Google Ads Action Plan
Generated from comprehensive review of all 21 analysis documents across
audit/,baseline/,reference/, and top-level PPC docs. February 2026.
Quick Reference
Campaign IDs
| ID | Campaign Name | Type | Status | Notes |
|---|---|---|---|---|
| 66 | Brand Search - SC | Search | Paused | Was £200/day. Hairpin ad group paused Feb 11, CPC cap cut to £2 |
| 67 | Shopping Top Performers - SC | Shopping | Active | Recommended for consolidation into C71. Per-ad-group tROAS, £80/day |
| 69 | Shopping Testing - SC | Shopping | Paused (Jul 2025) | Predecessor to C71 |
| 70 | General Search (SC) | Search | Paused (Jul 2025) | MAXIMIZE_CONVERSIONS, £120/day. Largest source of wasted spend |
| 71 | Shopping Catch All >£20 | Shopping | Active | Primary campaign. ~700+ SKUs, campaign-level tROAS, £150/day |
| 72 | Brand Shopping - SC | Shopping | Active | TARGET_ROAS 4.5x, £55/day |
| 75 | PMax: Knobs Only Shopping | PMax | Paused (Dec 2025) | 0.37x ROAS, 25.5% Display waste |
| 76 | PMax: Magnetic Knife Rack | PMax | Paused (Feb 2026) | 2.02x blended, 12.2% Display waste |
| 77 | PMax: Table Tops (SC) | PMax | Active | Started Jan 19 2026. 98% Search, volatile. Under review |
| 78 | Shopping Catch All <£20 | Shopping | Active | ~115 SKUs, new (Jan 2026). £323 spend, too early to judge |
| — | Branded Bid Test | Search | Active? | Manual CPC, £60/day, zero performance data. Purpose unclear |
Key Competitors
| Competitor | Where They Appear | Notes |
|---|---|---|
| Pipe Dream Furniture | Brand Search auctions | ~25-50% impression share daily. Persistent bidder on brand terms |
Key Dates
| Date | Event |
|---|---|
| Aug 18, 2023 | Snowball Creations (SC) took over account from Adpulse (AP) |
| Feb-Jul 2025 | Conversion tracking breakdown (Shopify checkout upgrade) |
| Jul 23, 2025 | General Search paused. Brand Search CPC dropped 55% overnight |
| Nov 14, 2025 | Brand Search switched from Manual CPC to Smart Bidding |
| Dec 12-23, 2025 | PMax Knobs experiment (12 days, 0.37x ROAS) |
| Jan 19, 2026 | PMax Table Tops started |
| Jan 27, 2026 | Brand Search CPC spike (£1.50 → £6.00) |
| Jan 29, 2026 | Generic “hairpin” keyword added to Brand Search |
| Jan 29-30, 2026 | GMC suspension (domain switch to hairpin.com) |
| Feb 3, 2026 | PMax Knife Rack paused |
| Feb 11, 2026 | Hairpin ad group paused, CPC cap cut to £2 |
| Feb 12, 2026 | ATC/BC promoted to primary conversion actions |
How to Use This Document
This is the master task list for Google Ads changes. Every item links back to the full analysis document(s) that justify it. Items are split by who does the work:
- Campaign Account Changes — agency executes in Google Ads
- Shopping Feed & Product Listing Changes — internal, feed/Shopify changes
- Engineering / Tooling — internal, dashboard/sync/automation work
- Product & Business Decisions — require business owner input
Each item has a numbered ID for easy reference in meetings. Status tracks progress. Evidence gives the specific data point so nobody has to re-read the source doc to understand why.
Campaign Account Changes (Agency)
Immediate (This Week)
1. Revert ATC/BC conversion actions to secondary (or apply fractional values)
Evidence: On Feb 12, the agency promoted Add to Cart and Begin Checkout to primary conversion actions. This inflated reported ROAS by ~2.7x (account ROAS jumped to 14.76x vs 5.42x purchase-only). Smart Bidding is now optimising toward ATC events valued at £70 each when the true expected value is ~£25 (30-35% cart completion rate). Three of seven campaigns already exceed Google’s 30-conversion threshold on purchases alone — they don’t need ATC/BC signal. The full analysis recommends: revert to secondary account-wide, or if signal is needed for low-volume campaigns (PMax Table Tops, Brand Shopping, C78), apply per-campaign with fractional values (ATC: £5-10, BC: £15-20). Source: atc-bc-primary-conversion-analysis.md, knobs-and-handles.md Status: Not started
2. Add negative keywords to Shopping campaigns for brand terms (limited scope)
Evidence: The brand analysis identified £3,564/year in “brand/hairpin” clicks across Shopping Top Performers and Catch All. However, that figure includes “hairpin legs” product searches — which is core revenue-generating traffic, not leakage. The brand name (“The Hairpin Leg Co”) is one word away from the product name (“hairpin legs”). Broad or phrase negatives would be catastrophic. Safe negatives (exact match only):
[hairpin leg company],[the hairpin leg company][hairpin leg co],[the hairpin leg co][thehairpinlegcompany]NOT safe — do not add:"hairpin leg"(phrase match) — would block all product searcheshairpin leg(broad match) — would block everything with both words Google does NOT apply close variant expansion to negative keywords, so[hairpin leg co]will not accidentally block “hairpin legs”. But the recoverable spend from these exact-match negatives is likely a small fraction of the £3,564 headline figure — most of that is legitimate product traffic. Source: brand-search-vs-shopping.md Status: Not started — low priority given limited recoverable spend
Short-term (This Month)
3. Restart Brand Search with proper controls (Manual CPC)
Current state: Brand Search is currently paused. The Hairpin ad group was paused on Feb 11, CPC cap cut from £5 to £2. The campaign had spiralled to £6+ CPC due to a combination of problems. What went wrong: Generic “hairpin” keyword added Jan 29 matched hair accessories at £14/click. Smart Bidding (Maximize Conversion Value) amplified costs. Budget at £200/day was 2-3x natural demand. A separate “Branded Bid Test” campaign may have been competing for the same auctions. Restart requirements (all must be in place before re-enabling):
- Manual CPC only — no Smart Bidding. The 22-month Manual CPC period (2024-Nov 2025) delivered £0.91 avg CPC and 9.68x ROAS. Smart Bidding has failed twice on this campaign.
- Max CPC cap of £2.00-2.50 — brand terms with QS 10 should never need more
- Budget £60-80/day — natural demand never exceeded £105/day even at peak
- No generic “hairpin” or “hairpin.com” keywords — these are not brand terms
- Delete or clarify “Branded Bid Test” campaign — if it targets the same keywords, it’s competing
- Negative keywords: screwfix, ikea, b&q, amazon, and other retailer/competitor brands (£1,837/year leaked to non-brand terms at £4.31 CPC, 0.85x ROAS)
- Phrase/exact match only on core brand terms — broad match was leaking 7.2% of spend
- Alternatively, consider Target Impression Share (95% absolute top, £2.50 max CPC) instead of Manual CPC — this directly optimises for brand presence Source: brand-search-cpc-analysis.md, brand-search-vs-shopping.md, account-review-2020-2026.md, ppc-risk-and-opportunities.md Status: Not started — campaign currently paused
4. Consolidate Shopping Top Performers (C67) into Catch All (C71)
Evidence: C71 beats C67 all 14 months compared. C71 CPC £0.73 vs C67 £1.24 (45% cheaper). C71 ROAS 4.50x vs C67 2.84x. 105 overlapping search terms cause £12,784/year in cannibalization. C67 has 69 paused ad groups — structural debt. Source: shopping-top-performers-analysis.md, shopping-catch-all-analysis.md, shopping-catch-all-vs-top-performers.md, shopping-campaign-structure.md Status: Not started
5. Stabilise C71 tROAS at 300% for minimum 6 weeks
Evidence: Campaign 71’s tROAS was changed 7 times in 4 months (Jun 2025 - Feb 2026), including three changes in one week (Nov 3-9: 250% → 200% → 250% → 280%). Google’s documentation states each significant tROAS change triggers a learning phase of up to 50 conversion events or 3 conversion cycles (typically 2-4 weeks). Google also notes “small, infrequent changes may not cause noticeable disruption” — the issue here is that 7 changes in 4 months is neither small nor infrequent.
What our data shows: Weekly ROAS coefficient of variation was 29.9% in the stable period (Jul-Sep, 0 changes), 18.9% in the unstable period (Oct-Dec, 4 changes), and 12.0% in Jan 2026 (0 changes). The data does NOT clearly show that tROAS changes increased volatility — in fact, the unstable period was less volatile at weekly level. Mean weekly ROAS declined across all three periods (5.04x → 4.55x → 4.01x), but this tracks with normal seasonality. Daily ROAS swings from 0.00x to 17.80x in all periods regardless of changes — at ~100 clicks/day, daily noise completely masks any learning period signal.
The argument is risk-based, not damage-based: We cannot prove past tROAS changes caused measurable harm. But with C71 delivering 3-5x ROAS even through constant changes, the prudent move is to stop introducing unnecessary risk. Set a target based on the Shopping consolidation analysis findings, and leave it alone long enough for the algorithm to have the best possible conditions to optimise.
Source: shopping-catch-all-vs-top-performers.md, shopping-campaign-structure.md Status: Not started
6. Redirect C67 budget to C71 — increase C71 from £150 to £230/day
Evidence: C71 impression share declining (75.8% to 66.7%). C71 drives 68% of account conversions. Budget is likely the constraint on profitable impressions. Source: shopping-catch-all-analysis.md, shopping-catch-all-vs-top-performers.md Status: Not started Blocked by: Item 4
7. If keeping C67 temporarily, pause Square Industrial Legs and Tapered Oak Legs
Evidence: Square Industrial Legs at 1.73x ROAS and Tapered Oak Legs at 1.06x ROAS — both below breakeven. Source: shopping-top-performers-analysis.md Status: Not started
8. Investigate loosening C71 tROAS to capture more profitable impressions
Evidence: Impression share declining despite being the best-performing campaign. If IS <60% or ROAS <2.5x, tROAS may be too tight. Source: shopping-catch-all-analysis.md Status: Not started Blocked by: Item 5
Medium-term (Q1 2026)
9. Run Manual CPC test on top 5-10 generic search terms
Evidence: Market inflation accounts for ~2-2.5x CPC increase; Maximise Conversions adds ~1.5x on top. Manual CPC at £1.20-1.50 cap on exact match, 3-month minimum test to quantify fixable vs unfixable CPC gap. Add negatives for competitor/retailer brands from day one (screwfix, ikea, b&q, amazon — previous campaigns wasted £5,768 on these). Source: search-campaign-cpc-analysis.md, account-review-2020-2026.md Status: Not started
10. Measure Brand Search incrementality while paused
Evidence: Brand Search is currently paused — this is an unplanned incrementality test. Monitor organic traffic, Brand Shopping performance, and total brand conversions during the pause to quantify how much brand traffic converts without paid search. Pipe Dream is bidding on brand terms at ~25-50% IS, so organic won’t absorb everything — expect ~25-30% organic pickup based on prior analysis. Source: brand-search-vs-shopping.md Status: In progress — campaign currently paused, measurement opportunity
11. Review PMax Table Tops at 6 weeks (early March)
Evidence: 98% Search allocation is good, 4.20x Search ROAS is promising, but volatile and early. Needs more data before judgement. Source: pmax-table-tops-analysis.md Status: Not started
12. Evaluate category-based Shopping segmentation 4-6 weeks post-consolidation
Evidence: After C67 merges into C71 and stabilises, review whether any product categories warrant separate campaigns due to genuinely different economics (margin, AOV, conversion rate). Source: shopping-campaign-structure.md Status: Not started Blocked by: Items 4, 5
13. Set up account-level placement exclusions
Evidence: Apply across PMax, Display, YouTube, Demand Gen. Google now allows PMax campaign-level negative keywords (up to 10,000). Source: market-intelligence.md Status: Not started
14. Lean into Brand Shopping while Brand Search is paused
Evidence: Brand Search pays 2-5x higher CPC than Brand Shopping for the same queries (£2.97 vs £1.14). With Brand Search paused, Brand Shopping is the primary brand defence. Monitor whether it absorbs brand traffic cost-effectively. If it does, this may become the permanent strategy with Brand Search as a tightly capped safety net only. Source: brand-search-cpc-analysis.md, brand-search-vs-shopping.md Status: In progress — Brand Search currently paused
Shopping Feed & Product Listing Changes (Internal)
Immediate (This Week)
15. Exclude colour-based hairpin SKUs from Shopping feed
Evidence: 159 duplicate listings. Size-based variants achieve 6.32x ROAS vs colour-based at 4.91x. Colour variants fragment impressions and dilute quality signals. Source: shopping-catch-all-vs-top-performers.md Status: Not started
Short-term (This Month)
16. Add “desk” to all 71cm product titles
Evidence: 38 products / 207 SKUs with zero “desk” coverage. “Desk legs” market is +123% above pre-COVID levels and growing. These products physically are desk legs — the titles just don’t say so. Source: shopping-category-gap-analysis.md Status: Not started
17. Optimise feed titles for growing search terms
Evidence: “Dining table legs” +81% growth, “oak table legs” +910% growth. Current product titles don’t align with these search patterns. Rewrite table leg titles for long-tail queries. Source: shopping-catch-all-vs-top-performers.md, shopping-category-gap-analysis.md Status: Not started
18. Update 10cm/20cm hairpin variant titles for sofa/cabinet leg searches
Evidence: Sofa legs market +53%, cabinet legs +66% growth. Short hairpin legs are physically suited for these uses but titles don’t reflect it. Source: shopping-category-gap-analysis.md Status: Not started
Medium-term (Q1 2026)
19. Enrich feed with brand, product_type, and custom_labels
Evidence: Feed enrichment enables better segmentation, reporting, and bidding. Custom labels can flag margin tiers, seasonal products, or promotional items. Source: product-performance-decision-framework.md Status: Not started
20. Create multi-pack feed listings for knobs/handles
Evidence: 86% of knobs orders are multi-unit (avg 8 per order). Multi-pack listings with quantity pricing, bundles, “most customers buy 6” messaging would better match purchase behaviour. Source: knobs-and-handles-analysis.md, knobs-and-handles.md Status: Not started
Engineering / Tooling (Internal)
Immediate (This Week)
21. Capture campaign tROAS values in Google Ads sync
Evidence: Target ROAS values are currently NULL in the sync. Without this, we can’t track when the agency changes bidding targets or correlate tROAS changes with performance shifts. Source: shopping-catch-all-analysis.md, shopping-top-performers-analysis.md Status: Not started
22. Capture campaign config snapshots (bidding strategy, budget, status)
Evidence: Google retains only 30-day change history. Without internal snapshots, we lose the ability to correlate performance changes with config changes beyond 30 days. Source: market-intelligence.md, operational-log.md Status: Not started
Short-term (This Month)
23. Build dashboard alert: C71 impression share <60% or ROAS <2.5x
Evidence: C71 IS declining (75.8% to 66.7%) and drives 68% of account conversions. Early warning needed before this becomes a revenue problem. Source: shopping-catch-all-analysis.md Status: Not started
24. Build dashboard alert: PMax non-Search spend >5% or Display >£5/day
Evidence: PMax Knobs had 25.5% Display waste, Knife Rack had 12.2%. Need automated detection if any PMax campaign starts leaking spend to non-Search channels. Source: pmax-is-it-worth-it.md, pmax-knobs-analysis.md, pmax-knife-rack-analysis.md Status: Not started
25. Build automated 69-click zero-conversion threshold flagging
Evidence: At account CVR of ~4.3%, need ~69 clicks with zero conversions for 95% confidence a term doesn’t convert. Automate this to flag candidates for negative keywords without premature action. Source: product-performance-decision-framework.md Status: Not started
26. Compare Google DDA attribution numbers with Laravel tracking
Evidence: Google’s Data-Driven Attribution and the Laravel app’s independent tracking should roughly agree. Persistent divergence indicates a tracking problem on one side. Source: attribution.md Status: Not started
Medium-term (Q1 2026)
27. Sync product-level search term data (Shopping Products Performance Report)
Evidence: Currently no product-to-search-term mapping in the dashboard. This data enables variant-level performance analysis and feed optimisation decisions. Source: product-performance-decision-framework.md Status: Not started
28. Build monthly CTR/CVR benchmark recalculation
Evidence: Benchmarks drift over time. Automated monthly recalculation ensures product performance thresholds stay current. Source: product-performance-decision-framework.md Status: Not started
29. Build variant sibling comparison tool
Evidence: Comparing performance of size/colour variants within the same product reveals feed optimisation opportunities (e.g., colour vs size variant ROAS gap in item 24). Source: product-performance-decision-framework.md Status: Not started
30. Sync November knobs/handles data to validate reported PMax ROAS
Evidence: Agency reported 4.5x PMax ROAS for knobs in November, but local data doesn’t cover that period. Need to backfill for independent verification. Source: knobs-and-handles-analysis.md Status: Not started
31. Establish internal config snapshot system
Evidence: Google retains only 30-day change history. Need daily/weekly snapshots of campaign settings, bidding strategies, budgets, and conversion action configuration. Source: market-intelligence.md, operational-log.md Status: Not started
Product & Business Decisions
Short-term (This Month)
32. Decide: incentivise multi-unit knobs purchases
Evidence: 86% of knobs orders are multi-unit (avg 8). Quantity pricing, bundles, and “most customers buy 6” messaging could increase AOV. PMax Knife Rack failed partly due to low AOV (£53). Source: knobs-and-handles-analysis.md, knobs-and-handles.md Status: Not started
33. Decide: knobs/handles campaign strategy
Evidence: 4 campaign structures tested over 2 years, all failed. The issue is the offer (low AOV, high CPC relative to margin), not the campaign structure. Shopping passively generates 15-25 conv/quarter at 2.5-4x ROAS. A dedicated campaign, if warranted, should be Standard Shopping only, £10-15/day, tROAS 3x, purchase-only conversions, 3-month minimum. Never PMax. Source: knobs-and-handles-analysis.md, knobs-and-handles.md, shopping-campaign-structure.md Status: Not started
34. “Brass” is not viable for paid search
Evidence: Only 16 clicks on brass search terms. Volume is too thin for any paid strategy. Source: knobs-and-handles-analysis.md Status: Not started — accepted as conclusion
Medium-term (Q1 2026)
35. Build compound brand keyword strategy for “Hairpin” rebrand
Evidence: Bare “hairpin” matches hair accessories at £6+ CPC. Strategy must use compound terms: “hairpin desk legs”, “hairpin table legs”, etc. Keep legacy “hairpin leg company” keywords active during transition. Source: ppc-risk-and-opportunities.md Status: Not started
36. Quantify brand search cost increase from “Hairpin” rebrand
Evidence: Estimated +£1,650-3,300/month if bidding on “Hairpin” brand terms. This is the ongoing cost of the brand name change for paid search. Source: ppc-risk-and-opportunities.md Status: Not started
Long-term / Strategic
37. Evaluate shelf bracket range expansion
Evidence: Shelf bracket market +27% above pre-COVID and growing. Product development opportunity that would open a new paid search category. Source: shopping-category-gap-analysis.md Status: Not started
38. Evaluate sofa legs, bed legs, cabinet legs product development
Evidence: Sofa legs +53%, bed legs +55%, cabinet legs +66% market growth. Existing short hairpin legs partially serve this market but dedicated products could capture more. Source: shopping-category-gap-analysis.md Status: Not started
39. Product direction toward furniture reduces brand search dependency
Evidence: As the product range moves beyond “hairpin legs” into broader furniture hardware, organic and Shopping traffic becomes proportionally more important than brand search. This reduces the rebrand cost risk. Source: ppc-risk-and-opportunities.md Status: Not started — strategic observation
Monitoring & Review Schedule
Weekly
| Check | Trigger for Action | Source |
|---|---|---|
| C71 impression share | <60% → review tROAS and budget | shopping-catch-all-analysis.md |
| C71 ROAS trend | <2.5x for 2 consecutive weeks → investigate | shopping-catch-all-analysis.md |
| CPC creep (all campaigns) | Week-over-week increase >15% → investigate | shopping-catch-all-analysis.md |
| Brand Search CPC | >£3.00 → check for generic keyword leakage | brand-search-cpc-analysis.md |
| PMax channel allocation | Non-Search >5% or Display >£5/day → pause or investigate | pmax-is-it-worth-it.md |
| Brand Search impression share | <90% → check for new competitor bidding | market-intelligence.md |
Monthly
| Check | Action | Source |
|---|---|---|
| Quality Score recovery | Track QS trend after keyword cleanup (items 1-4) | brand-search-cpc-analysis.md |
| Knobs/handles Shopping performance | Monitor passive conversions (target: 15-25/quarter at 2.5-4x) | knobs-and-handles.md |
| CTR/CVR benchmarks | Recalculate account-wide benchmarks | product-performance-decision-framework.md |
| Conversion action configuration | Verify no unintended ATC/BC inclusion changes | shopping-catch-all-analysis.md |
| Search term report review | Flag terms at 69+ clicks with zero conversions | product-performance-decision-framework.md |
Quarterly
| Check | Action | Source |
|---|---|---|
| Amazon Shopping presence in UK | Monitor for competitive bidding pressure | market-intelligence.md |
| Google Shopping free vs paid balance | Track ratio changes | market-intelligence.md |
| UK furniture spending seasonality | Adjust budgets for seasonal demand | market-intelligence.md |
| PMax feed-only viability | Check if standard Shopping remains the only way to guarantee Shopping-only | market-intelligence.md |
Data Hygiene
- Never make decisions on last 7 days of data — attribution hasn’t matured. (attribution.md)
- Re-fetch last 30 days on every sync — already implemented in the Google Ads sync. (attribution.md)
- Exclude/annotate Jan 29-30 data — GMC suspension period, not representative. (operational-log.md)
- PMax pause recovery takes ~5 weeks — factor this lag into any restart analysis. (operational-log.md)
- Attribution model changes cause 1-2 week Smart Bidding relearning — avoid during critical periods. (attribution.md)
Strategic Principles
These are hard-won lessons from six years of account history. They should guide all future decisions.
1. Campaigns are for different economics, not different products
A new campaign is only justified when you need a different budget + bidding target combination. Splitting by product category fragments the algorithm’s data and causes cannibalization. C67/C71 overlap is the textbook example: 105 shared terms, £12,784/year wasted. Source: shopping-campaign-structure.md
2. New campaign decision flowchart
Product economics (margin, AOV) → minimum spend need → structural targeting requirement → time-limited test with predefined success criteria. If you can’t articulate why the economics differ, it doesn’t need a separate campaign. Source: shopping-campaign-structure.md
3. Never use Smart Bidding on thin-volume campaigns
Smart Bidding needs ~30 conversions/month to optimise. Below that, use Manual CPC or don’t run the campaign. Knobs/handles PMax at £93/conv vs Shopping at £18/conv illustrates the cost of ignoring this. Source: search-campaign-cpc-analysis.md, knobs-and-handles-analysis.md
4. Never use Maximize Clicks for Shopping
March 2025: C67 on Maximize Clicks produced 0.70x ROAS. The strategy optimises for volume, not value. Source: shopping-campaign-structure.md
5. Never use per-ad-group bidding in Shopping
Fragments the algorithm’s data. Campaign-level bidding with product-group exclusions is always better. Source: shopping-campaign-structure.md
6. PMax rules for any future use
Apply Display/YouTube exclusions from day one. Use MAXIMIZE_CONVERSION_VALUE not MAXIMIZE_CONVERSIONS. Set budget to match realistic Search demand. Prefer Standard Shopping over PMax for niche products. Run controlled incrementality tests. Source: pmax-is-it-worth-it.md
7. Fix the offer before fixing the campaign
Knobs/handles had 4 campaign structures tested over 2 years. All failed. The problem was the offer (low AOV, no multi-unit incentive), not the campaign. Fix the product/pricing before spending more on ads. Source: knobs-and-handles.md, shopping-campaign-structure.md
8. Document reasons for all campaign pauses
Every pause should have a recorded reason and expected recovery timeline. PMax pause recovery takes ~5 weeks for Shopping to rebuild. Source: operational-log.md
9. Declining hairpin leg trend creates brand moat
The shrinking search volume for “hairpin legs” means fewer competitors enter the space. This makes existing brand authority more valuable over time, even as absolute volume declines. Source: ppc-risk-and-opportunities.md
Document Index
All source documents referenced in this plan, with brief descriptions.
Audit (Feb 2026)
| Document | Description |
|---|---|
| _summary.md | Executive summary of the Feb 2026 audit |
| brand-search-cpc-analysis.md | Brand Search CPC spike investigation — £1.47 to £6.05, Hairpin ad group contamination |
| knobs-and-handles-analysis.md | Knobs/handles campaign performance — PMax failure, Shopping comparison, multi-unit opportunity |
| pmax-is-it-worth-it.md | PMax assessment across all product categories — Display waste, channel allocation analysis |
| pmax-knife-rack-analysis.md | PMax Knife Rack deep dive — 2.02x blended ROAS, 12.2% Display waste, low AOV |
| pmax-knobs-analysis.md | PMax Knobs deep dive — 0.37x ROAS, 25.5% Display waste |
| pmax-table-tops-analysis.md | PMax Table Tops early performance — 98% Search, 4.20x ROAS, needs more data |
| shopping-catch-all-analysis.md | C71 Catch All performance — IS decline, ROAS trend, conversion action concerns |
| shopping-top-performers-analysis.md | C67 Top Performers assessment — underperformance vs C71, consolidation case |
Baseline
| Document | Description |
|---|---|
| account-review-2020-2026.md | Six-year account history — agencies, bidding strategies, conversion tracking breakdown |
| brand-search-vs-shopping.md | Brand traffic split analysis — Search vs Shopping costs, incrementality question |
| knobs-and-handles.md | Knobs/handles full history — four failed campaign structures, offer problem diagnosis |
| search-campaign-cpc-analysis.md | Generic search CPC analysis — market inflation vs Smart Bidding markup, Manual CPC test case |
| shopping-catch-all-vs-top-performers.md | Head-to-head C67 vs C71 — 14-month comparison, cannibalization quantification |
| shopping-category-gap-analysis.md | Category gap analysis — desk legs opportunity, growing markets, feed title gaps |
Reference
| Document | Description |
|---|---|
| atc-bc-primary-conversion-analysis.md | ATC/BC primary conversion analysis — Feb 12 change, 2.7x ROAS inflation, revert recommendation |
| attribution.md | Attribution model guide — DDA maturity windows, sync best practices |
| operational-log.md | Operational event log — GMC suspension, PMax pauses, CPC spikes |
| product-performance-decision-framework.md | Decision framework — statistical thresholds, product flagging criteria, tooling requirements |
| shopping-campaign-structure.md | Campaign structure principles — when to split, when to consolidate, bidding rules |
Top-Level PPC
| Document | Description |
|---|---|
| ppc-risk-and-opportunities.md | Brand migration PPC impact — “Hairpin” rebrand cost, compound keyword strategy |
| market-intelligence.md | External market events — PMax changes, Amazon, placement exclusions, seasonality |