SEM Audit Steps: Verify Conversion Goals Before Keywords

A modern SEM audit should begin with conversion goals, not keywords. When automated bidding uses a designated action as its objective, an account can appear efficient while optimizing for page views, unqualified leads, or another event that does not represent commercial success.

The familiar checks—search terms, targeting, ads, landing pages, costs, and returns—remain essential. What has changed is their context: current Performance Max evaluation guidance covers channel-level performance, granular search terms, placements, assets, landing pages, change history, and conversion lag, giving auditors more evidence than a campaign-level total alone.

1. Define the business result

Write down the outcome the account is supposed to produce and the financial measure used to judge it. An ecommerce account may prioritize completed purchases and contribution margin, while a lead-generation account may need qualified opportunities rather than every submitted form.

Turn that definition into a compact audit brief: the accounts and platforms in scope, the review and comparison periods, the principal conversion, and the acceptable cost per acquisition or return on ad spend. Avoid adopting one standard time window for every account. A longer range can stabilize a low-volume analysis, but it can also conceal a recent tracking failure, seasonal change, or campaign rebuild.

2. Preserve a baseline before making changes

Export campaign, cost, conversion, and conversion-value data before changing settings. Segment the baseline by campaign type, network, device, location, and time when the available volume supports a meaningful comparison. Keep brand and non-brand demand separate if their intent and economics differ.

Review change history alongside performance trends. A sudden improvement or decline may coincide with a budget adjustment, bid-strategy change, new landing page, altered conversion goal, or tracking deployment. Comparing periods without accounting for those interventions can misidentify a configuration change as a shift in demand.

Inspect conversion lag as well. Recent spend may have generated conversions that are not yet recorded, making the latest acquisition cost temporarily look worse than it is. Note the reporting cutoff and avoid treating incomplete days as directly comparable with mature periods.

3. Audit measurement and bidding inputs

For every conversion action, record its source, business meaning, value, counting method, attribution settings, status, and the campaigns that use it. Test the complete journey from an ad click through the website or app to the recorded outcome. Duplicate tags, reloadable confirmation pages, missing values, and imported events with different definitions can all distort performance.

The Google Ads conversion-goal documentation distinguishes customer-level defaults from campaign-level overrides and explains that primary actions enter bidding when the associated goal is biddable. Secondary actions generally remain outside bidding and the Conversions column unless they are included in a custom goal.

This goal map is the audit’s first decisive checkpoint. A newsletter signup, add-to-cart event, phone click, and completed sale should not be treated as interchangeable successes merely because all four are recorded. For lead generation, reconcile platform leads with qualified and closed records in the CRM where that data exists; for sales, compare reported value with the order system and document how cancellations, refunds, taxes, and shipping are handled.

4. Recalculate performance using business economics

Clicks and click-through rate can diagnose traffic and messaging, but they do not establish profitability. Calculate cost per qualified lead, cost per acquired customer, conversion value divided by cost, and contribution after variable costs when the required inputs are available. If revenue is absent from the advertising platform, perform the calculation outside it rather than substituting a traffic metric.

Consider a conditional example: a campaign produces 100 form submissions, of which 20 become qualified opportunities. Its cost per qualified opportunity is five times its cost per form. This is arithmetic, not an industry benchmark; an acceptable threshold depends on the company’s margins, close rate, capacity, and customer value.

Evaluate total and marginal performance separately. A campaign may have an acceptable historical average even as additional spending produces weaker conversions. Budget decisions should therefore consider recent incremental results and conversion quality, not only the account-wide average.

5. Inspect campaign controls and structure

Check whether each campaign has a clear job. Record its type, bidding strategy, selected goals, budget, geographic settings, languages, schedule, audiences, brand controls, exclusions, and final-URL behavior. Confirm that these settings match the locations the business can serve and the conversion definition established earlier.

Look for structural overlap instead of enforcing an arbitrary maximum number of keywords per ad group. Campaigns competing for the same intent can fragment budgets and reporting, while excessively broad groupings can weaken the relationship among a query, an ad, and its destination. A useful grouping should support a distinct intent, offer, economic target, or control requirement.

Review budget limitations together with conversion quality. A platform indication that a campaign could spend more does not show that additional spending would meet the company’s financial target. A profitable campaign constrained by budget may justify a controlled expansion, but the decision belongs in the economics review rather than being accepted automatically.

6. Examine the searches that triggered ads

Compare targeted keywords or search themes with the queries and categories that generated traffic. Classify material demand into groups appropriate to the business, such as brand, product, competitor, research, support, employment, and irrelevant intent. Add negative keywords only after checking whether they could also block useful variants or discovery traffic.

Use Performance Max query information together with channel, placement, landing-page, and conversion-quality data. A search term can reveal intent, but it does not by itself show which delivery context or destination produced a valuable result.

Do not expect every search-term report to reconcile with broader campaign totals. The Microsoft Advertising search-term report documentation states that the report includes terms with significant activity, so its impression count may represent only part of the impressions shown in other performance reports. Treat missing query detail as a reporting boundary, not as evidence that the remaining traffic was irrelevant.

7. Connect ads and landing pages to intent

Evaluate whether each ad’s promise matches the triggering intent and destination page. Check factual accuracy, offer consistency, approval status, asset coverage, and whether important prices or qualifications are disclosed where appropriate. Do not call one headline the winner merely because it appeared in a strong combination; automated delivery may expose assets to different audiences and contexts.

On landing pages, verify mobile usability, loading reliability, forms, checkout behavior, calls, and confirmation events. Trace parameters and redirects to ensure campaign identifiers survive the journey. Strong click-through rate paired with weak post-click performance often indicates a mismatch among the query, ad promise, page, or conversion experience rather than a bidding problem.

8. Convert findings into controlled changes

Rank findings by expected business impact, confidence, effort, and risk. Separate repairs, such as broken tracking or an incorrect primary goal, from experiments involving landing pages, targeting, creative, or bidding. Repairs restore the validity of measurement; experiments require a hypothesis and a defined success measure.

Avoid changing goals, bids, budgets, targeting, ads, and landing pages simultaneously unless an urgent failure requires coordinated action. For each change, record the owner, implementation date, affected campaigns, baseline, and review point. The audit is complete when it produces a defensible measurement model and prioritized change log—not when every platform recommendation has been accepted.

An SEM Audit Can Optimize the Wrong Goal—Check This Before Keywords