What key metrics should I regularly monitor to detect poor ad performance?

The Monday written summary

Every account I run gets a Monday morning written summary. Four metrics, one paragraph, sent before anything else opens. The point of the cadence is not the document. It is the forced act of reading the same four numbers the same way every seven days, so a drift in any one of them registers as a drift instead of a surprise at the end of the month. The four weekly metrics below are the ones I put in that paragraph. The four monthly metrics get a longer read on the first Monday of the month.

Weekly metric 1: CPA trend

Cost per acquisition pulled at the account level, then again at the campaign level for the top three spenders, on a rolling seven-day window compared against the prior seven days and the trailing twenty-eight. One week of CPA in isolation is noise. The trend across three windows is signal. A campaign whose CPA has risen twenty percent week over week and fifteen percent against the trailing twenty-eight is moving in the wrong direction even if the absolute number still looks acceptable.

Concern threshold: any top-three campaign with a seven-day CPA more than twenty percent above its trailing twenty-eight-day baseline. Typical first move: pull the search terms report for that campaign before touching bids.

Weekly metric 2: spend pacing

Pacing is the most boring metric in this list and the one that catches the most damage. Take month-to-date spend, divide by days elapsed, multiply by days in the month, and compare to budget. A campaign pacing to spend a hundred and forty percent of its monthly budget in the first ten days is either delivering against a temporary demand spike or running on a bid strategy that has slipped its leash. A campaign pacing to fifty percent is starving the auction and giving share away.

Concern threshold: any campaign pacing more than fifteen percent above or below its monthly target by the second Monday of the month. Typical first move: open change history on that campaign and read every change made in the prior fourteen days.

Weekly metric 3: search impression share

For any Search or Shopping campaign with meaningful spend, pull impression share, impression share lost to budget, and impression share lost to rank. Track each one week over week. A campaign whose total impression share has dropped ten points in seven days is losing the auction to either a competitor or a self-inflicted ad rank problem, and neither one fixes itself.

Concern threshold: impression share lost to rank climbing five points or more in a single week on a campaign that previously cleared the auction. Typical first move: auction insights for that campaign, then ad strength and landing page experience for the ad groups that lost rank.

Weekly metric 4: top-spend keyword performance

For Search campaigns, the top ten keywords by spend usually carry sixty to eighty percent of the cost. Read those ten every week: spend, clicks, conversions, conversion rate, and CPA. One keyword going off the rails inside that group will mask a healthy account average for two or three weeks before the damage shows up at the campaign level. Reading the top ten weekly catches it on day seven instead of day twenty-one.

Concern threshold: any keyword inside the top ten by spend with a seven-day conversion rate below half its trailing ninety-day rate. Typical first move: pull the search terms triggered by that keyword and check match type before considering a bid change.

Monthly metric 1: blended ROAS against Shopify

Once a month, on the first Monday, compare the ROAS Google Ads and Meta report against the revenue Shopify attributes to paid traffic. The two numbers will never match exactly. The size and direction of the gap is the metric. A gap inside ten percent says the tracking stack is honest. A gap above twenty-five percent says the platform is taking credit for revenue Shopify cannot see, and no bid change fixes a tracking problem.

Concern threshold: any month where platform-reported revenue exceeds Shopify-reported paid revenue by more than twenty-five percent. Typical first move: a tracking audit, not a campaign change. The free 25-page setup audit covers the conversion deduplication contract.

Monthly metric 2: new-customer ROAS

Total ROAS hides repeat purchases. A returning-customer order through a paid click is a margin loss most months because the same customer would have bought through email or direct. New-customer ROAS, isolated through Shopify customer reports or the platform’s new-customer acquisition column, is the metric that says whether paid is paying for itself. A blended ROAS of three-point-five with a new-customer ROAS of one-point-two means paid is subsidising the existing-customer base, not building one.

Concern threshold: new-customer ROAS below the contribution-margin breakeven for two consecutive months. Typical first move: a prospecting versus retargeting spend split and a check on whether retargeting is double-billing email.

Monthly metric 3: audience overlap

For Meta accounts running more than one prospecting ad set, the audience overlap tool catches one of the most common silent leaks. Two ad sets targeting overlapping audiences bid against each other in the same auction. The platform absorbs the cost and reports both ad sets as functional. Reviewing overlap monthly catches the duplication before it eats a quarter of the prospecting budget.

Concern threshold: any two prospecting ad sets with an overlap rate above twenty percent. Typical first move: consolidate the ad sets or rebuild the targeting so the audiences sit in distinct buckets.

Monthly metric 4: Quality Score distribution

For Google Search accounts, Quality Score is the leading indicator that the CPC line is about to move. Pull Quality Score at the keyword level for the top fifty keywords by spend, group into seven-or-higher, four-to-six, and three-or-lower buckets, and watch the distribution shift month over month. Twenty percent of top-spend keywords slipping into the three-or-lower bucket means CPCs are about to rise across the account regardless of what the bid strategy is set to.

Concern threshold: a five-point swing in the distribution toward the lower buckets across one month. Typical first move: ad relevance and landing page experience review on the keywords that fell. The Wasted Spend Calculator covers the Quality Score uplift maths.

The cadence is the product

On a roofing client I monitor weekly, the dashboard showed twenty-one thousand clicks at a dime each across the trailing ninety days. Top-decile by every surface metric. The Monday cadence caught what the dashboard did not: the campaigns were lead-form objectives where every form open registered as a click, and the form-completion rate against the twenty-one thousand opens was barely under five percent. Without the cadence, the founder would have read the click cost and kept scaling spend against the leak. The metric that matters is rarely the metric the platform highlights.

Founders who want me to run the eight-metric cadence against the live account instead of running it solo can send the account read-only via the contact form; the process page describes how the cadence fits inside a paid engagement.

A founder who pulls these eight metrics on the schedule above will catch most paid-ad problems in the week they start instead of the month they end. The metrics themselves matter less than the act of reading them on the same day, in the same order, every week. The full diagnostic library sits at /wasted-ad-spend/.