or normal seasonality moved the account. Evaluate the business outcome that the campaign is supposed to produce.

7. Rollback and disablement

[ ] A policy, brand, financial, targeting, measurement, or lead-quality stop condition is written down.
[ ] The affected category can be disabled from Manage or History.
[ ] Relevant Change history entries and before-and-after values are saved.
[ ] The rollback scope names the campaigns or objects to restore.
[ ] Related settings are checked before restoring a prior value.
[ ] The team understands that disabling future auto-apply does not erase past applications.
[ ] Spend, events, conversions, and quality are reconciled after rollback.
[ ] Re-enablement requires a new owner, narrower policy, and review date.

Google’s management guide describes disabling a recommendation from Manage or History. Use Change history for the specific automatic changes, and do not restore an old bid, conversion action, or keyword list from memory. If the category is repeatedly rolled back, remove it from the allowlist until the operating policy changes.

For detailed diagnosis, use the auto-apply troubleshooting guide. adz.so can audit supported Google Ads settings but does not approve, apply, or reverse the account’s recommendations on your behalf.

Research note: reviewed against Google Ads Help documentation for account scope, eligible recommendation behavior, Manage, History, Change history, dismissal, and disablement on 2026-08-03. Verify current UI labels and account availability before using this checklist.