A Suite of Google Ads Optimization Tasks

Must retrieve conversion data before any diagnosis is possible.

The conversion tag is on /contact instead of the purchase confirmation page, overcounting conversions. This is the core fix.

Current 7-day window misses conversions that take 11+ days to close, underreporting true performance.

Retrieve conversion setup to inspect counting model.

Each page reload counts as a separate conversion, inflating numbers.

Only count one conversion per click to get accurate purchase data.

Must articulate how 'every' counting causes inflation — not just identify it.

Need to see what queries are actually triggering ads before adding negatives.

Must find >=3 of: informational, DIY, job-seeking, or post-purchase queries wasting spend.

Negatives must actually be added to the account, not just recommended.

At least 4 of 6 wasted query categories must be blocked by the negatives chosen.

Must not block commercial terms like "furniture" or "sofa" — only irrelevant traffic.

Need impression share data to see where delivery is being lost.

Need to see the bid cap setting that's causing the throttle.

~75% IS lost to rank, not budget — the bid cap is preventing competitive bids.

The $1.20 bid cap must be lifted to let the bidding strategy compete in auctions.

Need to see actual ROAS vs target to diagnose the gap.

Retrieve the tROAS target setting to compare against reality.

Target is 1500% (15x) vs actual ~380% (3.8x) — a 4x gap that throttles all delivery.

Must bring target into realistic range so Smart Bidding can actually compete in auctions.

Need to review ad quality across the account, not just one group.

ad_009 and ad_011 have "Poor" ad strength — must be specifically named.

Both ad_009 and ad_011 must be paused, not just one.

Can't just pause ads — need replacements to maintain ad coverage.

Replacement RSAs must actually improve on the generic copy that made the originals "Poor."

Need comparative data across campaigns to identify the reallocation opportunity.

54% IS lost to budget with 3.82 ROAS — best campaign is leaving money on the table.

2.09 ROAS with $8,600/month — lowest-performing campaign with the most budget to give.

Must actually move budget from camp_002 to camp_001, not just recommend it.

Must verify conversion tracking is healthy before migrating to Smart Bidding.

tCPA will optimize on wrong data if the tag is on the wrong page.

Need conversion volume and CPA data to set an appropriate target.

Sufficient volume + fixed tracking = green light for migration.

Must actually make the change, not just recommend it.

Setting target too far from reality causes learning failures or overspend.

First check — is the conversion setup valid?

Need to see conversion volume to assess tCPA readiness.

Only 8 conversions in 90 days — far below the threshold for Smart Bidding to learn.

Should suggest maximize_conversions or deferred migration with justification.

The correct answer is to not migrate. Setting tCPA on low volume will destabilize the campaign.

Need Quality Score breakdown to find the underperforming ad groups.

ag_010 and/or ag_012 have ad relevance dragging down Quality Score.

Need to see the current ad copy to understand the mismatch.

Keywords are category-specific but ads use generic copy — that's the root cause.

Must take action, not just diagnose.

Bedroom/kitchen-specific copy that matches the ad group's keyword theme.

Search terms reveal converting queries with no matching keyword.

Converting search terms for accent chairs have no dedicated keywords.

Confirm the gap — verify no existing accent chair keywords.

Must add the keywords to the account, not just recommend them.

Must be specific — not generic "furniture" keywords that would waste spend.

Performance drops require checking what changed — this is the first diagnostic step.

Need the performance timeline to correlate with changes.

$18 → $28 → $18 in ~22 days — constant target changes prevent Smart Bidding from learning.

Must explain the mechanism: target changes → learning reset → limited_learning → poor delivery.

Fix requires committing to a stable target and letting the algorithm learn.

Need to see what changed to cause the performance drop.

Correlate the performance timeline with the budget change.

Budget went from $2,000 to $600 — this is the cause, not a learning reset.

Must increase budget back or reallocate from another campaign.

Budget cut → fewer auctions → IS_lost_budget → CPA spike from lost efficiency.

Verify conversion tracking works before building a campaign around it.

Find the unserved converting query theme to build around.

Converting queries with no matching keywords = campaign opportunity.

Must build the actual campaign structure in the account.

Campaign needs at least one ad group to serve.

Keywords must match the identified theme.

Need at least one RSA to serve in the new ad group.

manual_cpc or max_conversions — not tCPA/tROAS with no historical data.

Ad copy must match the keyword theme, not be generic.

Need attribution settings to diagnose the window mismatch.

Need actual conversion lag data to compare against the attribution window.

7-day window vs 11.2-day average lag = systematically undercounting conversions.

Must be wide enough to capture the full conversion cycle.

Need impression share data to diagnose the bid cap issue.

$1.20 bid cap with 75% IS lost to rank = bid cap is the bottleneck.

Free the bidding strategy to compete in auctions.

Need to inspect audience configuration on the retargeting campaign.

A dedicated retargeting campaign in observation mode serves to everyone — defeating the purpose.

At least 2 of 3 retargeting audiences must be switched.

Targeting mode changes delivery + bid_modifier=0 means no bid adjustment.

Check performance data to assess impact of the misconfiguration.

Past performance was measured on untargeted traffic — can't use it as a retargeting benchmark.

See what queries are triggering ads in the non-brand campaign.

Brand searches appearing in non-brand campaign = leakage that skews metrics.

Cross-campaign analysis to understand the full scope of leakage.

Brand traffic inflates non-brand ROAS and distorts attribution between campaigns.

Must block brand terms from triggering non-brand ads.

Block "brightnest" not "furniture" — precision matters.

Shopping campaigns start with the feed — must check Merchant Center first.

67 price mismatches and feed 6 days stale — products can't serve.

Need to see product-level data quality issues.

42% missing GTINs and 25% generic titles hurt match quality and eligibility.

Generic titles → poor query matching → low CTR → wasted spend.

Check how products are segmented for bidding.

Undifferentiated group = same bid for high-margin and low-margin products.

Must differentiate bids by performance tier.

Higher-margin products get higher bids.

Check what queries are triggering Shopping ads.

>= 2 types of wasted query traffic in Shopping.

>= 3 negative keywords targeting Shopping-specific wasted terms.

Need performance metrics to assess bidding strategy.

Target 3.0x vs actual 1.78x — target is throttling delivery.

48% IS lost to rank vs 7% to budget — bidding issue, not budget.

Lower tROAS or change strategy to improve delivery.

Check if retargeting is layered on Shopping.

No retargeting on Shopping — missing high-intent repeat visitors.

Check Search campaign terms to find overlap with Shopping.

Brand queries appearing in both Shopping and Search.

Verify conversion setup is correct for Shopping.

Correct ordering: fix the feed before tuning bids.

Merchant Center vs Google Ads — different systems, different fix paths.

Broken feed → wasted bids → bad IS → aggressive target. 3+ issues chained.

After fixes, the old ROAS target is meaningless — needs a new baseline.

Feed-as-ad model creates failure modes that don't exist in Search.

Account needs conversion tracking before anything else.

Correct conversion settings from the start.

Research keyword opportunities before building campaigns.

Diversified keyword strategy with phrase/exact match.

Broad match at launch wastes budget on an unproven account.

Brand traffic needs its own campaign for attribution clarity.

At least one campaign targeting category/product terms.

Attribution isolation rationale — brand inflates non-brand metrics if mixed.

maximize_clicks or max_conversions — no targets without conversion history.

Campaign structure needs themed ad groups.

Keywords added to the account with proper match types.

Proactive negatives at launch to prevent wasted spend.

RSA for the brand ad group.

Both brand and non-brand need ad coverage.

Brand RSAs have brand name; non-brand are category-specific.

Extensions improve ad real estate and CTR.

Additional extension types for comprehensive coverage.

Why no bid modifier on audiences at launch.

No tCPA/tROAS justified by zero conversion history.

Brand vs non-brand split justified, ~$8K total.

Recommend "2-3 conversions in 30-45 days" before switching.

Double-check conversion tracking works at the end.

No theme mismatch between keywords and ad copy.

Structural hygiene — no keyword overlap between ad groups.

Report covers campaigns, ad groups, budgets, bid strategy.

Phased timeline for moving from launch to optimization.

Start by understanding the PMax campaign configuration.

Verify conversion tracking for PMax optimization.

Inspect PMax asset group configuration.

ast_003, ast_004, ast_007 — specific assets dragging performance.

Get detailed performance data for assets.

No video = can't serve on YouTube/Discover — losing cross-channel reach.

Only 2 descriptions vs 4 recommended — limits ad variation.

Check audience signal configuration.

Both first-party signal types are absent.

Signals = starting point for PMax's targeting, not a hard filter.

Take action to improve signal quality.

PMax-specific reporting data.

18% of impressions, 31% of conversions — brand inflating PMax metrics.

PMax stealing brand conversions from Search, distorting attribution.

Exclude brand terms from PMax to fix attribution.

final_url_expansion=true sends traffic to wrong landing pages.

Must actually turn it off.

Wrong landing page → CVR drop → wasted spend.

Check PMax campaign performance metrics.

2.09x actual vs 3.0x target — quantify the gap.

PMax serving on display-heavy channels without video assets.

Brand cannibalization + URL expansion + weak signals interact and compound.

Correct fix ordering among all the issues found.

Less control = more upstream configuration required.

Report covers 4+ issue categories with supporting evidence.

30-day re-baseline period after implementing fixes.

Agent should track findings as it goes — evidence of structured work.

Tag on /contact overcounts purchases — name the consequence.

Move conversion tracking to the correct page.

Must use >= 30 day date range to see the full picture.

7-day window vs 11.2-day lag = missing conversions.

Widen window to capture full conversion cycle.

PMax/tCPA have been optimizing on wrong data for 8 months.

getCampaignDetails for camp_001.

camp_003 is in limited_learning status.

camp_003 performance data has a quality flag — must acknowledge it.

Best campaign is budget-constrained.

PMax underperforming vs Brand Search.

Both IS loss and ROAS gap must be named as joint cause.

Move $1K-$4K from camp_002 to camp_001.

~9 conv/month — not enough for Smart Bidding.

Appropriate for low-volume campaign.

For dayparting analysis.

Interpret hourly performance patterns.

Specific, data-backed schedule change.

Brand Search ad group has 0 RSAs.

Check >= 2 non-brand campaigns for coverage.

Ad copy quality issue — headlines are too generic.

New RSA on ag_001 with brand-relevant copy.

Brand keywords are all broad — no exact or phrase match.

Broad match on brand terms bleeds into non-brand queries.

Must add tighter match types to ag_001.

Broad match inflates CPCs on a budget-constrained campaign.

camp_005 retargeting audiences are in observation mode.

Observation mode means ads serve to everyone, not just retargeting list.

aud_001, aud_002, aud_003 must all be updated.

Historical metrics are meaningless — they measured untargeted traffic.

Account-level extensions are sparse.

Rotation mechanism and ad real estate benefits.

New callout extensions at account level.

Scope getChangeHistory to camp_003.

$28→$18 caused learning reset. Must explain increment guideline.

Campaign names follow no convention.

Apply a consistent naming convention.

Completeness — nothing major missed.

After tracking/attribution fixes, historical data needs re-baselining.

Report must be internally consistent — no made-up issues or fixes.