Google Ads Search common mistakes
Google Ads Search · Understand the traps and how to avoid them.
Applying literal, outdated definitions of match types - treating exact match as character-for-character, phrase match as strictly word-order-bound, or broad match as matching only close synonyms of every term.
Match type behaviour has broadened over time (exact and phrase both absorb close variants such as misspellings, plurals and same-meaning searches), so candidates answer from remembered definitions rather than current behaviour.
What to do instead
Memorise the current triad: exact = same meaning; phrase = same meaning with word order preserved; broad = related searches even when the terms are absent. Then read the stem to see whether it is testing reach or control before choosing.
Naming bid amount, keywords per ad group, or account history as inputs to Quality Score.
Bid and Quality Score both feed Ad Rank, so the two get mentally merged, and Quality Score is shown at keyword level with only three components visible.
What to do instead
Anchor on the three components - expected CTR, ad relevance, landing page experience - and state explicitly that bid contributes to Ad Rank but never to Quality Score.
Assuming a new Search campaign serves only on Google Search, or failing to connect symptoms (ads appearing on non-search sites) to a specific network setting.
Network options are opt-in toggles buried in campaign setup, and exam stems describe the symptom rather than naming the setting.
What to do instead
Build a symptom-to-setting map: Display expansion present = display placements enabled; partner-site impressions = Search Partners; wrong-geo complaints = location targeting set to 'interest' rather than 'presence'.
Assigning budgets, networks, location or language targeting at ad group level, or assuming an ad group setting overrides the campaign.
The Google Ads hierarchy is uneven - some settings exist only at campaign, others at both levels - so candidates generalise from one setting to all of them.
What to do instead
Fix a level table in memory: budget, networks, locations, languages and bid strategy live at campaign; keywords, ads and their bids live at ad group. Resolve 'the change did not apply' stems by checking which level actually owns the setting.
Recommending heavy pinning of every headline to position 1 as the way to control messaging, or assuming unpinned assets are rotated evenly.
Pinning is the most visible control in the responsive search ad editor, so it feels like the correct answer to any control-oriented stem; unpinned assets are selected by the system to optimise performance, not served in rotation.
What to do instead
Pin only where brand or legal requirements demand it, keep pins to a minimum, meet the asset minimums (at least three headlines and two descriptions), and treat Ad Strength as a diagnostic of asset diversity rather than a ranking factor.
Stating that created assets such as sitelinks, callouts or structured snippets always serve once added, or that they do not affect the auction.
The interface shows assets as 'eligible', so candidates read eligibility as a guarantee, while the Ad Rank formula's mention of expected impact of formats is over-extended.
What to do instead
Phrase it precisely: assets are eligible, not guaranteed - which asset shows depends on the query and position - and their effect on Ad Rank runs through expected impact of formats and CTR, not through a direct quality adjustment.
Treating the daily budget as a hard cap and computing monthly spend as daily budget × 30.
The interface displays a daily figure that reads like a ceiling, and the overdelivery rules are counterintuitive, which is exactly why they are tested.
What to do instead
Recall the three rules together: spend can reach up to 2× the daily budget on any single day, the monthly charge limit is daily budget × 30.4, and shared budgets pool spend across the campaigns that share them.
Describing Target CPA or Target ROAS as a hard per-auction cap, or recommending them for an account with little or no conversion history.
The strategy names sound like ceilings, and the conversion-volume prerequisite appears once in the documentation, while stems often describe a brand-new account and simply ask which strategy to pick.
What to do instead
Treat tCPA and tROAS as averages the system bids toward, not caps; start new accounts on Maximize clicks or Maximize conversions, move to tCPA/tROAS once conversion volume exists, and expect a learning period after strategy or target changes.
Comparing conversion counts produced under different attribution models, or expecting Google Ads conversion totals to match Analytics exactly.
Both tools label their output 'conversions', but credit is distributed differently by model, and tools differ in attribution windows, modelled conversions and session handling.
What to do instead
Compare like-for-like models and windows only, remember that data-driven is the current default rather than last-click, and treat cross-tool variance as expected rather than evidence of a broken tag.
Leaving every tracked action - page views, add-to-cart, newsletter signups - marked as a primary conversion included in bidding, then attributing poor performance to the bid strategy.
Conversion setup is usually done once and never revisited, and the 'primary vs secondary' control is easily conflated with the separate 'include in Conversions' setting.
What to do instead
Audit conversion actions before diagnosing bidding: keep only true business outcomes as primary and included in bidding, mark the rest secondary for observation only, and check the count setting (one vs every) and conversion window before drawing conclusions.