Google Ads Audit Summary — 11 Feb 2026
Data period: Dec 12, 2025 – Feb 9, 2026 (60 days). All figures verified against confirmed purchase conversions.
This is the first time we’ve been able to look at the account with channel-level data (network type segmentation was added Feb 2026). The purpose is to understand what the data tells us, capture the learnings, and identify where we can tighten things up further.
Account snapshot
| Campaign | Type | Status | Spend | Conv | Revenue | ROAS | CPC | Notes |
|---|---|---|---|---|---|---|---|---|
| Shopping Catch All >£20 | Shopping | SERVING | £5,464 | 237 | £21,305 | 3.90x | £0.73 | Star performer — worth monitoring IS decline (76% → 67%) |
| Shopping Top Performers | Shopping | SERVING | £2,783 | 95 | £7,588 | 2.73x | £1.24 | Discussion point: overlap with Catch All |
| Brand Search | Search | SERVING | £3,303 | 65 | £14,377 | — | £6.05 (post Jan 29) | CPC spike caused by domain migration + generic keyword — needs attention |
| PMax Table Tops | PMax | ENABLED | £701 | 19 | £2,892 | 4.12x | £0.84 | 98% Search — well configured; too early to fully evaluate |
| PMax Knife Rack | PMax | PAUSED | £2,265 | 86 | £4,578 | 2.02x | £0.39 | Correctly paused — data confirms the right call |
| PMax Knobs | PMax | PAUSED | £745 | 8 | £278 | 0.37x | £0.22 | Correctly paused — data confirms the right call |
Account totals (60 days): £37,525 spend / £118,688 revenue / 3.16x ROAS. Without non-Search waste: 3.26x.
Key findings
1. PMax: the right calls were already made — now we have the data to explain why
The team paused PMax Knobs (Dec 23) and PMax Knife Rack (Feb 3) based on poor performance. The new channel-level data confirms both were the right decisions and shows exactly what was happening under the hood.
Across all PMax campaigns, all time periods, non-Search channels (Display, YouTube, Gmail, Search Partners) produced zero conversions from 726K+ impressions and 14,300+ clicks. Not low-converting — literally zero.
| PMax Campaign | Status | Display % of spend | Display conversions | Bidding strategy |
|---|---|---|---|---|
| Knobs | Paused Dec 23 | 25.5% | 0 | MAXIMIZE_CONVERSIONS |
| Knife Rack | Paused Feb 3 | 12.2% | 0 | MAXIMIZE_CONVERSIONS |
| Table Tops | Active | 0.9% | 0 | MAXIMIZE_CONVERSION_VALUE |
What we learned: When PMax can’t find profitable Search auctions (niche products with limited search demand), it fills budget with remnant Display inventory at £0.04–0.07 CPC. MAXIMIZE_CONVERSIONS makes this worse; MAXIMIZE_CONVERSION_VALUE largely avoids it. Table Tops works because the bidding strategy is right and there’s enough search demand to keep the algorithm focused on Search.
The other thing the data revealed: PMax suppresses Standard Shopping for the same products. Documented for knobs — during the 12-day PMax campaign, Shopping Catch All was reduced to 1–5 impressions/day on knob products. After PMax was paused, Shopping took 5 weeks to rebuild its learning history and start converting. This is worth knowing for any future PMax decisions.
Overall PMax vs Shopping comparison: Even isolating Search-only traffic, PMax returned 3.66x ROAS vs Shopping’s 3.99x at a fraction of the volume. Shopping is the stronger channel for this account.
Going forward: These learnings give us a clear framework for any future PMax campaigns. If we use PMax again, the data says: use MAXIMIZE_CONVERSION_VALUE bidding, apply Display/YouTube exclusions from day one, and only use it for products with enough search demand to keep the algorithm on Search. For niche or low-volume products, Standard Shopping is the better fit.
2. Brand Search CPC: root cause identified via changelog
CPC jumped from £1.47 to £6.05 starting January 29. Daily spend nearly tripled while clicks dropped 32%.
| Metric | Before (Dec 12 – Jan 28) | After (Jan 29 – Feb 9) | Change |
|---|---|---|---|
| Avg CPC | £1.47 | £6.05 | +312% |
| Daily spend | £55 | £158 | +187% |
| Impression share | 93.5% | 74.7% | -20% |
| ROAS | 3.97x | 1.93x | -51% |
What happened: The Google Ads changelog (change_event API) and domain migration timeline together explain the full sequence:
- Jan 29 12:15pm — Primary domain switched from thehairpinlegcompany.co.uk to hairpin.com
- Jan 29 12:54pm — New “Hairpin” ad group created in Brand Search, with generic keywords “hairpin” (EXACT) and “hairpin.com” (EXACT) at £10 CPC bid. All 7 ads created with
thehairpinlegcompany.co.ukas the final URL. Google Ads ignores the “.com”, so both keywords effectively match “hairpin” — a generic term (hair accessories, road bends) with no brand intent - Jan 29–Feb 2 — Under Maximize Conversion Value (tROAS 4x), Smart Bidding tested the new generic keyword aggressively, bidding up to £14/click trying to find conversions from “hairpin” searchers. During this period, all ads still pointed to the old domain (301-redirecting to hairpin.com), which gradually degraded landing page quality signals — though this was a slow-burn factor, not the immediate cause of the spike
- Feb 3 12:24pm — tROAS raised from 4x to 5x (attempted fix)
- Feb 3 12:29pm — 2 new ads created in THLC ad group with
hairpin.comURLs (including “We’re Now Hairpin.com!”), 1 old ad paused - Feb 3 12:38pm — Bidding strategy switched from Maximize Conversion Value to Target Spend with £5 CPC ceiling (damage control)
The campaign was NOT on Target Spend when the spike began. It was on Maximize Conversion Value (tROAS 4x). The switch to Target Spend on Feb 3 was a reactive change — it stopped the £14 CPC blowouts but locked CPCs at ~£4.80, still 3x pre-spike levels.
The “competitor entry” theory was wrong. The impression share drop wasn’t from a competitor bidding on brand terms — it was the campaign’s own quality signals degrading after generic keywords were added to a brand campaign.
What didn’t cause the Brand Search spike: The GMC suspension (~24 hours starting Jan 29) impacted Shopping campaigns but not Brand Search. GMC and Google Ads Search are separate systems — there is no mechanism by which a Merchant Centre suspension degrades Search campaign quality or bidding. The domain change itself (new domain with no Ads quality history) was a background factor that slowed recovery over weeks, but would not cause an overnight 4x CPC jump — Quality Score recalculation is gradual. The immediate trigger was the generic keyword under aggressive Smart Bidding.
Estimated excess cost: ~£1,680 over 14 days (~£120/day).
Brand Search also pays 2–5x higher CPC than Brand Shopping for the same brand queries. On the term “hairpin leg company”: Brand Search CPC £2.97 vs Brand Shopping CPC £1.14, both converting. There may be an opportunity to let Brand Shopping carry more of the brand traffic load.
3. Shopping Top Performers: worth discussing the structure
| Metric | Shopping Catch All >£20 | Shopping Top Performers |
|---|---|---|
| CPC | £0.73 | £1.24 (+70%) |
| ROAS | 3.90x | 2.73x |
| Cost per conversion | £23 | £29 |
| Impression share | 68% | 86% |
105 search terms appear in both campaigns simultaneously. Top Performers is essentially a hairpin legs campaign — only 2 of 78 ad groups produce meaningful revenue. 69 ad groups are paused, including legacy mobile/desktop splits.
The same products would likely appear in Shopping Catch All if this campaign didn’t exist. Higher IS means it’s winning more auctions, but paying more to do so. “Hairpin legs” appears in 5 campaigns simultaneously with £287 combined spend on that single term.
This isn’t necessarily wrong — there may be strategic reasons for the separate structure that aren’t visible in the data alone. But it’s worth discussing whether consolidation would simplify the account and improve efficiency.
Knobs & handles: a positive story emerging
The PMax experiment didn’t work for knobs, but the data reveals the product category is more viable than the PMax results suggested. Per-unit price is £7–9, but customers buy an average of 8 units per order. Actual AOV: £46 median, £86 mean (from 3,033 historical orders). Monthly order volume is steady at ~55.
Shopping Catch All is already picking up knob conversions at 2.11x ROAS (early data, 4 conversions in 2 weeks, £18 cost per conversion vs PMax’s £93). With PMax out of the way, Shopping is building the learning history it needs to optimise. Standard Shopping looks like the right channel for this category — it just needs time to build data.
Suggested actions
Immediate (this week)
| # | Action | Impact | Detail |
|---|---|---|---|
| 1 | Remove “hairpin” and “hairpin.com” keywords from the Hairpin ad group — these are generic terms, not brand terms. “hairpin” matches hair accessories, road bends, etc. “hairpin.com” is treated as “hairpin” by Google Ads (punctuation ignored). Neither needs paid protection — organic handles navigational/generic queries | Primary cause of the CPC spike. Removes generic search traffic from the brand campaign | brand-search-cpc-analysis.md |
| 2 | Consider removing the entire Hairpin ad group — after removing the generic keywords, it has 315 keywords identical to the THLC ad group. Two near-identical ad groups create noise for any bidding strategy. If the Hairpin ad group has ads with copy tailored to the new brand name, move those ads into the THLC ad group instead | Simplifies campaign structure, reduces Smart Bidding noise | brand-search-cpc-analysis.md |
| 3 | Reduce Brand Search budget to £60–80/day | Natural demand never exceeded £105/day; £200 gives the algorithm too much room | brand-search-cpc-analysis.md |
After keyword removal (stagger changes to isolate impact)
| # | Action | Impact | Detail |
|---|---|---|---|
| 4 | Switch bidding strategy to Target Impression Share (95% absolute top, £2.50 max CPC) — do this a few days after removing the keywords, not simultaneously. Staggering changes lets you isolate the impact of each one | Directly optimises for brand presence. The £2.50 cap prevents overpaying while remaining above the pre-spike ~£1.50 average. Can tighten further as Quality Score recovers | brand-search-cpc-analysis.md |
| 5 | Clarify what “Branded Bid Test” campaign is doing — separate enabled campaign with Manual CPC at £60/day. Zero performance data synced. If targeting the same brand keywords, it’s either competing with Brand Search for the same auctions or serving no purpose | Could be duplicating spend or cannibalising Brand Search | brand-search-cpc-analysis.md |
| 6 | Monitor Quality Score recovery — after removing generic keywords, QS should gradually recover over 2–4 weeks. CPCs should drift back toward the £1–2 range. Don’t make further bidding changes during this recovery period | Patience required — Google recalculates QS gradually, not instantly | brand-search-cpc-analysis.md |
Short-term (next 2–4 weeks)
| # | Action | Impact | Detail |
|---|---|---|---|
| 7 | Discuss Shopping Top Performers structure — is there a strategic reason for the separate campaign, or would consolidation into Catch All simplify things? | 105 overlapping search terms; Catch All achieves better CPC and ROAS | shopping-top-performers-analysis.md |
| 8 | Clean up Top Performers if keeping — Square Industrial Legs (1.73x ROAS) and Tapered Oak Legs (1.06x ROAS) are below breakeven. 69 dead ad groups could be tidied | Focuses budget on the ad groups that perform | shopping-top-performers-analysis.md |
| 9 | Review PMax Table Tops at 6 weeks (early March) — check whether late-Jan attribution gap filled in. If cumulative ROAS stays above 3x, consider budget increase. If not, migrate products to Shopping. | Only remaining PMax campaign — well configured, just needs more data | pmax-table-tops-analysis.md |
Learnings for future PMax decisions
| # | Learning | What the data showed | Detail |
|---|---|---|---|
| 10 | Standard Shopping outperforms PMax for niche/low-volume products | PMax fills budget with Display when Search demand is limited. The team’s decision to pause Knobs and Knife Rack was correct — this explains why. | pmax-is-it-worth-it.md |
| 11 | MAXIMIZE_CONVERSION_VALUE is critical if using PMax | Table Tops (0.9% Display waste) vs Knife Rack (12.2%) — bidding strategy makes the difference | pmax-table-tops-analysis.md |
| 12 | PMax suppresses Shopping and the recovery takes weeks | After PMax Knobs was paused, Shopping took 5 weeks to start converting on those products. Factor this into any future PMax experiments. | knobs-and-handles-analysis.md |
| 13 | Apply Display/YouTube exclusions from day one on any future PMax campaign | Google now supports account-level placement exclusions. Zero conversions from these channels across 6 months of data. | pmax-is-it-worth-it.md |
Opportunities
| # | Opportunity | Why | Detail |
|---|---|---|---|
| 14 | Monitor Catch All >£20 impression share — declining from 76% → 67% | This campaign drives 68% of account conversions. Worth understanding whether loosening target ROAS slightly would capture profitable impressions being left on the table. | shopping-catch-all-analysis.md |
| 15 | Knobs/handles via Shopping — early signs are positive (2.11x ROAS, £18/conv). Give it 2–3 months to build data. | Now that PMax isn’t suppressing Shopping, the category can develop naturally. | knobs-and-handles-analysis.md |
| 16 | Multi-unit knob purchases — site doesn’t currently encourage bulk buying, but 86% of knob orders are multi-unit (avg 8). Quantity pricing, “most customers buy 6” messaging, or multi-pack Shopping feed listings could boost both conversion rate and AOV. | A site-side change that directly improves paid performance. | knobs-and-handles-analysis.md |
Data gaps to close
| Gap | Why it matters | Status |
|---|---|---|
| Target ROAS values (all campaigns showing NULL) | Can’t assess whether Catch All and Top Performers are optimising to the same target | Open — capture campaign config in sync |
| Closed — change_event API confirmed: tROAS 4x → Target Spend switch on Feb 3 was reactive. Root cause was “Hairpin” ad group + domain migration on Jan 29 | ||
| November Google Ads data (pre-Dec 12) | PMax Knobs reportedly ran at 4.5x ROAS in Nov — would confirm whether the product works at lower budget | Open — historical backfill if possible |
| Closed — keyword_view synced. Both ad groups have 315+ keywords. Hairpin ad group has 2 unique generic keywords (“hairpin” EXACT, “hairpin.com” EXACT) causing the issue |
Per-campaign detail
Each campaign has a full analysis document in this folder:
| Document | Campaign | Key finding |
|---|---|---|
pmax-is-it-worth-it.md | All PMax (comprehensive) | Full PMax vs Shopping comparison using 6 months of data. Includes channel waste analysis (merged from former pmax-display-waste-analysis.md). Shopping wins on every metric. |
brand-search-cpc-analysis.md | Brand Search | CPC spike from Jan 29. Root cause: generic “hairpin” keyword + domain migration. Bidding strategy switch (Feb 3) was reactive. |
knobs-and-handles-analysis.md | Knobs (cross-campaign) | Real AOV is £46–86 (multi-unit buying). Viable for Shopping. |
pmax-knobs-analysis.md | PMax Knobs | Correctly paused. 31% Display waste. Channel data confirms the decision. |
pmax-knife-rack-analysis.md | PMax Knife Rack | Correctly paused. 15% Display waste, £53 AOV marginal for PMax. |
pmax-table-tops-analysis.md | PMax Table Tops | Well configured (98% Search, 4.12x ROAS). Needs more time for evaluation. |
shopping-catch-all-analysis.md | Shopping Catch All >£20 | Account’s best campaign. 3.90x ROAS, 237 conversions. Worth monitoring trends. |
shopping-top-performers-analysis.md | Shopping Top Performers | Structure discussion — overlapping terms with Catch All, higher CPC. |