Google Ads Creative common mistakes
Google Ads Creative · Understand the traps and how to avoid them.
Treating responsive search ad asset count as the goal itself, pinning every headline and description to a fixed position so the ad cannot rotate combinations, and then assuming the ad will optimize automatically.
Candidates memorize 'provide many assets' and 'pin to control messaging' as separate rules, then over-apply pinning to guarantee a specific message appears, which suppresses the combination testing that responsive ads rely on.
What to do instead
Pin only when a legal, brand, or regulatory requirement demands a specific position, and pin the minimum number of assets. For everything else, let assets rotate and judge performance by asset-level ratings (Low/Good/Best) rather than by manually forcing order.
Choosing a broad match keyword plus a broad 'Drive conversions' bidding goal for a brand-new campaign with no conversion history, then expecting the campaign to spend efficiently in week one.
Candidates conflate 'broad match plus smart bidding is the modern best practice' with 'it works from day one.' Smart bidding needs conversion volume and signal quality before it can bid well, and broad match widens the query surface at the same time.
What to do instead
Match bidding strategy to data maturity: start with a volume-appropriate strategy or a target that reflects available conversion data, keep match types tighter while the conversion signal is thin, and only widen match type and loosen targets after the account has accumulated stable conversion history.
Reading a Search campaign's reported conversions at face value without checking whether conversion actions are counted as 'Primary' vs 'Secondary' or whether the count setting is 'One' vs 'Every' for the same action.
Candidates learn that conversions drive bidding but skip the account-level conversion setup, so they misattribute performance and misdiagnose why a campaign is or is not optimizing.
What to do instead
Before diagnosing performance, inspect conversion actions: which are primary (used for bidding), which are secondary (observation only), and whether each uses one-per-click or every-conversion counting. Reconcile the campaign's reported numbers to that configuration before drawing conclusions.
Assuming Display campaigns can be judged with the same click-through and conversion metrics as Search, and concluding Display 'doesn't work' when CTR is far lower.
Candidates carry Search mental models into Display, where intent, placement context, and view-through behavior differ fundamentally. Low CTR on Display is normal and not by itself evidence of poor performance.
What to do instead
Evaluate Display on view-through conversions, impression share and reach, frequency, and audience-segment performance, and use appropriate bidding for awareness versus conversion goals. Compare Display to its own benchmarks, not to Search CTR.
Building a Display campaign with only keyword contextual targeting and no audience layers, then wondering why placements are irrelevant or why remarketing scale is missing.
Candidates default to the targeting method they know from Search and under-use audience solutions (affinity, in-market, custom intent, remarketing, customer match) and placement exclusions.
What to do instead
Choose targeting by funnel stage: audience segments for prospecting and remarketing, custom intent or in-market for intent-driven reach, and contextual or placement targeting as a supplement. Add placement, topic, and app exclusions, and review the placements report to prune waste.
Judging a Video campaign on impressions and views alone, or setting a TrueView-style cost-per-view target while the actual business goal is conversions, then reporting the campaign as a success.
Candidates pick the metric that is easiest to see in the interface rather than the metric aligned to the campaign's stated objective, and they do not map Video campaign subtypes and bid strategies to funnel stage.
What to do instead
State the objective first, then select the matching campaign subtype and bid strategy, then the metric. For upper funnel, use view rate, reach, and brand-lift style measures; for lower funnel, use conversions and cost per conversion, and check view-through and assisted conversions where available.
Treating a Shopping campaign's product feed as a fixed input and trying to fix poor Shopping performance only through bids, campaign structure, and negative keywords.
Candidates focus on the campaign UI because that is where the exam's bidding and structure concepts live, and they underestimate how much Shopping performance is determined by feed quality: titles, descriptions, product type, GTIN, availability, and price accuracy.
What to do instead
Audit the feed first: title structure leading with brand plus key attributes, complete and correct identifiers, accurate price and availability, and sensible product type and custom labels for segmentation. Then use campaign priority, listing groups, and negative keywords to steer traffic.
Confusing attribution model with conversion window, or assuming a data-driven attribution model changes the historical numbers already reported rather than how credit is assigned going forward.
Both settings live near each other in conversion setup and both affect reported conversions, so candidates collapse them into one concept and mis-answer questions about why reported conversions differ between reports.
What to do instead
Separate the two: the attribution model decides how credit is split across touchpoints, the conversion window decides how long after a click or view a conversion is counted. Also confirm that model changes apply prospectively and that reports use a consistent model and window before comparing periods.
Answering questions about incrementality or experiment design by citing last-click conversion lift, without accounting for a control group or for cross-channel effects.
Candidates equate 'conversions went up after launch' with causation and skip the measurement-design concepts that distinguish observational reporting from true lift measurement.
What to do instead
For causal claims, describe a controlled design: holdout or control group, randomized assignment where possible, and a pre-registered success metric. Use last-click and attribution reports for descriptive analysis, and reserve lift language for experiments with a comparison baseline.
Applying one campaign structure template (for example, one keyword theme per ad group with single-theme ads) to every account regardless of budget, geography, or product catalog size.
Candidates memorize a 'correct' structure from study material and treat it as universal, ignoring that structure should follow spend concentration, language and geography splits, and catalog segmentation.
What to do instead
Derive structure from constraints: split by budget ownership, geography and language, and product or service margin where those differ; consolidate where volume is too low to support separate campaigns. Justify each split by a decision it enables, not by a template.