A useful weekly Google Ads audit should answer three questions: what changed, why it may have changed, and what deserves action next. Start with a like-for-like performance comparison, then review visibility, mature search terms, keywords, ads, and bidding. Finish with a decision queue – not a pile of automated edits.

AI is most useful here as an evidence organizer. It can pull the same checks every week, connect findings across reports, and draft recommended actions. It should not decide that a query is irrelevant, a keyword deserves pausing, or a budget needs increasing without business context.

A daily check catches operational exceptions: campaigns that stopped serving, unusual spend, and pacing problems. A weekly audit waits for more data and looks for optimization opportunities. A monthly or quarterly health audit goes wider into account structure, configuration, targeting, and measurement.

Mixing all three cadences creates noise. If Monday’s review includes every setting in the account, the operator spends more time sorting findings than making decisions.

Use complete, comparable periods. A Monday review can compare the previous Monday–Sunday with the Monday–Sunday before it. If you run the audit midweek, specify exact dates instead of comparing a partial week with a complete one.

Keep the search-term window separate from the headline performance window. Search-term decisions often need more volume, while the newest conversions may still be arriving. Google calls this conversion lag: recent CPA can look too high and ROAS too low because conversions are reported back to the date of the ad interaction.

Adapt the lag buffer to the business. A lead-generation account with a long qualification path may need more caution than an ecommerce account with same-session purchases.

Begin with enabled campaigns. Compare cost, impressions, clicks, conversions, and conversion value with the preceding period. Ratios are useful for spotting movement, but always keep the absolute numbers visible.

Do not diagnose the cause from campaign totals alone. Drill into ad groups, search terms, devices, locations, or conversion actions based on the pattern you find.

Search impression share explains how much eligible visibility you captured. Lost impression share due to budget suggests a funding constraint. Lost impression share due to rank points toward Ad Rank, which can involve bids, quality, expected impact, and auction context.

Neither metric is a command. More budget is only useful when the marginal traffic is valuable. A rank problem should not trigger a blind bid increase before you review query quality, ads, landing pages, and economics.

Split search terms into three queues.

The report is incomplete by design. Google omits some low-volume queries for privacy, so visible search terms should not be treated as the complete universe of demand.

Require an evidence threshold before evaluating keywords. Ten clicks can be a practical minimum for triage, but it is not a universal statistical rule. Compare zero-conversion spend with an approved CPA benchmark, and compare converting keywords against the account or campaign target.

Use Quality Score as a diagnostic filter, not the optimization objective. Its components – expected CTR, ad relevance, and landing-page experience – help indicate where to investigate. They do not tell you whether a keyword is profitable.

Only review keyword or ad-group bids where Manual CPC actually controls them. For automated bidding campaigns, changing a stored keyword bid may have no practical effect. Keep bid candidates separate from keyword-status decisions so a low bid is not mistaken for poor intent.

Start with policy status. A disapproved enabled ad is an operational issue before it is a performance experiment.

For performance, compare responsive search ads with other eligible ads in the same ad group after they have meaningful impressions. An account-wide CTR threshold ignores differences in queries, brand demand, position, and market. Relative comparison is a triage signal, not proof that an ad caused the result.

Sort the output into three buckets:

Every item should include the evidence, affected entity, proposed next step, expected impact, and whether the change requires approval.

Confirm the customer ID, account time zone, current period, comparison period, mature search-term period, and enabled campaign count. If a section failed, the output is partial – not a complete audit.

Defaults make a process repeatable, but they are not business strategy. Replace generic thresholds with the account’s target CPA, minimum meaningful spend, lead quality requirements, and conversion lag where possible.

“Three clicks and no conversions” is evidence. “Add this as a negative” is interpretation. A term can be commercially relevant and still lack enough data. It can also convert offline without that result reaching Google Ads.

Avoid changing negatives, ads, bids, and budgets in one sweep. Group related decisions, record them, and preserve enough time to interpret the result.

HireOtto’s weekly optimization audit runs the core checks above as one read-only workflow. It compares enabled-campaign performance with the preceding equal-length period, reviews Search visibility, and evaluates mature search terms, enabled Search keywords, Manual CPC bid candidates, disapproved responsive search ads, and ads trailing their own ad group’s CTR.

By default, it reviews the last seven days, creates the preceding comparison window automatically, and uses a 30-day search-term lookback with a three-day conversion-lag buffer. Thresholds can be changed for the account. The response prioritizes findings, shows supporting evidence, reports passed checks, flags partial execution or truncation, and provides CSV exports for the fuller candidate tables.

The audit never adds negatives, creates keywords, changes bids, pauses ads, or adjusts budgets. Those are separate HireOtto actions that should happen only after you review the candidate, choose the correct level and match type, and approve the exact change.

The weekly audit is available on the Agency plan. For the surrounding workflows, see HireOtto’s reporting guide, daily optimization guide, Google Ads MCP tools reference, and quickstart guide.

Use daily checks for delivery and pacing, weekly reviews for optimization, and monthly or quarterly audits for structure and configuration. High-spend or rapidly changing accounts may need more frequent reviews.

No. Use a complete period that matches the account’s volume and sales cycle. Low-volume accounts may need 14 or 30 days. Always compare like with like and note promotions, outages, budget changes, and seasonality.

There is no universal number. Review the account’s conversion-lag reporting. The three-day buffer in HireOtto is a practical default, not a promise that all conversions have matured.

HireOtto’s weekly audit is read-only. It surfaces candidates and can suggest the next action, but adding a negative is a separate write operation that should be explicitly approved.

The headline campaign comparison may include enabled campaign performance, but the detailed search-term, keyword, Manual CPC, impression-share, and RSA checks are Search-focused. Use dedicated Performance Max reports for PMax optimization.