Google Ads Operational Log
Events, incidents, and changes that affect campaign performance. Use this to correlate performance anomalies with known operational causes.
When adding entries: include the date, what happened, which campaigns were affected, and the observable impact in the data.
2026
Jan 29–30: Google Merchant Center suspension (domain switch)
What happened: Preparing for the domain switch to hairpin.com, but pages weren’t yet indexed under the new domain. GMC flagged the mismatch and suspended the product feed for approximately one day.
Campaigns affected: All Shopping and PMax campaigns (anything relying on the product feed):
- Shopping Catch All >£20 (ID 71): impressions dropped to 4,252 then 1,323 (vs ~10K normal)
- Shopping Catch All <£20 (ID 78): similar drop
- Shopping Top Performers (ID 67): reduced impressions
- PMax: Knife Rack (ID 76): 12,984 then 847 impressions
- PMax: Table Tops (ID 77): minimal spend those days
- Brand Shopping (ID 72): affected
Not affected: Brand Search (ID 66) — text ads don’t rely on GMC feed. Note: Brand Search’s CPC spike started Jan 27, two days before the GMC issue. These are independent events.
Data impact: Jan 29-30 should be excluded or annotated when calculating weekly/period averages for Shopping and PMax campaigns. The dip is operational, not a performance signal.
Dec 23: PMax Knobs Only paused — Shopping learning lag
What happened: PMax: Knobs Only Shopping (ID 75) was paused after 12 days (Dec 12–23) of loss-making performance (0.37x ROAS, £745 spend, 31% Display waste).
Campaigns affected: Shopping Catch All >£20 (ID 71) and Shopping Catch All <£20 (ID 78) for knob/handle products only.
Impact: PMax takes auction priority over Standard Shopping. During the 12 days PMax ran, Shopping Catch All was suppressed to 1–5 impressions/day on knob products with zero clicks. After PMax was paused, Shopping did not immediately backfill — it took 5 weeks for Shopping to build enough product performance history to start converting on knob terms (first conversions: Jan 26).
Lesson: Running PMax alongside Shopping for niche product categories doesn’t just cannibalise — it prevents Shopping from building optimisation data. The learning gap persists well beyond the PMax pause date. For future product-specific PMax campaigns, be aware that turning them off creates a multi-week gap before Shopping can recover coverage for those products.
Jan 27: Brand Search CPC regime change
What happened: Brand Search CPC jumped from ~£1.50 to ~£6.00 overnight. Cause not fully confirmed — likely a combination of competitor entering brand auctions (impression share dropped 93.5% → 74.7%) and TARGET_SPEND bidding with £200/day budget responding by dramatically overbidding. A new “Hairpin” generic ad group was also added around this time.
See: brand-search-cpc-analysis.md for full analysis.
Status: Ongoing — campaign still running at elevated CPCs as of Feb 9.
Jan 19: PMax Table Tops campaign started
What happened: PMax: Table Tops (SC) (ID 77) first data appears Jan 19. This is a new campaign, not a continuation of an older one.
Note: When evaluating this campaign’s performance, the data window starts Jan 19 — not Jan 11 like the rest of the account’s 30-day audit period.
Feb 3: PMax Knife Rack paused
What happened: PMax: Magnetic Knife Rack (ID 76) was paused. Last day with data is Feb 3. The campaign had been loss-making for three consecutive weeks (0.79x, 0.91x, 1.14x ROAS) and was showing PMax junk-placement behaviour (100K+ impression spikes at <£0.10 CPC with zero conversions).
Decision: Correct. Low-AOV product (~£40-50) couldn’t support paid acquisition at the campaign’s cost structure.