The Digital Traction paid search diagnostic

Most Google Ads underperformance is misread. A drop gets blamed on "the algorithm" or on seasonality, a bid gets changed in response, and the real cause stays in place. The diagnostic exists to replace that guesswork with a fixed order of questions, each answered from the account's own evidence, so the cause is named before anything is touched. It is the method underneath the Google Ads account audit and every diagnostic page here.

The order is not cosmetic. Each layer depends on the one beneath it, so the diagnostic starts at measurement validity and works up. If the account is counting the wrong thing as a conversion, then every judgement above it — bidding, budget allocation, even lead quality — is built on a signal that does not mean what people think it means. Better bidding on a wrong conversion just finds more of the wrong outcome faster. So we verify what the account measures before we ever judge how it spends.

Each layer answers one question with evidence pulled from the account. Measurement validity — is a real lead being counted, or an automatically-collected engagement event. Query and demand quality — do the actual search terms show commercial intent, or demand that was never yours to win. Account structure and control — can the account be steered, or is one ad group holding a thousand mixed-intent keywords, which makes Google predict a low click-through rate and suppress how often the ads show. Ad and page relevance — are the query, the ad, and the landing page one conversation or three. Bidding and budget allocation — is spend pointed at value, and is impression share lost to rank or to a budget cap. Lead quality — do form fills become customers. CRM and revenue feedback — does the qualified outcome return to the platform so bidding can learn from it. Reporting and decision use — does anyone actually act on what the numbers say.

When performance drops, guessing is the enemy. We line up the account's timestamped change-history log against the weekly funnel trend and tie the decline to specific dated actions. Then we rule out the competing explanations one at a time, each with evidence rather than a shrug: seasonality (is the whole category down, or just this account), auction pressure (did cost-per-click actually rise), bid changes (did the strategy or its targets move), budget (was there a real cut), and landing-page edits (do they pre-date or post-date the drop). What survives that process is the cause. A change that lines up with the decline and cannot be explained away is where you look first — the reasoning, not the conclusion, is the deliverable.

A leading indicator is the intermediate step that moves before the headline metric does — usually a mid-funnel action, like a form-page view, that leads the reported conversion by a few days. It matters for two reasons. On the way down, it is often the first place a problem shows, before the conversion count catches up. On the way back, it tells you a fix is working days before the outcome you actually care about responds, so you are not waiting blind on a lagging number. The right indicator is specific to the account; the principle — find the step that moves first — holds everywhere.

The method depends on two things it cannot manufacture: access and a baseline. Without read access to the account, its change history, and the tag-manager container, parts of the diagnosis are simply invisible — the analytics API shows which events arrived, not whether a submit event fires on submit or on page load. And without a clean weekly baseline, a drop cannot be dated or compared against normal. Everything here is also read from accounts we have reviewed; single-account findings are observations that sharpen the method, not benchmarks that predict your result.