brand-search-vs-shopping

Brand Shopping vs Brand Search: Cannibalization & CPC Analysis

Context

Two brand campaigns run simultaneously:

• Brand Search - SC (id 66): Search, currently Maximize Clicks (Target Spend) with £2 CPC ceiling, £200/day budget. Bidding history per change_event API: Manual CPC (pre-2024 through Nov 14, 2025; ECPC state unknown) → Maximize Conversion Value tROAS 4x (Nov 14, 2025 - Feb 3, 2026) → Maximize Clicks £5 ceiling (Feb 3) → £2 ceiling (Feb 11)
• Brand Shopping - SC (id 72): Shopping, TARGET_ROAS at 4.5x, £55/day budget. Created Dec 11, 2024 (tROAS 3x, £25/day). Activated Dec 16, 2024.

Also investigated: General Search (SC) (id 70): Search, MAXIMIZE_CONVERSIONS, £120/day, active Oct 2024 - Jul 23 2025, now PAUSED.

Questions: Is there significant cannibalization? Why is Brand Search CPC so much higher than Brand Shopping? Is Shopping sitting above Search on the SERP cannibalizing clicks?

Data sources: 26 months of campaign-level daily metrics (Jan 2024 - Feb 2026), search term overlap, impression share, conversion actions, device/network breakdowns, Google Ads change history (Feb 2024 - present), Google Search Console organic data (Oct 2024 - Feb 2026, 7.8M rows), auction insights (Dec 2023 - Feb 2026, Brand Search).

---

Answer

Is Brand Shopping cannibalizing Brand Search? There is overlap, but it is not the problem. Both campaigns bid on “hairpin leg company” every month — that’s unavoidable when you run brand campaigns in both Search and Shopping. But Shopping handles that term at £1.19 CPC vs Search at £1.84, and Shopping is 26% cheaper per conversion on the same query. The overlap costs something (~£1,700/year on the Shopping side), but it’s buying cheaper conversions, not wasting money.

Why is Brand Search CPC so much higher? Not because of Shopping. Brand Search CPC has spiked twice — to £4.21 avg in Feb 2025 and £5.50 in Feb 2026. Both crises were caused entirely by Brand Search’s own bidding strategy and budget decisions:

• Crisis 1 (Feb 2025): A Shopify checkout upgrade broke conversion tracking. Enhanced CPC on Brand Search reacted by bidding aggressively into the signal void. Subsequent budget increases (£70→£133 over 8 weeks) funded the elevated CPCs instead of fixing them. A separate campaign (General Search) amplified the damage by competing on the same brand terms with broken conversion data.
• Crisis 2 (Jan 2026): Brand Search had been switched from Manual CPC to Smart Bidding (Maximize Conversion Value) in Nov 2025. When a generic keyword (“hairpin”) was added on Jan 29, Smart Bidding tested it at up to £14/click. The oversized £200/day budget gave it room to do so.

Shopping CPCs stayed in a £0.67-1.12 band through both crises. Shopping did not cause or contribute to either spike.

Can they work together? Yes — and during stable periods (Jul-Nov 2025), they did. Brand Search averaged £1.28 CPC, Brand Shopping £0.97 CPC, both running at healthy ROAS with no conflict. They serve different roles: Search places a text ad for brand defence (38% CTR, single landing page). Shopping places product cards for visual discovery (886 products, no single product above 3.5% of traffic). On the same brand query, Search captures the “I want this brand” click while Shopping captures the “I want this product” click.

Should one be turned off? Not necessarily. But if forced to choose, Shopping is the more reliable campaign: half the CPC, stable ROAS (3.36-5.52x vs 2.01-10.91x), and self-regulating via tROAS. Every time Brand Search has been disrupted, Shopping has naturally absorbed brand inventory at controlled CPCs. The data supports keeping both — but with Brand Search on Manual CPC with a tight budget (£60-80) and hard CPC ceiling (£2), not on Smart Bidding with an oversized budget.

What actually needs fixing? Brand Search’s bidding strategy and budget. The 22-month Manual CPC period (2024 through Nov 2025) delivered avg £0.91 CPC and 9.68x ROAS. The switch to Smart Bidding and budget inflation to £200/day created the vulnerability that both crises exploited. The recommendations section details the specific changes.

The rest of this document is the evidence behind these answers.

---

The Full Picture: 26-Month Monthly Performance

2024 Brand Search Baseline (Pre-Shopping, Pre-General Search)

Brand Search was the only brand campaign from Jan-Nov 2024. Manual CPC bidding, £55-70/day budget.

Month	Clicks	Cost	Avg CPC	CVR	ROAS	Impression Share
Jan 2024	2,058	£1,130	£0.55	6.8%	12.33x	98.6%
Feb 2024	2,051	£1,627	£0.79	5.9%	9.12x	98.0%
Mar 2024	1,852	£1,352	£0.73	5.8%	9.56x	98.6%
Apr 2024	1,656	£1,349	£0.81	8.9%	13.61x	97.6%
May 2024	1,472	£1,483	£1.01	7.4%	8.54x	98.3%
Jun 2024	1,362	£1,388	£1.02	8.1%	10.64x	98.6%
Jul 2024	1,311	£1,661	£1.27	7.7%	7.87x	96.4%
Aug 2024	1,213	£1,432	£1.18	9.9%	10.44x	98.1%
Sep 2024	1,351	£1,050	£0.78	8.1%	9.90x	98.5%
Oct 2024	1,411	£1,519	£1.08	8.8%	10.58x	98.9%
Nov 2024	1,370	£1,072	£0.78	7.9%	9.17x	98.3%
Dec 2024	1,081	£1,470	£1.36	6.6%	5.60x	97.7%

2024 summary: £16,533 spend, 18,188 clicks, avg CPC £0.91, 1,371 conversions, ROAS 9.68x. Rock-solid 97-99% IS on Manual CPC with £55-70/day budget. This is the healthy baseline everything should be measured against.

Auction insights confirm this performance was not due to low competition. 2024 actually had more unique competitors per quarter (10-13) than the 2025 stable period (6-8), with comparable average competitor impression share (14-20%). The strong performance came from three internal factors: Manual CPC with fixed bids that didn’t react to competitive pressure, right-sized budgets (£55-70) that matched natural demand, and no internal competition from other campaigns (General Search didn’t start until Oct 2024, Brand Shopping until Dec 2024).

Attribution model note: The 2024 and 2025-26 ROAS figures are directly comparable. Daily conversion data shows fractional values (e.g. 4.72, 2.14, 0.37) from the very first day of 2024, confirming data-driven attribution (DDA) was already active. Google auto-migrated remaining last-click conversion actions to DDA in September 2023. The 9.68x ROAS in 2024 is not inflated by last-click attribution — the performance difference versus 2025 is real.

2025-2026 Brand Search

Brand Search - SC

Month	Clicks	Cost	Avg CPC	CVR	ROAS	Impression Share
Jan 2025	1,490	£1,691	£1.14	8.8%	9.88x	98.6%
Feb 2025	534	£2,246	£4.21	9.5%	2.62x	54.8%
Mar 2025	886	£2,653	£2.99	8.5%	3.45x	78.4%
Apr 2025	1,182	£3,840	£3.25	8.6%	3.61x	96.6%
May 2025	1,148	£3,919	£3.41	7.6%	2.96x	99.0%
Jun 2025	1,045	£2,881	£2.76	7.2%	3.37x	98.5%
Jul 2025	1,111	£1,850	£1.66	7.3%	5.55x	99.3%
Aug 2025	1,021	£1,080	£1.06	9.4%	10.91x	98.9%
Sep 2025	1,149	£1,528	£1.33	10.8%	8.26x	97.5%
Oct 2025	1,363	£2,356	£1.73	7.3%	4.80x	97.1%
Nov 2025	1,289	£1,572	£1.22	8.0%	5.12x	95.8%
Dec 2025	996	£1,604	£1.61	7.1%	3.93x	96.8%
Jan 2026	1,333	£2,721	£2.04	6.4%	3.06x	91.2%
Feb 2026*	264	£1,452	£5.50	6.8%	2.01x	73.8%

Brand Shopping - SC

Created Dec 11, 2024. tROAS 3x initially, activated Dec 16.

Month	Clicks	Cost	Avg CPC	CVR	ROAS	Impression Share
Dec 2024	1,112	£853	£0.77	4.6%	4.83x	52.8%
Jan 2025	1,125	£759	£0.67	5.5%	5.52x	54.0%
Feb 2025	2,704	£2,033	£0.75	3.6%	3.98x	62.7%
Mar 2025	2,538	£2,463	£0.97	5.3%	5.21x	55.5%
Apr 2025	2,726	£2,989	£1.10	3.4%	2.93x	57.9%
May 2025	3,039	£2,989	£0.98	4.2%	4.36x	57.0%
Jun 2025	2,746	£2,989	£1.09	3.1%	3.26x	55.9%
Jul 2025	2,436	£2,727	£1.12	4.3%	3.98x	56.6%
Aug 2025	1,934	£1,773	£0.92	3.8%	3.86x	59.5%
Sep 2025	2,011	£2,024	£1.01	4.1%	3.73x	62.2%
Oct 2025	1,186	£1,101	£0.93	3.1%	3.61x	64.1%
Nov 2025	1,465	£1,368	£0.93	5.0%	4.81x	63.9%
Dec 2025	1,387	£1,227	£0.88	4.4%	4.47x	61.7%
Jan 2026	1,307	£1,032	£0.79	2.9%	3.36x	71.5%
Feb 2026*	95	£94	£0.99	6.1%	3.87x	88.6%

*Partial month (11 days)

---

Period Analysis

The data splits into four distinct phases. The story is not what it first appears.

Period 1: January 2025 — Good But Already Degraded

Metric	Brand Search	Brand Shopping	Combined
CPC	£1.14	£0.67	£0.94
Clicks	1,490	1,125	2,615
Cost	£1,691	£759	£2,450
ROAS	9.88x	5.52x	8.53x
Impression Share	98.6%	54.0%	—

Both campaigns performing well. But with the 2024 baseline now available, Jan 2025 was already significantly degraded:

Metric	2024 avg (Brand Search only)	Jan 2025 (both campaigns)
Brand Search clicks/month	1,516	1,490 (-2%)
Brand Search CPC	£0.91	£1.14 (+25%)
Brand Search ROAS	9.68x	9.88x

Brand Search click volume was comparable to 2024, but CPC had already risen 25% from the 2024 average. Three factors were already in play:

• General Search (active since Oct 2024) was absorbing “hairpin legs” traffic at 1.95-2.77x ROAS
• Brand Shopping (active since Dec 16, 2024) was capturing some brand impressions
• Seasonal decline (Jan is typically lower than Nov-Dec)

---

Period 2: February - June 2025 — Cascading Failure (Tracking Break → Agency Interventions → Entrenched Damage)

Metric	Brand Search	Brand Shopping
Avg CPC	£2.99-4.21	£0.75-1.10
Monthly Cost	£2,246-3,919	£2,033-2,989
ROAS	2.62-3.61x	2.93-5.21x
IS	54.8-99.0%	55.5-62.7%

Brand Search CPCs quadruple overnight in February (£1.14 → £4.21). The impression share crashes to 54.8% in Feb before recovering. ROAS drops from 9.88x to 2.62x. This was not one event — it was a cascade where each factor compounded the original problem rather than fixing it.

Trigger: Shopify Checkout Update Broke Conversion Tracking (~Feb 5-10)

A Shopify checkout upgrade the weekend of Feb 8-9 broke the Analyzify-Purchase conversion action — the primary conversion used for bidding. Daily Analyzify-Purchase data shows the timeline precisely:

Date	Brand Search	General Search	Brand Shopping	Notes
Feb 1-4	3.3-5.0/day	1.0-7.0/day	1.0-5.0/day	Normal
Feb 5	4.0	0.1	2.0	General Search signal collapses
Feb 6	0.6	0	4.2	Brand Search signal collapses
Feb 7	0	0.1	0	Full blackout
Feb 8	1.0	0	0	Blackout continues
Feb 9	0	0	2.0	Blackout continues
Feb 10	1.0	0	1.5	First trickle
Feb 11-13	0.5-1.1	1.0-1.1	2.2-4.0	Slow recovery
Feb 14-15	1.0	2.0-2.3	3.0-5.2	Approaching normal

The change history table now covers this period. The conversion break itself (Shopify checkout upgrade) doesn’t appear in Google Ads change events, but the budget changes in response do — see below. The conversion data is unambiguous: Analyzify-Purchase went to zero across all campaigns simultaneously around Feb 7.

Critically, it wasn’t just Analyzify. GA4 Purchase (observation-only, not used for bidding) also crashed from ~10 all-conversions/day to ~1/day during the same window. Meanwhile, Analyzify events before checkout continued firing normally:

Analyzify Event	Pre-Break (Feb 1-4)	Blackout (Feb 7-10)	Affected?
Page View	233-270/day	212-342/day	No
Add To Cart	18-18/day	13-25/day	No
Begin Checkout	5.5-7.3/day	2.5-11.5/day	Partially
Purchase	6-10/day	0.1-2.5/day	Yes — ~90% drop

This confirms the Shopify checkout upgrade specifically broke the purchase confirmation step (thank-you page data layer), while pre-checkout pages were unaffected. Both Analyzify and GA4 tracking were affected simultaneously.

Budget Reallocation (Feb 5, 3:21pm)

The CPC spike started before any budget changes were made. Daily data shows:

Date	Brand Search CPC	Shopping Imps	Notes
Feb 1	£0.61	4,554	Normal
Feb 2	£0.81	5,359	Normal
Feb 3	£1.55	5,891	First CPC rise — conversion signal degrading
Feb 4	£5.19	4,726	CPC spike already happening — Shopping still on £25/day
Feb 5	£5.85	12,158	Agency triples Shopping budget at 3:21pm
Feb 6	£3.73	12,088	Shopping absorbing more brand auctions
Feb 7	£4.86	13,143	Full conversion blackout

The conversion signal was already degrading by Feb 3. ECPC reacted by Feb 4 — CPC hit £5.19 before any budget changes were made and while Shopping was still at its original £25/day budget. The Feb 5 budget response:

Campaign	Old Budget	New Budget	Change
Brand Shopping	£25/day	£50 → £80/day	+220%
Shopping Catch All	£33/day	£67 → £83/day	+150%
Brand Search	£70/day	£73/day	+4%
Shopping Top Performers	£367/day	£215/day	-41%

The Shopping budget tripling was arguably the best single move in this crisis. Shopping uses tROAS (3x), which self-regulates — it bids more widely (entering more auctions) but not more expensively. Shopping CPCs stayed at £0.55-0.87 throughout the entire crisis while Brand Search CPCs were £4-9. Shopping also had partial signals to work with: page views and add-to-cart events weren’t broken, only the purchase event, giving tROAS some data even during the blackout. The result was that Shopping caught brand traffic that Search was losing, at controlled CPCs.

Brand Shopping impressions tripled from ~4K/day to ~12-15K/day. This did increase internal auction competition for Brand Search, but that’s a trade-off: the alternative was losing those brand impressions entirely while Search was unable to win them cost-effectively.

The Shopping budget stayed elevated for several months. It went from £80 to £98 (raised further on Mar 31), and returned to £58 on Jul 23 as part of the broader restructure when General Search was paused. During this period the campaign was performing well on its own terms (controlled CPCs, tROAS-regulated), so there was no urgent signal to reduce it.

How Each Campaign Responded

The three bidding strategies each interacted with the budget changes and conversion break differently:

Brand Shopping (TARGET_ROAS 3x, budget tripled to £80) — the combination of zero conversion signal and tripled budget caused spending to explode:

• Impressions tripled: ~4K/day (Jan) → 12-15K/day (Feb 5-10)
• Clicks tripled: 28-41/day → 92-143/day
• Daily spend jumped from ~£25 to £80-93
• CPCs stayed controlled (£0.55-0.87) — the tROAS algorithm bid more widely, not more expensively
• Note: the tROAS was 3x at this point (not the current 4.5x — that changed Jul 23, 2025)

Brand Search (Manual CPC per change history, £12.01 max CPC bid, budget raised to £73) — CPC spiked in lockstep with the conversion break:

• CPC spiked: £1.55 (Feb 3) → £5.19 (Feb 4) → £5.85 (Feb 5) → £8.83 (Feb 15)
• Impressions collapsed: 62/day → 10-30/day
• Clicks halved: 39/day → 10-22/day
• The £12.01 ad group max CPC allowed these CPCs — the cap was never the constraint
• The timing resolves the ECPC question. Under pure Manual CPC (no ECPC), losing conversion data has zero effect on CPC — the bid is fixed, the auction is unchanged. The fact that CPC spiked immediately when conversions broke means some automated bidding component was reacting to the signal loss. This strongly implies Enhanced CPC was enabled — ECPC adjusts bids autonomously based on conversion likelihood, and behaves unpredictably when the conversion signal disappears. The change history logs this as “Manual CPC” because ECPC is a sub-setting, not a separate strategy type.
• Desktop bore the brunt: Desktop CPC went from ~£2.00 to £8-10+, while mobile went from ~£1.20 to £2-4

General Search (MAXIMIZE_CONVERSIONS) — completely lost its way. Optimising for conversions that no longer existed:

• CPC doubled: £1.48 → £2.36 → £4.25 (Feb 6-7)
• Impressions collapsed: 1,916 (Feb 3) → 648 (Feb 7)
• Had 4 days with zero or near-zero Analyzify conversions (Feb 5-8)
• This campaign was already cannibalizing brand terms — now it was cannibalizing them at double the CPC with no conversion signal

Daily impression share shows the timeline precisely:

Date	Brand Search IS	Brand Shopping IS	Notes
Jan 27-Feb 3	95-100%	40-64%	Normal
Feb 4	55.7%	60.4%	IS crashes — same day CPC hits £5.19
Feb 5	93.6%	62.9%	Brief recovery
Feb 6-7	43-49%	62-64%	Sustained collapse
Feb 8-10	48-58%	60-63%	Still degraded
Feb 12	23.8%	61.9%	Lowest point — only 1 in 4 auctions won
Feb 13-15	50%	56-64%	Stabilizing at half the baseline

Brand Shopping IS was completely unaffected — it held at 54-64% throughout. The auction disruption was entirely in the Search auction, where Brand Search and General Search were competing against each other (and potentially external competitors) with broken conversion signals.

Compounding Factor: General Search Was Already Cannibalizing Brand Terms

Even before the conversion break, General Search (SC) (id 70) was active and bidding on brand terms. Over its entire lifetime (Jan-Jul 2025), it captured 5,528 clicks on brand terms at £14,426 in spend — more than Brand Search itself spent in several months.

The conversion break made the cannibalization worse: General Search’s MAXIMIZE_CONVERSIONS algorithm, now blind to conversions, started bidding more aggressively on the highest-CTR terms it knew — which were brand terms. This pushed Brand Search’s auction competition higher, driving up CPCs further.

Evidence of the dual failure:

• Brand Search IS crashed to 54.8% in Feb — lost nearly half its auctions (partly to General Search, partly to reduced impressions)
• IS recovered to 96-99% by April as Brand Search’s budget expanded to compete, but at 3x the CPC cost
• Brand Shopping absorbed spillover (impressions tripled from ~4K to ~12K/day) but at low CPCs thanks to its ROAS constraint
• Even after conversion tracking recovered (~mid-Feb), CPCs remained elevated through June because the algorithms had learned elevated bid levels and General Search was still active

Search term data confirms this was pure CPC inflation on identical queries, not a search term mix change:

Term	Week Before (Jan 27-Feb 3)	Break Week (Feb 4-11)	CPC Change
”hairpin leg company” (Brand Search)	212 clicks, £110 (£0.52/clk)	113 clicks, £501 (£4.43/clk)	+8.5x
”hairpin legs” (General Search)	198 clicks, £265 (£1.34/clk)	162 clicks, £466 (£2.88/clk)	+2.1x
”hairpin leg company” (Brand Shopping)	9 clicks, £7.10 (£0.79/clk)	14 clicks, £9.70 (£0.69/clk)	Stable

The same brand term in Brand Search went from £0.52 to £4.43 per click. Shopping, bidding on the exact same term, stayed at £0.69-0.79. This is the TARGET_SPEND vulnerability in its purest form.

Subsequent Budget Adjustments (Feb 18, Mar 31)

The conversion break lasted roughly one week (~Feb 5-15). Further budget adjustments followed:

Feb 18 — Brand Search budget £73 → £87: By this point CPC was at £5-6. The additional budget headroom gave ECPC room to continue spending at the elevated rates.

Mar 31 — Brand Search budget £87 → £133 (and Brand Shopping £80 → £98): This increase was likely intended to regain impression share and volume after the disruption. The daily data shows the algorithm’s response:

Date	Brand Search CPC	Daily Spend	Notes
Mar 30	£2.60	£75	Pre-increase
Mar 31	£2.97	£92	Budget raised to £133
Apr 1	£3.91	£133	Algorithm fills new budget at higher CPC
Apr 2	£5.05	£146	Continued escalation
Apr 7	£4.01	£177	Approaching budget ceiling
Apr 8	£4.11	£185	Above budget ceiling (Google allows 2x overspend daily)

Before the increase: ~£85/day at ~£2.80 CPC. After: ~£130/day at ~£3.50 CPC. The algorithm used the extra budget to bid more aggressively rather than to win more auctions cheaply.

Monthly spend progression:

Month	Avg Daily Spend	Budget Ceiling	CPC	vs 2024 baseline
Jan 2025	£55/day	£70	£1.14	+25% CPC
Feb 2025	£80/day	£70→73→87	£4.21	+363% CPC
Mar 2025	£86/day	£87→133	£2.99	+229% CPC
Apr 2025	£128/day	£133	£3.25	+257% CPC
May 2025	£126/day	£133	£3.41	+275% CPC
Jun 2025	£96/day	£133	£2.76	+203% CPC

By April-May, Brand Search was spending £128/day at £3.25+ CPC versus the Jan baseline of £55/day at £1.14 CPC — 2.3x the spend for fewer clicks. The budget increases didn’t generate more traffic; they funded higher CPCs on existing traffic. This is a known challenge with automated bidding: the algorithm optimises to spend the available budget, so increasing budgets during a CPC spike tends to entrench the elevated rates rather than resolve them.

Learning for future disruptions: When a conversion signal breaks under automated bidding, the priority is to limit budget exposure until data recovers — reduce budgets, lower CPC ceilings, and restore them only after tracking is confirmed. The Shopping budget tripling was the right call (tROAS self-regulates). The Brand Search budget increases were intended to regain volume, but the combination of elevated CPCs and higher budgets meant the algorithm spent more at the inflated rates rather than finding cheaper clicks.

---

Period 3: July - November 2025 — Recovery (Stable)

Metric	Brand Search	Brand Shopping
Avg CPC	£1.06-1.73	£0.88-1.12
Monthly Cost	£1,080-2,356	£1,101-2,727
ROAS	4.80-10.91x	3.61-4.81x
IS	95.8-99.3%	56.6-64.1%

General Search was paused in late July 2025. The effect was immediate:

• Brand Search CPC dropped from £2.76 (Jun) → £1.66 (Jul) → £1.06 (Aug)
• Brand Search ROAS recovered to 10.91x in August — better than the January baseline
• Impression share returned to 95-99%
• August was the best month of the entire year for Brand Search efficiency

This five-month window is the true “stable” baseline for what these campaigns look like without internal competition. Brand Search CPC of £1.06-1.73 and Brand Shopping CPC of £0.88-1.12 gives a ratio of ~1.2-1.6x — actually below the industry-expected 2-3x. Both campaigns were running efficiently.

Brand Shopping volume gradually declined through this period (impressions: 280K Jul → 122K Oct) as seasonal demand shifted and the ROAS target constrained spending to efficient auctions only.

---

Period 4: December 2025 - February 2026 — Second CPC Spike (Self-Inflicted Again)

Metric	Brand Search	Brand Shopping
Avg CPC	£1.61 → £2.04 → £5.50	£0.88 → £0.79 → £0.99
Monthly Cost	£1,604 → £2,721 → £1,452*	£1,227 → £1,032 → £94*
ROAS	3.93x → 3.06x → 2.01x	4.47x → 3.36x → 3.87x
IS	96.8% → 91.2% → 73.8%	61.7% → 71.5% → 88.6%

*Feb partial (11 days)

Initially this spike appeared to be external pressure (competitor entering brand auctions). The brand-search-cpc-analysis identified the actual root cause via the Google Ads change_event API: it was self-inflicted. Auction insights confirm competition was at historically low levels during this period (Finding 9).

Root cause confirmed (from Google Ads changelog):

1. Jan 29 12:15pm — Primary domain switched from thehairpinlegcompany.co.uk to hairpin.com
2. Jan 29 12:54pm — New “Hairpin” ad group created with generic keywords “hairpin” (EXACT) and “hairpin.com” (EXACT) at £10 CPC bid. Google ignores the “.com”, so both effectively match “hairpin” — a generic word (hair accessories, road bends, crochet stitches) with no brand intent
3. Jan 29–Feb 2 — Under Maximize Conversion Value (tROAS 4x) (NOT Target Spend — the campaign was changed to Target Spend on Feb 3 as damage control), Smart Bidding tested the new generic keyword aggressively, bidding up to £14/click trying to find conversions from “hairpin” searchers. The ad URLs still pointed to the old domain (301-redirecting), gradually degrading landing page quality
4. Feb 3 — tROAS raised 4x → 5x (failed fix), then 14 minutes later switched to Target Spend with £5 CPC ceiling. This stopped the £14 blowouts but locked CPCs at ~£4.80
5. Jan 29–30 — GMC was also suspended for ~24 hours due to the domain switch, which impacted all Shopping campaigns (including Brand Shopping) but NOT Brand Search

The “competitor entry” theory was wrong. The impression share drop was caused by the campaign’s own quality signals degrading after generic keywords were added, not by a competitor bidding on brand terms.

Daily data showing the spike:

Date	Brand Search CPC	Notes
Jan 25	£0.38-0.43	Normal
Jan 26	£1.16-1.27	Normal
Jan 27	£3.10-3.25	First warning — domain prep may have begun
Jan 29	£4.28-5.26	Hairpin ad group created + domain switched
Jan 30	£9.20-16.93	Peak — Smart Bidding testing generic keyword
Jan 31	£8.66-17.79	Peak
Feb 1-2	£6.81-12.68	Settling
Feb 3	~£4.80	Target Spend switch — caps the damage
Feb 4-11	£3.71-5.62	New elevated baseline under Target Spend

Estimated excess cost: ~£1,680 over 14 days (~£120/day).

Brand Shopping volume also collapsed from late January onward (GMC suspension + seasonal). Daily Shopping clicks fell from 80+ to single digits. But Shopping CPCs barely moved (stayed £0.65-1.55) and IS actually increased to 88.6%, meaning it was capturing a larger share of a smaller pie.

The impression share data tells the same story as Period 2:

• Brand Search IS: 96.8% → 73.8% — losing auctions (quality degradation from generic keyword)
• Brand Shopping IS: 61.7% → 88.6% — gaining auctions (unaffected by the Brand Search issue)

Shopping was again naturally absorbing brand inventory that Search was being priced out of.

---

Change History Timeline

Complete bidding strategy and budget changes from the google_ads_change_events table (Feb 2024 - present). Operators: nick@ and daniel@ are from Snowball Creations (the PPC agency).

Brand Search - SC

Date	Change	Who
Pre-2024	Manual CPC, budget ~£73/day	—
Apr 17, 2024	Budget £73 → £55	nick@
Jul 22, 2024	THLC ad group max CPC £8.51 → £10.01	—
Jul 29, 2024	Budget £55 → £60	nick@
Aug 9, 2024	Budget £60 → £70	nick@
Oct 2, 2024	THLC ad group max CPC £10.01 → £12.01	—
Feb 5, 2025	Budget £70 → £73 (during conversion break)	nick@
Feb 18, 2025	Budget £73 → £87	nick@
Mar 31, 2025	Budget £87 → £142 → £133 (same day)	nick@
Jul 23, 2025	THLC ad group max CPC £12.01 → £10 (same day as General Search pause)	—
Sep 9, 2025	Budget £133 → £150	daniel@
Nov 1, 2025	Budget £150 → £200	daniel@
Nov 14, 2025	Manual CPC → Maximize Conversion Value, tROAS 4x	daniel@
Jan 29, 2026	Hairpin ad group created with £10 max CPC, generic keywords	daniel@
Feb 3, 2026	tROAS 4x → 5x, then same day → Maximize Clicks, £5 CPC ceiling	daniel@
Feb 11, 2026	CPC ceiling £5 → £2, Hairpin ad group paused	daniel@

Key insight: Brand Search was on Manual CPC for 22 months per the change history (pre-2024 through Nov 14, 2025). Whether Enhanced CPC was enabled during this period is unknown — the change history doesn’t capture ECPC state. The switch to Smart Bidding in November was a major change — within 2.5 months it led to the worst CPC spike in the account’s history. The broadly stable CPC performance of 2024 is consistent with Manual CPC (with or without ECPC) in the absence of major disruptions.

Brand Shopping - SC

Date	Change	Who
Dec 11, 2024	Campaign created: tROAS 3x, £25/day, paused immediately	daniel@
Dec 16, 2024	Activated (unpaused)	nick@
Feb 5, 2025	Budget £25 → £50 → £80 (tripled during conversion break)	nick@
Mar 31, 2025	Budget £80 → £100 → £98 (same day)	nick@
Jul 23, 2025	tROAS 3x → 4.5x, budget £98 → £58 (same day as General Search pause)	daniel@
Sep 9, 2025	Budget £58 → £75, tROAS 4.5x → 4x	daniel@
Sep 10, 2025	tROAS 4x → 4.5x (next day flip-back)	daniel@
Oct 10, 2025	Budget £75 → £35	daniel@
Nov 14, 2025	Budget £35 → £55	daniel@

Coordinated Changes

Three dates had changes across multiple campaigns — these were deliberate restructurings:

Feb 5, 2025 (during conversion break): Brand Shopping budget tripled (£25→£80), Shopping Catch All doubled (£33→£83), Shopping Top Performers cut 41% (£367→£215), Brand Search barely touched (£70→£73). All changes at 3:21-3:27pm. Note: Brand Search CPC had already spiked to £5.19 the previous day (Feb 4) before these changes — the tracking break, not the budget reallocation, was the initial trigger. But the Shopping flood compounded it (see Period 2 detail).

Mar 31, 2025: Brand Search budget jumped £87→£133, Brand Shopping £80→£98. Brand Search daily spend immediately jumped from £184 to £266-370 at higher CPCs — the algorithm filled the bigger budget at inflated rates rather than finding cheaper clicks.

Jul 23, 2025 (General Search paused): Brand Search ad group CPC dropped £12→£10, Brand Shopping tROAS tightened 3x→4.5x and budget cut £98→£58. This was a deliberate restructuring — General Search paused, Shopping tightened, Search slightly relaxed. CPC dropped 55% overnight the next day.

---

Key Findings

1. Two CPC Crises — Both Were Cascades, Not Single Events

Period 2 (Feb-Jun 2025) — Three stages:

1. Trigger (Feb 3-4): Conversion tracking started degrading. Brand Search CPC spiked from £0.81 to £5.19 on Feb 4 — before any budget changes were made and while Brand Shopping was still on its original £25/day budget. The change history records “Manual CPC” bidding, but pure Manual CPC can’t produce CPC increases from a conversion break. The timing proves Enhanced CPC was enabled, reacting to signal loss by bidding aggressively within its £12.01 ceiling.

2. Budget response (Feb 5 onward): Brand Shopping’s budget was tripled (£25→£80) — a reasonable move that kept CPCs controlled via tROAS while capturing brand traffic that Search was losing (see Period 2 for full analysis). The subsequent Brand Search budget increases (£73→£87 on Feb 18, £87→£133 on Mar 31) had the unintended effect of giving the ECPC algorithm more room to spend at elevated CPCs. By April, daily spend was £128 (vs £55 in Jan) at £3.25 CPC (vs £1.14 in Jan).

3. General Search amplified the duration: Already active on brand terms before the break, its MAXIMIZE_CONVERSIONS algorithm went blind when conversions disappeared and bid aggressively on the highest-CTR terms it knew — brand terms. This internal competition kept Brand Search CPCs elevated even after tracking recovered. Pausing General Search on Jul 23 was the actual turning point (see Period 3), and the 55% overnight CPC drop confirmed it was the main amplifier.

Period 4 (Jan-Feb 2026) — Two stages:

1. Trigger (Jan 29): Brand Search had been switched to Maximize Conversion Value (tROAS 4x) on Nov 14, 2025. On Jan 29, a new “Hairpin” ad group with the generic keyword “hairpin” (EXACT) was created. Smart Bidding tested it aggressively at up to £14/click, degrading campaign quality signals.

2. Reactive strategy switch (Feb 3): The tROAS was raised from 4x to 5x, then 14 minutes later switched to Maximize Clicks with a £5 CPC ceiling. This capped the worst blowouts but locked CPCs at ~£4.80. See brand-search-cpc-analysis.md for the full changelog.

Common thread: In both crises, the initial trigger (tracking break / generic keyword) was external or operational, not a strategic error. The key learning is that automated bidding strategies (ECPC, Smart Bidding) react unpredictably to signal disruptions, and the account lacked structural protections (hard CPC ceilings, conservative budgets) to contain the damage. The Jul 23 restructure — pausing General Search, tightening tROAS, reducing budgets — was the most effective set of changes made during the entire period.

2. Shopping CPC Is Structurally Lower Than Search — and That’s Normal

Over the full 14 months:

• Brand Search weighted avg CPC: £2.18
• Brand Shopping weighted avg CPC: £0.93
• Ratio: 2.3x

This is within the industry-expected 2-3x range. Shopping ads use a product-feed auction with different quality signals, visual product cards, and lower engagement expectations. Industry benchmarks show Search CPC ~$2.69vsShoppingCPC$0.66 globally (WordStream 2025). Branded search CPCs have risen 34% industry-wide in the last 12 months (Dreamdata).

During the stable period (Jul-Nov 2025), the ratio compressed to just 1.2-1.6x — unusually efficient for Search.

3. Cannibalization Is Real, Concentrated, and Quantifiable

Over 14 months, the top overlapping brand terms appeared in both campaigns:

Term	Search Clicks	Search Cost	Shopping Clicks	Shopping Cost	Combined
hairpin leg company	8,419	£15,509	858	£1,022	£16,531
the hairpin leg company	2,344	£3,947	187	£243	£4,190
hairpin leg co	729	£1,430	74	£69	£1,499
the hairpin leg co	333	£869	28	£36	£905
hairpin legs	52	£195	328	£358	£553
Total top 5	11,877	£21,950	1,475	£1,728	£23,678

On the dominant term “hairpin leg company”:

• Search CPC: £1.84 vs Shopping CPC: £1.19 (1.5x ratio)
• Search CVR: 8.2% vs Shopping CVR: 7.2% (Search converts marginally better)
• But the CPC difference means cost per conversion is £22.57 (Search) vs £16.60 (Shopping) — Shopping is 26% more efficient per conversion

The monthly overlap data shows this is consistent across the entire year — “hairpin leg company” appears in both campaigns every single month.

However, product-level data reveals the two campaigns serve different roles. Brand Shopping spreads clicks across 886 products with no single product exceeding 3.5% of traffic — functioning as visual product discovery within brand search results. Search drives to a single landing page. On the same brand queries, Shopping acts as a product catalogue while Search acts as brand defence. They are more complementary than duplicative.

4. Brand Search Bidding Strategy History — From Stable to Chaotic

The change history reveals the complete bidding timeline (see Change History Timeline above for full detail):

Period	Brand Search Bidding	Budget	CPC Result
Pre-2024 through Nov 14, 2025	Manual CPC (£10-12 max)	£55-200/day	£0.55-1.36 avg (22 months of stability)
Nov 14, 2025 - Feb 3, 2026	Max Conv Value (tROAS 4x)	£200/day	£1.22 → £5.50+ (escalated after generic keyword added Jan 29)
Feb 3, 2026 - present	Maximize Clicks (£2-5 cap)	£200/day	£4-5 (damage control)

The change history shows Brand Search on Manual CPC for 22 months (pre-2024 through Nov 14, 2025). During this period, CPCs were broadly stable in 2024 (£0.55-1.36 avg). The switch to Maximize Conversion Value (tROAS 4x) on Nov 14, 2025 was a major change. Within 10 weeks, a single operational error (adding a generic keyword) caused CPC to spike 4-5x because Smart Bidding had no guardrails against it.

Why was the switch made? The data shows no performance problem that warranted the change. In the weeks before Nov 14: CPC was £1.22, ROAS was 5.12x, impression share was 96-97%. The campaign was healthy. The likely motivation was budget utilisation — Brand Search was spending an average of £73/day against a £200 budget (37% utilisation). In Oct-Nov, 57 of 61 days spent under £100. From a spend perspective, the campaign looked like it was leaving money on the table.

The budget had been steadily inflated during the recovery period: £58 (Jul 23) → £133 → £150 → £200 over four months, during a period when natural brand search demand never exceeded ~£80/day. Smart Bidding was likely introduced to close the gap between budget and spend — but for a brand campaign with finite demand, underspending the budget is a feature, not a bug. There are only so many people searching “hairpin leg company” on a given day. Manual CPC was winning 97% of those auctions at £1.22. There was nothing left to find.

The oversized budget + Smart Bidding combination created the vulnerability that the Jan 29 generic keyword exploited. Smart Bidding’s response to a £200 budget and a new keyword was to test it aggressively — exactly what it’s designed to do, but catastrophic on a generic term.

Note: The change history does not capture whether Enhanced CPC (ECPC) was enabled during the Manual CPC period. However, the timing evidence strongly implies ECPC was on: Brand Search CPC spiked from £0.81 to £5.19 on Feb 4, 2025, in lockstep with the conversion tracking break. Pure Manual CPC cannot produce CPC increases from a conversion break (bids are fixed). The spike requires an automated bidding component reacting to the signal loss. See Period 2 and Finding 1 for the full analysis.

Brand Shopping has been on TARGET_ROAS throughout (3x initially, 4.5x since Jul 2025). This self-regulates: it only bids where it expects a return. When disruptions occur, Shopping bids less and wins fewer auctions — but never overpays. This is why Shopping CPC stayed in a narrow £0.67-1.12 band while Search CPC ranged from £0.55 to £5.50+.

5. Other Campaigns Are Still Leaking Brand Traffic

Beyond the two brand campaigns, other active campaigns capture brand searches:

Campaign	Status	Brand/Hairpin Clicks	Cost on Overlap	Avg CPC
General Search (SC)	Paused Jul 2025	6,625 (hairpin terms)	£17,273	£2.61
Shopping Top Performers - SC	ENABLED	2,209	£2,459	£1.11
Shopping Catch All - Over 20	ENABLED	1,140	£1,105	£0.97
Shopping Testing - SC	Paused Jul 2025	469	£706	£1.51

Shopping Top Performers and Shopping Catch All are still active and collectively capturing 3,349 brand/hairpin clicks at £3,564. This is additional cannibalization beyond the Brand Shopping campaign.

5a. General Search Deep Dive — Why It Never Worked

General Search (SC) (id 70) ran from October 2024 to July 23, 2025 (10 months), with MAXIMIZE_CONVERSIONS bidding at £120/day. It was the single largest source of wasted spend in the account.

2024 performance (Oct-Dec):

Month	Clicks	Cost	CPC	Conv	ROAS
Oct 2024	867	£1,424	£1.64	21	1.95x
Nov 2024	2,685	£4,053	£1.51	102	2.72x
Dec 2024	2,318	£4,041	£1.74	97	2.77x

Loss-making from day one. The campaign ramped quickly to £4K/month spend but never achieved breakeven ROAS in any month.

Lifetime totals (Oct 2024 - Jul 2025): £39,290 spend, 19,491 clicks, 715 conversions at 2.01x ROAS. At 40% gross margin, the campaign lost an estimated £7,700 (gross profit £31,500 vs £39,290 ad spend). It never turned a profit in any single month.

Structure: 11 ad groups, massive keyword duplication

Ad Group	Clicks	Cost	CPC	Conv	CVR	ROAS
Hairpin Legs	5,311	£13,870	£2.61	269	5.1%	1.96x
Hairpin Table Legs	1,893	£4,686	£2.48	86	4.5%	1.76x
Table Legs	2,088	£3,848	£1.84	58	2.8%	1.78x
Metal Table Legs	2,029	£3,611	£1.78	46	2.2%	2.36x
Desk Legs	467	£868	£1.86	11	2.3%	1.11x
Coffee Table Legs	496	£758	£1.53	6	1.2%	0.77x
Outdoor Table Legs	503	£693	£1.38	3	0.6%	0.26x
Dining Table Legs	405	£579	£1.43	5	1.3%	1.35x
Metal Desk Legs	284	£468	£1.65	7	2.6%	1.21x
Black Hairpin Legs	137	£380	£2.77	4	2.9%	0.52x
J Wall Hooks	8	£11	£1.32	0	0.8%	0.26x

Every ad group was below breakeven. The “Hairpin Legs” group at 1.96x was the best of a bad lot. Coffee Table Legs (0.77x), Outdoor Table Legs (0.26x), and Black Hairpin Legs (0.52x) were pure waste.

221 keywords total, with “hairpin legs”, “table legs”, and “desk legs” duplicated across 9 of 11 ad groups. The campaign also contained BROAD match keywords including “coffee”, “dining”, “outdoor”, “bench”, “chair”, “wire”, “needle”, “bobby pin”, and “paperclip” — which would have matched on irrelevant searches and wasted budget.

The fundamental problem: text ads lose to Shopping on the same terms

On the term “hairpin legs” over the same period (Jan-Jul 2025):

Campaign	Clicks	Cost	CPC	CVR	ROAS
General Search	4,077	£10,695	£2.62	4.8%	1.76x
Shopping Top Performers	421	£425	£1.01	5.8%	5.08x
Brand Shopping	149	£146	£0.98	5.3%	5.02x
Shopping Catch All	54	£59	£1.09	6.0%	10.17x

Shopping campaigns delivered 3-6x better ROAS on the exact same search term. The product listing format (visual, with price and image) naturally self-selects buyers. Text ads on generic product terms attract browsers.

General Search was spending £10,695 on “hairpin legs” at 1.76x ROAS while Shopping was delivering 5-10x ROAS on the same term for a fraction of the cost. Every click General Search won on “hairpin legs” was a click that Shopping could have won more cheaply and more profitably.

The cannibalization impact: Brand Search CPC dropped 55% when General Search was paused

Date	Brand Search CPC	Notes
Jul 10-23 (GS active)	£1.49-2.64 avg £1.75	Normal for period
Jul 23	£1.49	Last day of General Search
Jul 24	£0.67	55% drop overnight
Jul 25-31	£0.60-1.41 avg £0.94	New lower baseline
Aug 1-10	£0.38-1.18 avg £0.80	Continued improvement

Brand Search CPC dropped from £1.49 to £0.67 the day after General Search was paused. This was not gradual — it was immediate. By August, Brand Search was running at £1.06 average CPC, the most efficient it had been all year.

The mechanism: General Search and Brand Search were both eligible for “hairpin” product terms. Even though they targeted different keywords, Google’s auction considers all eligible ads from the same advertiser. With General Search bidding aggressively on overlapping terms under MAXIMIZE_CONVERSIONS, it created internal auction pressure that elevated CPCs for Brand Search too. Removing General Search from the auction immediately released that pressure.

Was General Search the main cause of Period 2’s bad performance?

No — but it was a major amplifier. The sequence was:

1. Conversion tracking broke (Feb 5-10) — this was the trigger
2. General Search (MAXIMIZE_CONVERSIONS) lost its conversion signal — it started bidding blind, aggressively, on the highest-CTR terms it knew (hairpin/brand terms)
3. This internal competition drove up Brand Search CPCs — Brand Search couldn’t win auctions cheaply because General Search was overbidding in the same space
4. Even after conversion tracking recovered (~mid-Feb), General Search kept cannibalizing — the algorithm had learned elevated bid levels and continued competing for brand/hairpin auctions
5. Budgets were pulled back around May-June as poor performance was identified
6. General Search was paused Jul 23 — Brand Search CPC dropped immediately

The conversion break would have been a ~1-2 week disruption without General Search. With General Search in the mix, the damage lasted 5 months and cost the account significantly more.

Why General Search was structurally never going to work for this account

1. Product-market fit for text ads is wrong. Hairpin leg buyers are predominantly visual shoppers — they need to see the product style, finish, and size. A text ad saying “Buy Table Legs” doesn’t communicate what makes hairpin legs different from any other table legs. Shopping ads with product images solve this naturally.

2. The non-hairpin terms were loss-making. “Table legs” (1.58x), “coffee table legs” (0.62x), “outdoor table legs” (0.38x) — these generic terms attract buyers for all types of table legs, not specifically hairpin legs. The conversion rate is low because most “table legs” searchers want standard wooden or metal legs, not a niche design product.

3. MAXIMIZE_CONVERSIONS with £120/day on 1.82x ROAS terms is a money furnace. The algorithm optimised for volume, not profitability. It found conversions — 990 of them — but at a cost that exceeded the gross profit on those sales.

4. The campaign cannibalised Shopping. Every “hairpin legs” click that General Search won at £2.62 CPC was one that Shopping could have won at £1.01 CPC with better ROAS. General Search was not finding incremental customers — it was stealing clicks from a more efficient channel and paying more for them.

6. Impression Share Tells the Whole Story

Month	Search IS	Shopping IS	Interpretation
Jan 2025	98.6%	54.0%	Baseline: Search dominant
Feb 2025	54.8%	62.7%	Conversion break + General Search cannibalizing
Mar 2025	78.4%	55.5%	Partial recovery
Apr-Jun 2025	96-99%	56-58%	Budget expanded to compete
Jul-Nov 2025	96-99%	57-64%	Stable (General Search paused)
Dec 2025	96.8%	61.7%	Stable
Jan 2026	91.2%	71.5%	Hairpin keyword + domain migration destabilize
Feb 2026	73.8%	88.6%	Quality degradation; Shopping absorbing

The inverse correlation in the last two months is the clearest evidence of Shopping backfilling Search losses. As Search gets priced out, Shopping naturally picks up the slack at lower CPCs.

7. Conversion Attribution — Both Campaigns Drive Real Purchases

Campaign	Purchase Conversions	Conv Value	Cost	ROAS
Brand Search	1,199	£138,957	£31,394	4.43x
Brand Shopping	1,074	£101,745	£24,580	4.14x

Both campaigns are attribution-credited with real Analyzify purchase conversions. GA4 Purchase is tracked as an all-conversion but excluded from bidding. The conversion counts are close (1,199 vs 1,074) despite very different traffic patterns, suggesting both campaigns reach buyers — but a significant portion of these may overlap (same customer sees both ads, converts, both get attributed).

Note: The Feb 2025 conversion tracking break (~Feb 5-10) means conversions for that week are understated across all campaigns. The monthly Feb 2025 CVR figures (Brand Search 9.5%, Brand Shopping 3.6%) reflect approximately 3 weeks of accurate tracking plus 1 week of near-zero recording. True February conversion rates were likely higher.

8. Incrementality: GSC Organic Data Shows ~25-30% Absorption

We have 7.8 million rows of Google Search Console data (Oct 2024 - Feb 2026). This lets us measure what actually happens to organic brand clicks when paid brand search retreats — a natural incrementality test across three distinct events.

Event 1: Feb 2025 — Brand Search IS crashed from 99% to 50%

Week of	Organic brand clicks	Paid brand clicks	Paid spend	Brand Search IS
Jan 6	136	704	£727	99.2%
Jan 13	133	528	£556	99.7%
Jan 20	124	598	£898	96.1%
Jan 27	114	574	£760	99.2%
Feb 3	215	334	£1,282	62.1%
Feb 10	224	212	£1,058	50.7%
Feb 17	234	224	£1,236	49.7%
Feb 24	190	278	£1,139	47.0%
Mar 3	165	396	£1,216	81.9%

Pre-break average (Jan weeks): organic ~127/week, paid ~601/week. During break (Feb weeks): organic ~216/week, paid ~262/week.

• Paid lost 339 clicks/week (-56%)
• Organic gained 89 clicks/week (+70%)
• Organic absorbed 26% of lost paid clicks. 74% was lost entirely.
• Organic CTR jumped from 1.0% to 1.8% on the same impression base — the organic result moved up the page when the paid ad above it disappeared.

Event 2: Jul 2025 — General Search paused (Jul 23)

Week of	Organic brand clicks	Paid brand spend	Brand Search IS
Jun 16	73	£1,193	—
Jun 23	89	£1,144	—
Jul 7	64	£788	—
Jul 14	96	£954	—
Jul 21	118	£766	—
Jul 28	145	£412	—
Aug 4	171	£415	—
Aug 11	138	£439	—
Aug 18	182	£479	—

Pre-pause: organic ~80/week. Post-pause: organic ~161/week — a 101% increase. Brand Search IS stayed at 99% throughout, so this isn’t Brand Search retreating. The organic lift came from General Search text ads no longer occupying the SERP for “hairpin legs” and related terms. This is evidence that General Search was suppressing organic clicks while it was running.

Event 3: Jan 2026 — CPC spike (generic keyword + domain migration)

Week of	Organic brand clicks	Paid brand clicks	Paid spend
Jan 5	156	704	£991
Jan 12	123	624	£981
Jan 19	89	596	£899
Jan 26	155	430	£2,463
Feb 2	198	332	£1,857

Pre-spike: organic ~123/week, paid ~641/week. During spike: organic ~177/week, paid ~381/week.

• Paid lost 260 clicks/week (-41%) while spend doubled (£957 → £2,160)
• Organic gained 54 clicks/week (+44%)
• Organic absorbed 21% of lost paid clicks.
• Organic CTR jumped from 1.2% to 2.6%.

Synthesis: Brand Search is ~70-75% incremental at the click level. When Brand Search retreats, organic absorbs 25-30% of the lost clicks, the rest disappears entirely. This is substantially higher incrementality than external research suggests (Haus: 0-1%, INCRMNTAL: near zero for strong organic brands). Auction insights (Finding 9) explain the gap: a persistent competitor (Pipe Dream Furniture) bids on brand terms daily at ~25-50% IS, so when Brand Search retreats, their ad captures clicks — not organic.

However, the cost per incremental click changes the picture:

CPC scenario	Nominal CPC	True cost per incremental click (÷ 0.70)	Assessment
Stable (Aug 2025)	£1.06	£1.51	Reasonable insurance
Moderate (Sep-Dec)	£1.22-1.73	£1.74-2.47	Acceptable
Inflated (Feb 2026)	£5.50	£7.86	Uneconomic

At stable CPCs (£1-1.50), Brand Search is justifiable insurance. At current inflated CPCs (£5+), the true cost per incremental click is ~£8 — far above what the traffic is worth.

9. Auction Insights: Competitor Landscape and Baseline CPC

Auction insights data (Dec 2023 - Feb 2026, daily) reveals the competitive landscape. Competition did not cause either CPC crisis — those were driven by internal factors (tracking break + ECPC in Crisis #1, generic keyword + Smart Bidding in Crisis #2). But competition does correlate with the baseline CPC level across multi-month periods.

One Persistent Competitor: Pipe Dream Furniture

pipedreamfurniture.co.uk bids on Brand Search terms nearly every day (767 of 784 days). They are the only competitor with meaningful sustained presence — the rest (Amazon, Etsy, cutmy.co.uk, etc.) appear intermittently at 10-13% IS.

Pipe Dream’s presence correlates with baseline CPC across broad periods:

Period	Pipe Dream Avg IS	Brand Search Avg CPC	Notes
Jan-Sep 2024	25%	£0.89	Stable baseline
Oct 2024 - Jan 2025	48%	£1.11	PD doubled presence; CPC crept up
Mar-Jun 2025	29%	£3.02	Post-crisis (CPC still elevated from other factors)
Jul-Nov 2025	25%	£1.28	PD back to baseline; CPC recovered
Dec 2025 - Feb 2026	19%	£2.76	PD at lowest; CPC elevated from internal causes

When Pipe Dream is at ~25% IS, baseline CPC sits around £0.89-1.28. When they doubled to ~48%, CPC crept to £1.11. But the CPC crises (£4-5+ averages) occurred independent of competitive pressure — the Dec 2025-Feb 2026 period had the lowest Pipe Dream presence in the dataset yet the highest CPCs.

Why This Matters: Incrementality

The auction insights do explain one genuinely important finding. Finding 8 measured Brand Search at ~70-75% incremental — far higher than external research suggests (Haus: 0-1%, INCRMNTAL: near zero for strong organic brands). The gap was unexplained.

Now we know: Pipe Dream is bidding on brand terms every day at ~25-50% IS. When Brand Search retreats, Pipe Dream’s ad captures those clicks — not organic. This is why organic only absorbed 25-30% of lost clicks rather than the 70-90% that incrementality research predicts for uncontested brands.

Strategic implication: Brand Search is more incremental than expected specifically because a competitor is bidding. If Pipe Dream stopped, incrementality would drop and the ROI case for Brand Search would weaken. The appropriate response is not to match their aggression (that’s how you get £5+ CPCs), but to defend brand position cost-effectively with a hard CPC ceiling.

---

Recommendations

Immediate (this week)

1. Remove “hairpin” and “hairpin.com” keywords and consider removing the entire Hairpin ad group. The Hairpin ad group was paused on Feb 11 (change history confirms), which is good. But it should be permanently removed — after removing the generic keywords, it has 315 keywords identical to the THLC ad group. If the Hairpin ad group has brand-name-specific ad copy, move those ads into the THLC ad group before deletion. (See brand-search-cpc-analysis.md for details.)

2. Switch Brand Search bidding to either Manual CPC (£1.50 max, ECPC off) or Target Impression Share (95% absolute top, £2.00 max). The 2024 data under Manual CPC shows broadly stable performance (avg CPC £0.91, ROAS 9.46x). The switch to Smart Bidding in Nov 2025 preceded the worst CPC crisis in the account’s history — though the generic keyword was the direct trigger, Smart Bidding amplified it. Either option restores a hard CPC ceiling. The CPC ceiling was already reduced to £2 on Feb 11 — the key remaining question is which bidding strategy to pair it with. Stagger this a few days after the keyword removal to isolate impact.

3. Reduce Brand Search daily budget to £60-80. The budget utilisation data is clear: in Oct-Nov 2025, Brand Search spent under £100 on 57 of 61 days, averaging £73/day against a £200 budget (37% utilisation). Even in peak January, natural brand demand never exceeded ~£105/day. The £200 budget was inherited from the recovery period (£58 → £133 → £150 → £200 over four months) when it served as headroom for regaining impression share — that purpose is long past. An oversized budget on any automated bidding strategy is an invitation to overspend: the algorithm will fill whatever room it’s given, at whatever CPC it takes. £60-80 covers natural brand demand with a comfortable margin.

Short-term (next 2-4 weeks)

4. Lower keyword CPC cap from £10 to £2. Brand terms with QS 10 should never need a £10 bid.

5. Add negative keywords to Shopping Top Performers and Shopping Catch All for core brand terms (“hairpin leg company”, “the hairpin leg company”, etc.) to stop £3,564/year in brand leakage.

6. Monitor Quality Score recovery. After removing the generic keywords, campaign quality signals should recover over 2-4 weeks. CPCs should drift back toward the £1-2 range. Don’t make further bidding changes during this period.

7. Do not use Smart Bidding on Brand Search. The Nov 14 switch from Manual CPC to Maximize Conversion Value was the strategic decision that set up the Jan 29 crisis. Smart Bidding is designed to find volume — it explores aggressively, tests new keywords at high bids, and uses the full budget to maximise its target metric. These are exactly the wrong behaviours for a brand campaign with finite, predictable demand. There are only ~1,300 people searching “hairpin leg company” per month. Manual CPC wins 97%+ of those auctions at £1.22. Smart Bidding’s response to the same situation was to chase the remaining 3% at escalating CPCs, then test a generic keyword at £14/click. Manual CPC with ECPC off and a £2 ceiling is the right strategy for a campaign where the goal is “be present for every branded search at a reasonable cost.”

8. Match budgets to demand, not ambition. Brand campaigns have natural spending ceilings dictated by search volume. If the campaign consistently spends 30-50% of its budget, that is not a problem to solve with smarter bidding — it means the budget exceeds demand. The correct response is to lower the budget to match actual spend with a 20-30% margin (e.g. £90-100 for a campaign averaging £73/day). Treating underspend as a performance issue is how the Nov 14 switch happened: a healthy campaign was moved to Smart Bidding because it “looked like” it was leaving money on the table.

Medium-term (next 1-2 months)

9. Run a controlled incrementality test. The GSC data already shows organic absorbs ~25-30% of lost paid clicks naturally (Finding 8). A deliberate 2-week pause of Brand Search (with Brand Shopping still active) would confirm this with clean data and measure the conversion impact — which the natural experiments couldn’t because they coincided with other disruptions (conversion break, CPC spikes). Monitor:

   • Organic brand CTR (should jump ~70-100% based on historical pattern)
   • Brand Shopping click volume (trend suggests it will absorb more)
   • Total brand conversions across all channels

10. Consider Brand Shopping as primary brand defense. Over 14 months it has been: half the CPC (£0.93 vs £2.18), more stable ROAS (2.93-5.52x vs 2.01-10.91x), self-regulating against competition, and trending toward higher impression share naturally.

---

Summary

Brand Search: 26-Month View (Jan 2024 - Feb 2026)

Metric	2024 (Manual CPC)	2025-26 (Mixed)
Total spend	£16,533	£31,394
Total clicks	18,188	14,811
Avg CPC	£0.91	£2.12
Total conversions	1,371	1,198
ROAS	9.68x	4.41x
Impression Share	97-99%	55-99%

Year-on-year degradation: 19% fewer clicks, 133% higher CPC, 54% lower ROAS, while spending nearly double (£16.5K → £31.4K). The 2024 baseline (Manual CPC, no competing campaigns) delivered consistent 9-10x ROAS at sub-£1 CPC. Everything since has been worse — more money for less traffic at lower returns.

Brand Search vs Brand Shopping (Jan 2025 - Feb 2026)

Metric	Brand Search	Brand Shopping
Total spend	£31,394	£24,580
Total clicks	14,811	27,403
Weighted avg CPC	£2.18	£0.93
Total conversions	1,199	1,074
CVR	8.1%	3.9%
Conv value	£138,957	£101,745
ROAS	4.43x	4.14x
Cost per conversion	£26.18	£22.88
CPC range	£1.06 - £5.50	£0.67 - £1.12
CPC stability	Volatile (5.2x range)	Stable (1.7x range)

Conclusion: The core issue is not Shopping cannibalizing Search — it’s that Brand Search has been repeatedly destabilized by cascading failures where an initial disruption compounded beyond what anyone anticipated:

1. Feb 2025 — Tracking break + ECPC: The Shopify checkout upgrade broke conversion tracking on ~Feb 5. ECPC spiked CPCs to £5+ on Feb 4 — before any budget changes could take effect. The Shopping budget tripling (£25→£80) was the right crisis move (Shopping caught brand traffic at controlled CPCs via tROAS). The subsequent Brand Search budget increases (£73→£87→£133 over 8 weeks) were intended to regain volume but had the unintended effect of funding elevated CPCs. Tighter budgets during the disruption would have contained the damage faster.
2. Feb-Jul 2025 — General Search amplified the duration: Already active on brand terms before the break, General Search (MAXIMIZE_CONVERSIONS) went blind when conversions disappeared and bid aggressively on brand terms at inflated CPCs. This internal competition kept Brand Search CPCs elevated even after tracking recovered. Pausing General Search on Jul 23 was the turning point — Brand Search CPC dropped 55% overnight, confirming General Search was the main amplifier.
3. Jul 23, 2025 — The restructure that worked: General Search was paused, Brand Shopping’s tROAS was tightened (3x→4.5x), and budgets were reduced. Brand Search recovered to £1.06 CPC by August — the most efficient month of the year. This was the most effective set of changes in the entire period.
4. Nov 2025 — Strategy switch to Smart Bidding: Manual CPC switched to Maximize Conversion Value (tROAS 4x). This removed the per-click cost control that had kept CPCs broadly stable for 22 months — a strategic bet that a smarter bidding algorithm could do better.
5. Jan 2026 — Generic keyword under Smart Bidding: The “hairpin” (EXACT) keyword addition on Jan 29 gave Smart Bidding an expensive generic term to test at up to £14/click. The reactive switch to Maximize Clicks on Feb 3 capped the worst blowouts. The CPC ceiling reduction to £2 on Feb 11 was the right follow-up.

Shopping has been the more reliable campaign throughout: cheaper per click, cheaper per conversion, stable ROAS, and self-regulating via tROAS. Every time Brand Search has been disrupted, Shopping has naturally absorbed brand inventory at lower CPCs. GSC data shows organic absorbs another 25-30% of lost Brand Search clicks. Auction insights (Finding 9) show a persistent competitor (Pipe Dream Furniture) bidding on brand terms daily, which explains the higher-than-expected incrementality — Brand Search remains ~70-75% incremental specifically because competitor ads fill the gap when it retreats.

The data suggests Shopping should be the primary brand defense, with Brand Search serving as a tightly capped safety net.

General Search (SC) separately cost the account an estimated £7,700 in net losses on £39,290 spend over 10 months (Oct 2024 - Jul 2025). It generated £79,000 in conversion value (2.01x ROAS), but at 40% gross margin that’s only £31,500 in gross profit — £7,700 less than the ad spend. It was loss-making in every month it ran, cannibalised both Brand Search and Shopping on “hairpin” terms, and its presence amplified the damage from the Feb 2025 conversion tracking break. The lesson: text ads for generic product searches do not work for this business. Shopping delivers 3-6x better ROAS on the same terms.

---

Data Gaps & Additional Sources

Completed

1. Auction Insights data — ✅ Populated for Brand Search (id 66), Dec 2023 - Feb 2026. See Finding 9. Confirms Pipe Dream Furniture as persistent brand bidder. Competition correlates with baseline CPC but did not cause either crisis. Explains the high incrementality finding (competitor ads capture clicks when Brand Search retreats).

To verify with agency

2. Enhanced CPC (ECPC) state during Manual CPC period — The change history records “Manual CPC” as the bidding strategy from pre-2024 through Nov 14, 2025, but does not capture whether Enhanced CPC was enabled. ECPC lets Google adjust bids autonomously and would explain the Feb 2025 CPC spike. Snowball Creations (nick@/daniel@) would know whether ECPC was on. This is a simple yes/no question — the timing evidence strongly supports ECPC being enabled, but formal confirmation would close it definitively.

Nice to have (incremental improvements)

2. Hour-of-day / day-of-week data — Not currently synced. Would reveal whether CPC spikes concentrate at specific times and enable ad scheduling.

3. Geographic breakdown — Not synced at campaign level. Would enable a geo-based incrementality test (pause Brand Search in Scotland only, measure organic lift vs. England control).

4. Cross-campaign attribution / user journey overlap — The tracking system has tracking_journeys and tracking_touches tables. Linking a user who saw both a Shopping ad and a Search ad to a single purchase would quantify true cannibalization at the user level, not just the query level.

5. Brand Shopping product-level breakdown — Which products are getting brand traffic in Shopping? Data available in google_ads_product_raw_daily (185K rows for Brand Shopping) but not yet analysed for this report.

What we have

• Campaign-level daily metrics — 26 months (Jan 2024 - Feb 2026). Complete for all campaigns.
• 2024 baseline data — Full year of Brand Search on Manual CPC. Establishes healthy benchmarks: £0.91 CPC, 9.46x ROAS, 97-99% IS.
• Search term overlap data — Comprehensive. Exact terms, monthly, with full cost/conversion breakdowns.
• Conversion action breakdown — Clear that only Analyzify Purchase counts for bidding.
• Device/network breakdowns — Available. No surprising patterns.
• Google Search Console (organic data) — 7.8M rows, Oct 2024 - Feb 2026. Used in Finding 8 for natural incrementality measurement.
• Google Ads change history — Persisted in google_ads_change_events table, Feb 2024 - present (1,197 events). Used to build the complete bidding strategy and budget timeline for all campaigns. Also available via live API query at /a/google/ads/change-history (29-day window).
• Auction insights (Brand Search) — Daily competitor data, Dec 2023 - Feb 2026. 22 competitors identified, Pipe Dream Furniture dominant. Used in Finding 9 for competitor landscape and incrementality explanation.

Remaining question: ECPC state (ask agency) — was Enhanced CPC enabled during the Manual CPC period? The auction insights data (Finding 9) has now confirmed competitor activity on brand terms, and the timing evidence strongly supports ECPC being enabled, but formal confirmation from the agency would close this definitively.