Do a full audit of our Google Ads account. Check: (1) campaign structure — are campaigns well-organized? (2) quality scores — which keywords are below 5? (3) wasted spend — search terms burning budget

Starting the data pull now. Running all queries in parallel.

I have all the data I need. Now compiling the full audit report.

The audit is complete. The report has been saved as google_ads_audit_march2026.md. Here's the full summary:

Google Ads Audit — March 2026

KPI Snapshot

MoM is stable. The YoY drop is the critical issue — 81% fewer conversions and CPA nearly doubled. This needs to be the first investigation before any budget decisions.

1. Campaign Structure

10 active campaigns, all DK V4 — naming is consistent. Main problems:

• 4 enabled campaigns spending DKK 0 (old "Focused Campaign" experiments left live)
• 2 campaigns with zero conversions in 30 days: Privatundervisning (DKK 652 wasted) and Cities (DKK 226 wasted)
• 57 paused legacy campaigns cluttering the account
• Same keywords active in 3-4 campaigns simultaneously causing cannibalization

Fix: Pause the zero-spend and zero-conversion campaigns, archive legacy, separate the shared budget, add cross-campaign negatives.

2. Quality Scores

• QS 1: "matematik undervisning" — all three components below average
• ~20 keywords at QS 3, ~8 at QS 4

The pattern is clear and consistent: landing page experience is BELOW AVERAGE on nearly every low-QS keyword. Ad copy is mostly Average or Above Average. This is not an ad problem — it's a landing page problem. A PageSpeed fix and subject-specific landing pages (e.g. /matematik, /engelsk) are the lever here. Expected QS improvement: +2-3 points, which translates to -15-25% CPC on affected keywords.

3. Wasted Spend

• "abacus" — DKK 103, 0 conversions. Add as negative immediately.
• "hhx", "hf", "stx" — DKK 176 combined, 0 conversions. School abbreviations catching irrelevant traffic.
• "hjælp" (Danish for "help") — too generic, high risk
• Keyword "hjælp til matematik" appears in 4 separate campaigns competing against each other

Fix: Add negatives this week. Build a shared negative keyword list. Total estimated savings: DKK 300-500/month minimum.

4. Ad Extensions

All campaigns use sitelinks, callouts, and structured snippets. Missing:

5. Bidding Strategy

Critical finding: All V4 Search campaigns are named "tCPA" but have no target CPA set. They run as unconstrained Maximize Conversions. This is why V4: Educational Levels hit DKK 590 CPA with no guardrail.

Best CPA by campaign:

• PMax: DKK 120 (efficient)
• Generic: DKK 253
• Consolidated: DKK 265

Worst: Educational Levels at DKK 590, Privatundervisning and Cities at infinity (0 conversions).

Fix: Set tCPA targets immediately (recommend DKK 300 to start), add tROAS target to PMax (10x), separate Generic from the shared budget and give it room to scale.

Top 5 Actions This Week

1. Investigate the YoY conversion collapse — compare conversion actions 2025 vs 2026
2. Set tCPA = DKK 300 on all V4 Search campaigns
3. Pause zero-conversion campaigns + zero-spend enabled experiments
4. Add lead form extensions to all search campaigns
5. Add "abacus", "hhx", "hf", "stx" as negative keywords

Pulling all 90-day data in parallel now.

Search terms timed out on 300 rows. Pulling in batches.

Now writing the full report.

Done. Full report saved as google_ads_90day_analysis_march2026.md. Here's what the longer window revealed:

Key Findings from the 90-Day View

The Impression Share Crisis (Biggest Finding)

Every V4 Search campaign is stuck at ≤10% impression share — meaning Google is only showing your ads in 1 in 10 eligible searches. The surprising part: this is not a budget problem. The campaigns are only spending 71-78% of their available budgets. The campaigns are losing auctions before running out of money. The fix is setting proper tCPA targets, which gives Google a clear value signal to bid more aggressively. Without that target, it bids conservatively.

Negative Keywords — DKK 5,400+ in Identifiable 90-Day Waste

The longer window exposed 4 clear patterns of wasted spend that repeat every month:

The full upload-ready negative keyword list (4 lists, ~50 terms) is in the appendix of the report.

Cross-Campaign Cannibalisation

"Matematik tutor" triggers in 4 campaigns simultaneously — DKK 2,106 spend across 3 months. "Lektiehjælp matematik" in 4 campaigns. This drives up your own CPCs and splits QS signals. The report includes a keyword ownership map to fix this.

Biggest Opportunity: SRP/SOP Terms

"Srp tutor" converted at DKK 128 CPA — the best in the account. SRP/SOP are high-school projects where students urgently need intensive help and are clearly willing to pay. This cluster is barely touched. Adding "srp hjælp", "sop hjælp", "projekthjælp" etc. to V4: Hjælp could capture high-converting volume at well below your current average CPA.

Top 3 actions this week

1. Upload the negative keyword lists from the appendix — ~DKK 21k/year in savings
2. Set tCPA targets on all V4 campaigns (table in report) — unlock the 90% of missed impressions
3. Increase PMax budget from DKK 1,100 → DKK 1,300 — it's the only campaign actually spending its full budget and at DKK 176 CPA it's your best performer