Free AI Max Signal Audit

Is your AI optimising on garbage data?

AI Max and Smart Bidding are only as good as the signals they train on. Bad conversion tracking, phantom signals, and degraded campaigns mean the algorithm is spending your budget based on lies. Find out the truth.

Signal Health Score & Training Noise %
Machine vs Human Control CPA Gap
Algorithm Stage Distribution
256-bit SSL EncryptedGoogle OAuth — No passwords storedRead-only accessPrivacy Policy
AI Max Signal Diagnostics

SIGNAL HEALTH SCORE

41 / 100

TRAINING NOISE %

36.2%

HUMAN CPA

₹820

MACHINE CPA

₹2,140

Campaigns · Algorithm Stage
DEGRADATIONGeneric — Shopping
Score: 28
EXPLORATION███████████████████
Score: ███
EXPLOITATION████████████████
Score: ███
Example preview only — the full audit reveals every campaign, signal score, and override action

FAQs

Override in four scenarios: (1) major business changes Smart Bidding hasn't been trained on — flash sales, inventory shortages, pricing changes, new product launches; (2) tracking breaks where conversion signal has been disrupted and the algorithm is optimising on bad data; (3) competitive disruption where a competitor has exited or entered aggressively and the auction dynamics have shifted faster than Smart Bidding can adapt; (4) insufficient conversion volume (fewer than 30/month for Target CPA, 50/month for Target ROAS) where the algorithm is operating below its reliability threshold. Outside these cases, manual override usually degrades performance because Smart Bidding sees auction-time signals humans can't. The AI Max Signal Audit evaluates all four conditions and produces an Override Action checklist when any apply.

Google states 7 days as the learning period, but the realistic time to stabilised performance is 14–28 days for most campaigns and 30–60 days for low-conversion-volume campaigns. During learning, CPA volatility is typically 30–50% above stabilised levels and ROAS can swing dramatically week-to-week. Avoid making changes during learning (budget, bid target, creative, structure) because each change resets the clock. The AI Max Signal Audit flags which of your campaigns are in active learning phase and estimates the realistic stabilisation date based on your conversion volume.

Three options, in order of preference: (1) use Maximise Conversions without a target CPA — removes the target that requires volume to calibrate while still letting the algorithm learn; (2) switch to Enhanced CPC with manual base bids — gives some algorithmic benefit without Smart Bidding's volume requirement; (3) move to pure Manual CPC if conversion volume is below 10/month. Running Target CPA or Target ROAS with insufficient volume produces erratic performance, extended learning periods, and systematic waste. The AI Max Signal Audit recommends the bidding strategy appropriate for your actual conversion volume, not Google's suggested default.

AI Max refers to Google's expanded AI-driven features: AI-rewritten ad copy (dynamic headlines and descriptions generated at auction time), keyword expansion beyond your provided list, and broader audience matching beyond your specified signals. When AI Max is active, Google controls progressively more of the campaign — keyword selection, creative combinations, audience targeting. The practical effect is a move from advertiser-controlled targeting to Google-controlled targeting, with corresponding reductions in transparency. The AI Max Signal Audit measures what share of your spend is driven by AI-expanded decisions versus explicit human choices, and whether those AI decisions are generating returns above or below your target.

In accounts with 50+ monthly conversions, strong conversion signal quality, and sufficient budget, Smart Bidding typically outperforms manual CPC by 15–30% on CPA and 10–25% on conversion volume because it sees auction-time signals humans cannot (device, time, demographic, query intent, prior search history). In accounts with fewer than 30 monthly conversions or polluted signals (view-through, modeled, duplicated), manual CPC often outperforms because the algorithm is learning on bad data. The comparison is conditional, not absolute. The AI Max Signal Audit estimates your specific account's likely performance delta between Smart Bidding and manual based on your conversion volume and signal quality.

Only when the conversion rate improvement justifies it — if Smart Bidding delivers 3x the CPC but also 3x the conversion rate, CPA is unchanged and the volume gain is what matters. If conversion rate is flat and CPC is 3x, Smart Bidding is actively destroying value. The honest calculation: (Smart Bidding CPA) ÷ (Manual CPC CPA). If above 1.0, manual wins; if below, Smart Bidding wins. Practitioners often compare Smart Bidding CPC to manual CPC in isolation, which is the wrong comparison — the right one is CPA. The AI Max Signal Audit runs the correct CPA comparison between Smart Bidding and the closest manual CPC benchmark for your account.

Change the bidding strategy from Target CPA/ROAS/Maximise Conversions to Manual CPC in the campaign settings — this triggers a new learning phase (2–4 weeks) during which performance will be volatile. Set initial manual bids at 70% of your historical average CPC to avoid paying premium rates while you recalibrate. Keep Smart Bidding paused for at least 30 days before retrying — switching back and forth within a short window resets learning repeatedly. The AI Max Signal Audit recommends the specific bid starting points based on your historical CPC and click distribution.

Signal Health Score (0–100) measures the quality and volume of conversion data feeding Smart Bidding. It combines conversion volume (30+ per campaign per month), verified event accuracy (low Ads-vs-GA4 gap), high-value conversion actions as Primary (not view-through or micro-conversions), and tracking consistency (no gaps, resets, or duplicate fires). Low Signal Health means the algorithm is making bidding decisions on unreliable data, which systematically degrades performance regardless of budget or target. The AI Max Signal Audit calculates your per-campaign Signal Health Score and identifies the specific signal issues dragging it down.

Machine CPA is the cost per conversion from traffic the algorithm expanded into without explicit human direction — Broad Match query expansion, AI-rewritten ad variants, PMax channel selection, audience expansion beyond your specified signals. Human CPA is CPA from traffic matching your deliberately chosen keywords, approved ad copy, and explicitly targeted audiences. In healthy AI-assisted accounts, Machine CPA is within 20% of Human CPA. When Machine CPA runs 2x or more above Human CPA, the algorithm is systematically overpaying for lower-quality conversions. The AI Max Signal Audit calculates both figures side by side per campaign.

Three diagnostic signals: (1) Machine vs Human CPA divergence — Exact and Phrase match conversions have much lower CPA than Broad Match and AI-expanded conversions; (2) phantom conversion spend — growing budget share going to placements or queries that Google attributes conversions to but GA4 cannot verify; (3) ROAS instability — large week-to-week swings without corresponding campaign changes, indicating exploratory rather than exploiting behaviour. Any two of the three indicate AI-driven degradation. The AI Max Signal Audit quantifies all three and produces a single 'AI drag' estimate — the monthly spend being wasted on algorithmic decisions.

Override immediately at the start of a promotion, not after — Smart Bidding cannot anticipate a conversion-rate spike and will underbid during the first 1–3 days of the promotion, missing the highest-volume window. Switch to Maximise Conversions (no target) for the duration, increase daily budget by 30–50%, and revert afterward. For flash sales under 48 hours, Manual CPC with elevated bids is safer than any Smart Bidding strategy. After the promotion, give Smart Bidding 14 days to re-stabilise before making further changes. The AI Max Signal Audit produces a promotion-override playbook calibrated to your account's historical conversion-rate spike size.

Override Actions are specific configuration changes that reduce AI autonomy in areas generating the most waste — switching keywords from Broad to Exact match (overriding query expansion), pinning RSA headlines (overriding copy selection), adding audience exclusions (overriding audience expansion), enabling brand exclusions in PMax (overriding branded traffic capture). Each Override Action has an estimated monthly recovery value so you can apply the highest-impact ones first. The AI Max Signal Audit produces the Override Action checklist sorted by recovery value, so a 15-minute intervention on the top three typically captures 60–80% of the recoverable waste.

Free AI Max Signal Audit — Is Your Algorithm Optimising on Garbage? | ClickCatalyst | ClickHub