Knobs & Handles — Cross-Campaign Product Analysis
Summary
Knobs and handles are a higher-value category than per-unit pricing suggests. Order data shows customers buy an average of 8 units per order, giving a real AOV of ~£86 (median £46) for knob/handle-only orders — comparable to table legs. However, Google Ads performance has been poor: the dedicated PMax campaign (Dec 12–23) returned 0.37x ROAS while consuming £745, and Shopping Catch All campaigns show early promise at 2.11x ROAS but on only 4 conversions over 6 weeks. The PMax campaign likely cannibalised Shopping traffic — turning it off in January did not reduce overall knob/handle conversions.
The category is potentially viable through Shopping at the right scale. The problem was the PMax campaign structure, not the product economics.
Data Coverage
- Google Ads period: Dec 12, 2025 – Feb 10, 2026 (61 days)
- Order data: Full order history (3,033 orders containing knobs/handles)
- Missing: November 2025 Google Ads data (PMax was reportedly running at ~4.5x ROAS in Nov). Data sync begins Dec 12.
- Source:
google_ads_product_raw_daily,google_ads_raw_daily,google_ads_search_term_daily,orders(full_order JSON) - Products: 7 knobs (Bowl 40mm, Conical 18mm, Disk 30mm, Hex 30mm, Knurl 18mm, Knurl 40mm, Square 30mm) and 3 pull handles (Bar Trim, Knurl 15mm, Spiral 12mm)
- Finishes: Matt Black, Brushed Brass, Satin Copper, Industrial Nickel (knobs); same plus size variations for handles
Product Pricing & Actual Order Behaviour
Unit prices
| Product | Unit Price | Type |
|---|---|---|
| Conical 18mm / Knurl 18mm Knob | £6.95 | Knob |
| Disk 30mm / Hex 30mm / Square 30mm Knob | £7.95 | Knob |
| Bowl 40mm / Knurl 40mm Knob | £8.95 | Knob |
| Pull Handle (T-Bar 55mm) | £9.95 | Handle |
| Pull Handle (140mm) | £12.95 | Handle |
| Pull Handle (180mm) | £15.95 | Handle |
| Pull Handle (300mm) | £19.95 | Handle |
What customers actually buy (3,033 orders analysed)
| Metric | Value |
|---|---|
| Knob/handle-only orders | 84% (2,538 of 3,033) |
| Mixed orders (knobs + other products) | 16% (495) |
| Average knob/handle qty per order | 7.8 units |
| Average order subtotal | £98 |
| Average knob/handle value per order | £83 |
| Knob/handle % of total order value | 84% |
Knob/handle-only orders:
| Metric | Value |
|---|---|
| Average order value | £86.45 |
| Median order value | £45.65 |
| Median quantity | 4 units |
| Average quantity | 8.4 units |
Order value distribution (knob-only orders):
| Value Range | Orders | % |
|---|---|---|
| Under £10 | 185 | 7% |
| £10–19 | 445 | 18% |
| £20–29 | 257 | 10% |
| £30–49 | 465 | 18% |
| £50–74 | 391 | 15% |
| £75–99 | 182 | 7% |
| £100–149 | 212 | 8% |
| £150–199 | 108 | 4% |
| £200+ | 293 | 12% |
Quantity distribution (all orders):
| Qty | Orders | % | Cumulative |
|---|---|---|---|
| 1 | 423 | 14% | 14% |
| 2 | 565 | 19% | 33% |
| 3–4 | 636 | 21% | 54% |
| 5–6 | 403 | 13% | 67% |
| 7–10 | 372 | 12% | 79% |
| 11–20 | 371 | 12% | 91% |
| 21+ | 263 | 9% | 100% |
People buy knobs in quantity. Only 14% buy a single unit. Two-thirds buy 3 or more. Nearly a quarter buy 10+. This means the real AOV is not £7–9 — it’s £46–86 depending on whether you use median or mean.
Monthly order volume (organic + all channels)
| Month | Orders | Knob-Only Avg Value | Avg Qty |
|---|---|---|---|
| 2024-12 | 39 | £54 | 5.3 |
| 2025-01 | 56 | £84 | 8.1 |
| 2025-02 | 56 | £82 | 7.0 |
| 2025-03 | 69 | £80 | 7.9 |
| 2025-04 | 66 | £117 | 11.2 |
| 2025-05 | 54 | £62 | 6.8 |
| 2025-06 | 71 | £105 | 13.9 |
| 2025-07 | 55 | £107 | 9.6 |
| 2025-08 | 42 | £83 | 7.8 |
| 2025-09 | 61 | £69 | 6.1 |
| 2025-10 | 44 | £113 | 11.3 |
| 2025-11 | 55 | £77 | 6.8 |
Consistent ~50–70 orders/month. November (when PMax was reportedly at 4.5x ROAS) had 55 orders — not materially different from months before or after PMax was running. This supports the cannibalization theory.
Campaign Overview
All campaigns touching knobs & handles products
| Campaign | Type | Period | Impr | Clicks | CPC | Cost | Conv | Revenue | ROAS |
|---|---|---|---|---|---|---|---|---|---|
| PMax: Knobs Only (SC) | PMax | Dec 12–23 | 281K | 3,434 | £0.22 | £745 | 8 | £278 | 0.37x |
| Shopping Catch All <£20 | Shopping | Jan 26–Feb 10 | 4,837 | 78 | £0.71 | £55 | 3 | £208 | 3.77x |
| Brand Shopping | Shopping | Dec 12–Jan 20 | 4,232 | 50 | £0.78 | £39 | 1 | £25 | 0.64x |
| Shopping Catch All >£20 | Shopping | Dec 12–Feb 10 | 1,789 | 22 | £0.73 | £16 | 0 | £0 | 0.00x |
| TOTAL | — | Dec 12–Feb 10 | 292K | 3,584 | £0.24 | ~£855 | 12 | ~£511 | 0.60x |
The blended 0.60x ROAS is dragged down almost entirely by the PMax campaign. Shopping-only performance (excl. PMax) is £110 cost / £233 revenue = 2.11x ROAS.
PMax Channel Breakdown (Dec 12–23)
| Channel | % Spend | Clicks | Cost | CPC | Conv | Revenue | ROAS |
|---|---|---|---|---|---|---|---|
| Search | 68.9% | 374 | £514 | £1.37 | 8 | £278 | 0.54x |
| Display | 25.5% | 3,046 | £190 | £0.06 | 0 | £0 | 0.00x |
| YouTube | 5.4% | 13 | £41 | £3.12 | 0 | £0 | 0.00x |
| Partners | 0.2% | 1 | £1 | £0.54 | 0 | £0 | 0.00x |
31% of PMax budget went to non-Search channels with zero conversions. The worst Display waste ratio of any PMax campaign in the account.
The PMax Cannibalization Question
Timeline
| Period | PMax Status | PMax Budget | Shopping Active? | What Happened |
|---|---|---|---|---|
| Nov 2025 | Running | ~£10–20/day? | Yes | Reportedly ~4.5x ROAS. 55 total knob orders that month. |
| Dec 12–23 | Running | £60/day | Yes | Budget tripled+. 0.37x ROAS. 8 attributed conversions. |
| Dec 24–Jan 25 | Paused | — | Yes | Shopping Catch All handles knob traffic passively. |
| Jan 26–Feb 10 | Paused | — | Yes | Shopping <£20 picks up. 3 conversions in first week. |
What the data suggests
The PMax budget increase (Nov → Dec) did not increase conversions. If November had ~4.5x ROAS at a lower budget, and December had 0.37x ROAS at £60/day, the higher budget bought mostly junk Display/YouTube impressions. The Search portion (£514 for 8 conversions = £64/conv) was far more expensive than November’s implied cost per conversion.
Turning PMax off did not reduce conversions. Shopping Catch All <£20 picked up 3 conversions in its first two weeks of serving knobs/handles (Jan 26–Feb 10) at £55 cost, compared to PMax’s 8 conversions in 12 days at £745 cost. Even adjusting for the shorter period, the Shopping campaign’s cost per conversion (£18) is dramatically better than PMax’s (£93).
The organic order volume was stable throughout. ~55 orders/month with or without PMax running. This doesn’t prove PMax had zero incremental impact (we can’t see channel attribution for organic orders), but it does mean the £745 PMax spend didn’t produce a visible uplift in total order volume.
Conclusion: PMax was likely buying conversions that would have happened through Shopping anyway — at 5–8x the cost. The algorithm, faced with limited Search inventory for a niche product, dumped budget into Display and YouTube fill. The conversions it did capture on Search were at £64/conv — viable if the AOV is £86 (1.3x ROAS), but poor given Shopping captures similar conversions at £18/conv.
PMax suppressed Shopping during the overlap
PMax takes auction priority over Standard Shopping — this is Google’s documented behaviour. The data confirms it: during the 12 days PMax was running (Dec 12–23), Shopping Catch All served knob products with only 1–5 impressions per day and zero clicks. PMax completely suppressed Shopping from learning or serving on these products.
| Period | PMax Status | Shopping knob impressions/week | Shopping knob clicks | Shopping conversions |
|---|---|---|---|---|
| Dec 12–23 | Running | 1–5/day (~28 total) | 0 | 0 |
| Dec 24–Jan 25 | Paused | 6–93/week (ramping slowly) | 5 total | 0 |
| Jan 26–Feb 10 | Paused | 113–264/week | 34 total | 3 (7.25x ROAS) |
After PMax was paused, Shopping didn’t immediately backfill — it took 5 weeks for Shopping to start converting on knob terms. Google’s algorithm had no performance history for these products in Shopping because PMax had been claiming all the auctions. The Jan 26–Feb 10 conversions represent Shopping finally building enough data to optimise.
This is a critical finding for any future product-specific PMax campaigns: running PMax alongside Standard Shopping for niche products actively harms Shopping’s ability to learn and optimise, and the damage persists for weeks after PMax is removed.
Product-Level Performance
All campaigns combined, grouped by product (product-level data only — lower than campaign totals because PMax Display/YouTube isn’t product-attributed):
| Product | Impressions | Clicks | Cost | Conv | Revenue | ROAS |
|---|---|---|---|---|---|---|
| Knurl 15mm Pull Handle | 4,469 | 62 | £69.13 | 1 | £24.85 | 0.36x |
| Conical 18mm Knob | 152,993 | 450 | £68.78 | 1 | £34.95 | 0.51x |
| Knurl 18mm Knob | 158,890 | 644 | £60.28 | 0 | £0.00 | 0.00x |
| Disk 30mm Knob | 162,020 | 365 | £58.82 | 0 | £0.00 | 0.00x |
| Square 30mm Knob | 181,384 | 607 | £51.14 | 2 | £79.18 | 1.55x |
| Bowl 40mm Knob | 163,640 | 357 | £46.47 | 1 | £13.90 | 0.30x |
| Hex 30mm Knob | 150,211 | 471 | £32.85 | 0 | £0.00 | 0.00x |
| Bar Trim Pull Handle | 2,938 | 39 | £20.99 | 1 | £128.79 | 6.14x |
| Spiral 12mm Pull Handle | 1,133 | 17 | £6.55 | 0 | £0.00 | 0.00x |
| Knurl 40mm Knob (new) | 418 | 8 | £3.86 | 0 | £0.00 | 0.00x |
The high impression counts (150K+) are overwhelmingly from PMax’s 12-day run. Strip out PMax and most individual knobs have under 100 Shopping impressions.
Conversion Detail
| Date | Product | Campaign | Clicks | Cost | Revenue | Notes |
|---|---|---|---|---|---|---|
| Dec 14 | Knurl 15mm Pull Handle | Brand Shopping | 1 | £3.54 | £24.85 | ~2 handles |
| Dec 18 | Conical 18mm Knob | PMax | 2 | £2.34 | £34.95 | 5x £6.95 |
| Dec 22 | Bowl 40mm Knob | PMax | 10 | £7.90 | £13.90 | ~1-2 knobs (small order) |
| Jan 26 | Bar Trim Pull Handle | Shopping <£20 | 1 | £1.91 | £128.79 | Multi-item order |
| Jan 28 | Square 30mm Knob | Shopping <£20 | 1 | £0.54 | £79.18 | ~10x £7.95 |
The conversion values broadly align with the multi-unit buying pattern seen in order data. The £34.95 and £79.18 values represent 5-knob and 10-knob purchases respectively — exactly the kind of bulk buying the order data shows.
Shopping Campaign Weekly Progression (excl. PMax)
| Week | Impr | Clicks | CPC | Cost | Conv | Revenue | ROAS |
|---|---|---|---|---|---|---|---|
| Dec 12–14 | 355 | 5 | £1.06 | £5.32 | 1.0 | £24.85 | 4.67x |
| Dec 15–21 | 506 | 3 | £0.52 | £1.55 | 0 | £0 | 0.00x |
| Dec 22–28 | 554 | 8 | £0.86 | £6.88 | 0 | £0 | 0.00x |
| Dec 29–Jan 4 | 936 | 14 | £0.81 | £11.40 | 0 | £0 | 0.00x |
| Jan 5–11 | 1,029 | 15 | £0.55 | £8.22 | 0 | £0 | 0.00x |
| Jan 12–18 | 1,342 | 17 | £0.87 | £14.86 | 0 | £0 | 0.00x |
| Jan 19–25 | 1,292 | 10 | £0.68 | £6.83 | 0 | £0 | 0.00x |
| Jan 26–Feb 1 | 1,864 | 33 | £0.67 | £22.04 | 3.0 | £207.97 | 9.44x |
| Feb 2–8 | 2,273 | 25 | £0.80 | £20.00 | 0 | £0 | 0.00x |
| Feb 9–10 | 707 | 20 | £0.65 | £13.07 | 0 | £0 | 0.00x |
| Totals | 10,858 | 150 | £0.73 | £110.17 | 4 | £232.82 | 2.11x |
Impressions and clicks are growing week-on-week as Shopping algorithms learn the product data. With only 4 conversions over 6 weeks it’s too early to declare this viable, but the trajectory is positive.
Search Term Analysis
Top search terms by clicks
| Search Term | Clicks | Impressions | Cost | Conv | Revenue |
|---|---|---|---|---|---|
| copper kitchen handles | 8 | 106 | £10.52 | 0 | £0 |
| copper handles for kitchen cabinets | 4 | 4 | £4.39 | 0 | £0 |
| knurled copper kitchen handles | 3 | 7 | £5.08 | 0 | £0 |
| brushed copper kitchen handles | 3 | 11 | £2.01 | 0 | £0 |
| copper drawer handles | 2 | 2 | £3.46 | 0 | £0 |
| copper door handles | 2 | 10 | £2.10 | 0 | £0 |
| copper cupboard handles | 2 | 9 | £1.42 | 0 | £0 |
Converting search terms
| Search Term | Clicks | Cost | Conv | Revenue |
|---|---|---|---|---|
| brushed brass cabinet handles | 1 | £1.91 | 1.0 | £128.79 |
| square black knob | 1 | £0.54 | 2.0 | £79.18 |
Both from single clicks — insufficient for any statistical conclusion.
Brass-specific search terms
“Brass” terms had only 16 clicks and £12.70 total spend across the entire period. “Copper” dominates the search demand for this product type. Several “brass” terms are actually matching to wall hooks rather than knobs/handles.
Search intent observations
“Copper kitchen handles” and “cabinet handles” dominate — generic terms where Hairpin competes against B&Q, Screwfix, and Dunelm. The search terms show limited branded or product-specific demand (“knurl knob,” “hex knob” etc. don’t appear). This suggests paid search captures people looking for generic hardware, not people specifically seeking Hairpin’s design-led range. Whether that’s good (new customer acquisition) or bad (low intent match) depends on the conversion rate with more data.
Revised Assessment
The AOV changes everything
| Category | Per-Unit Price | Actual AOV | PMax ROAS | Shopping ROAS |
|---|---|---|---|---|
| Table Tops | £50–150 | ~£150 | 4.20x | Healthy |
| Knobs & Handles | £7–9 | £46–86 | 0.54x (Search) | 2.11x (early) |
| Knife Rack | £40–50 | ~£50 | 2.38x | Marginal |
At an AOV of £46–86, knobs/handles sit alongside or above knife racks in terms of order value. The PMax failure was a campaign structure problem (budget dumped into Display waste), not a product economics problem.
What needs investigation
1. Are multi-unit buyers converting through paid search? The 4 Shopping conversions show values of £25, £35, £80, and £129 — suggesting yes, multi-unit buyers are reaching the site through ads. But 4 conversions isn’t proof.
2. Is there a quantity incentive gap? The order data shows people buy in bulk, but nothing in the current ad or site experience specifically encourages this. Multi-pack pricing, quantity discounts, or “complete your kitchen” messaging could increase both conversion rate and AOV from paid traffic.
3. What was the November performance really? We don’t have Nov Google Ads data in the system. If PMax genuinely returned 4.5x ROAS at a lower budget, the product works in paid — the December failure was about budget scaling, not product fit. Getting November data synced would confirm this.
Recommendations
1. Do not restart PMax for knobs/handles. The PMax format is wrong for this product: limited Search inventory means the algorithm fills budget with Display/YouTube waste. The December experiment proved this at scale.
2. Monitor Shopping Catch All performance. The <£20 campaign is picking up knob/handle conversions at reasonable CPCs (£0.71). Give it 2–3 more months of data before deciding whether to create a dedicated Shopping campaign. Target: 20+ conversions for statistical significance.
3. Consider a dedicated Standard Shopping campaign at low budget. If Shopping Catch All continues to perform, a dedicated “Knobs & Handles” Standard Shopping campaign at £10–15/day with TARGET_ROAS bidding could scale the category without PMax’s Display bleed. This should only be tried after confirming at least 15–20 conversions through Catch All.
4. Incentivise multi-unit purchases on site. The data shows people naturally buy 4–8 units, but the site doesn’t actively encourage this. Options:
- Quantity pricing tiers (e.g., “Buy 6+ save 10%”)
- “Complete your kitchen” bundles with curated knob sets
- Cross-sell on product pages showing “most customers buy X units”
- Multi-pack listings in the Shopify feed (6-pack, 10-pack) which would also improve Shopping ad AOV visibility
5. “Brass” is not a viable paid search angle. Only 16 clicks on all brass-specific terms. “Copper” has more search demand but also hasn’t converted. Content/organic is the right channel for finish-specific marketing.
6. Sync November Google Ads data if possible. Confirming the reported 4.5x ROAS in November would validate that the product works in paid search at the right budget level, and that December’s failure was a scaling mistake rather than a fundamental problem.
Related Documents
./pmax-knobs-analysis.md— PMax: Knobs Only Shopping analysis (paused campaign)./shopping-catch-all-analysis.md— Shopping Catch All >£20 analysis./shopping-top-performers-analysis.md— Shopping Top Performers analysis