Google Ads conversion tracking audit

Your conversion actions are reporting values in more than one currency. USD, EUR, sometimes blank. Google Ads doesn't normalise across currencies for bidding — Smart Bidding sums €100 and $100 as 200, and your aggregate ROAS column is meaningless until you fix this.

Your Google Ads account has no enabled Primary conversion action. Not one. Every action is either disabled or marked Secondary, so Smart Bidding has zero signal to optimise against. Maximize Conversions, Target CPA, Target ROAS — they're all bidding blind right now. Fix this before you change anything else.

Your Google Ads purchase action is set to "Don't use a value" (or fixed at zero). To Smart Bidding, every order looks identical. A $10 sale and a $1,000 sale weigh the same. If you're running Target ROAS or Maximize Conversion Value, the algorithm has no revenue signal to optimise against, and it falls back to chasing the cheapest conversions it can find.

One or more Google Ads conversion actions use click-through windows misaligned with the sales-cycle context configured in the audit (short window on a long-consideration purchase, or vice versa). Either direction degrades data quality: a window too short drops valid conversions; a window too long pulls in delayed activity that is unlikely to be ad-attributable.

One or more Google Ads conversion actions are currently disabled despite having significant historical conversion value. The disabled actions are not feeding Smart Bidding, and the underlying business outcome is invisible to optimisation. Either re-enable them or move the value capture to a different active action.

A Google Ads purchase or sale conversion action is configured with a fixed value (e.g. $50 per conversion) even though the underlying transaction value varies meaningfully across customers. Smart Bidding cannot distinguish high-value from low-value customers and optimises for volume of conversions rather than revenue.

Different conversion actions in this account use different attribution models. Some Data-Driven, some Last-Click, some Position-Based. When these actions roll up into the same campaign or goal report, the displayed conversion value mixes attribution logic in ways that make period-over-period comparison and bidding decisions unreliable.

One or more lead-style conversion actions (form submits, demo requests, signups) are configured with monetary values. Unless the team has consciously assigned a "lead value" for value-based bidding, this is almost always accidental: a sale-conversion template was copy-pasted and the value field was forgotten. The result is Smart Bidding treating leads as direct revenue and over-prioritising lead campaigns against actual sales.

No enabled Google Ads conversion action in this account is suitable to serve as a Primary bidding goal. Either nothing is marked Primary, or the actions marked Primary are micro-conversions (signups, page views, add-to-carts) rather than the macro business outcome. Smart Bidding optimises toward Primary actions, so campaigns are currently optimising toward the wrong target.

One or more campaigns use Target ROAS values that diverge materially from the account's historical ROAS performance. Targets set well above achieved ROAS will cause Smart Bidding to suppress spend; targets set well below will leave revenue on the table by over-spending on marginal traffic.

One or more Google Ads conversion actions use a click-through window shorter than the typical click-to-conversion delay for this business. Conversions that happen after the window closes are not attributed to the ad click, which underreports campaign performance. Especially for upper-funnel and remarketing traffic where delay is normal.

This account's Primary conversion volume is below the threshold Google recommends for stable Smart Bidding (~15-30 conversions per bidding portfolio per month). With sparse signal, automated bidding strategies like Target CPA and Target ROAS optimise from noise, produce volatile CPA, and enter extended learning periods after every change.

Every enabled conversion action in this account uses Last-Click attribution. Several high-volume actions would qualify for Data-Driven Attribution (DDA), which is generally more accurate. The account is leaving attribution accuracy on the table.

This account uses many different attribution models across its conversion actions. Last-Click, Position-Based, Data-Driven, Linear, Time-Decay. Without a discernible governance pattern. Account-level reports that mix these are not interpretable; campaign decisions made against the mixed signal are unreliable.

A conversion action has a counting method that does not match its category. a Lead action set to "Every" (inflates lead counts when users retry submissions), or a Purchase action set to "One" (undercounts repeat purchases in the same click session). Whichever way it is misaligned, the action systematically misrepresents volume.

Two or more conversion actions in this account have different names but appear to track the same business event ("Purchase," "purchase-web," "Website Purchase"). Unless explicitly de-duplicated, these inflate measurement and confuse campaign-goal selection.

One or more conversion actions in this account have extreme value outliers. Individual conversions reporting values 10×, 100×, or more above the median. Outliers can corrupt Smart Bidding when the algorithm treats a single $50,000 conversion as evidence for a pattern that will not repeat. The fix is to investigate the outliers, fix the data pipeline if they are bugs, or cap conversion values if they are real-but-misleading.

One or more Google Ads conversion actions use a counting method that does not match the underlying business event. Lead-style actions (form submits, signups) should typically use "One". Count one conversion per click no matter how many times the user submits. Sale-style actions (purchases, transactions) should use "Every". Count every transaction. AdLint flags this when the configured counting method is the opposite of what the category implies.

A conversion action in this account is firing regularly (count > 0) but every conversion reports a value of zero. The action is configured to accept dynamic values, but the upstream pipeline (GTM or import) is sending zero. Likely because a Data Layer Variable is unresolved or the import column is missing.

A significant share of conversions in this account are reported with substantial delay after the click. Typically because they come from offline imports or attribution windows that allow late conversions. Smart Bidding's learning speed is gated on how fresh the feedback signal is; high reporting lag means the algorithm is making decisions based on a stale picture.

Several conversion action names in this account are non-descriptive. Generic strings like "Conversion 1," "Lead," or "Website Lead Action 2025-03." These names make audits harder because a reviewer cannot tell from the name alone what business event the action represents or which source it pulls from.

Google Ads matches conversion tags by conversion ID and label, not by friendly name, so a naming mismatch between GTM tag names and Google Ads action names does not break tracking on its own. It does make the setup hard to audit. A reviewer cannot tell at a glance which Ads action receives each tag hit, which increases the chance of editing the wrong tag, leaving renamed actions live, or missing duplicates during cleanup.

Multiple conversion actions in this account track what appears to be the same business event but report different sources. Some Website (tag-fired), some Import (offline upload), some Phone Calls. Without explicit deduplication, the same conversion can be counted across sources and inflate measurement.

One or more conversion actions in this account use Data-Driven Attribution (DDA), which assigns credit using machine-learned path analysis. DDA requires enough recent conversion volume to produce stable credit assignment; below that threshold, the model is noisy or falls back to last-click, making attribution-based decisions unreliable.

One or more conversion actions use click-through windows longer than the realistic causal window for this business. Long windows recover late conversions but increasingly attribute conversions to clicks that may have had no influence. a user who clicked an ad 87 days ago and bought today probably would have bought anyway. The result: inflated reported campaign value and weaker correlation between ad spend and revenue.

This account has accumulated many conversion actions that show no recent volume. They exist in Tools & Settings → Measurement → Conversions but have not fired in 30+ days. Inactive actions clutter campaign goal selection, slow down audits, and create the risk that a future engineer wires a new campaign to a long-dead action by mistake.

One or more conversion actions have categories that contradict their names. an action named "Purchase. Website" categorised as "Lead," or "Lead. Salesforce" categorised as "Other." Either the category is wrong, the name is wrong, or the action genuinely represents something the team has not been clear about.

Multiple conversion actions in this account use identical fixed values (e.g. Three different lead actions all set to $50). The pattern usually indicates each action was created by copying a template and the value field was not customised per the actual business meaning. Even if the values happen to be the right amount, the lack of per-action calibration is a governance smell.

Conversion actions in this account use inconsistent naming patterns. The runtime cost is zero, but inconsistent names slow every audit and increase the risk that a campaign gets wired to the wrong action because two similarly-named actions are hard to distinguish at a glance.

A conversion action's click-through and view-through windows are set in a relative configuration that does not follow the typical pattern. View-through windows are normally shorter than click-through windows because view-through evidence is weaker, but some actions in this account configure them inverted or near-identical, which produces inconsistent attribution treatment.

One or more high-volume conversion actions in this account use Last-Click attribution despite having enough conversion volume to qualify for Data-Driven Attribution (DDA). Last-Click systematically over-credits the closing-touch and under-credits assist touches, which biases bidding toward bottom-of-funnel campaigns and away from upper-funnel traffic that actually drove demand.

A meaningful share of this account's conversion actions are categorised as "Other" rather than a specific Google Ads category (Purchase, Lead, Submit Lead Form, Sign-up, etc.). The "Other" category disables several Google Ads features. Including the per-category bidding optimisations, automatic value-based recommendations, and category-aware reporting filters.

Within a single Google Ads conversion category (e.g. "Lead" or "Sign-up"), individual actions report values that vary materially. This may be intentional (different lead types have different qualified-revenue contributions) or a sign of inconsistent value methodology across actions.

View-through conversion windows determine how long after an ad impression. Not a click. a conversion can still be attributed to that impression. AdLint flags configurations where the VTC window is either too long (likely over-attributing) or too short (likely missing genuine display- and YouTube-driven conversions).