Each finding is tagged Confirmed (shown directly by the data), Observation (a pattern in the data, cause not proven) or Hypothesis (likely explanation that needs a check). No target ROAS, CPA or margin was provided, so the audit judges performance against the account's own history and against itself. It does not use an assumed break-even.
Spend grew 53% and purchases grew 56%, but revenue fell 15%. The account is buying more orders that are each worth half as much. Almost all of the lost revenue is one product line. Fuel injector sets made A$7,051 in the previous 60 days and A$1,920 in the last 60. Many of these listings are disapproved in Merchant Center for promotional overlays on their images. That includes the previous period's top seller, whose listings are 18 of 19 disapproved. Conversion tracking is healthy, so the drop is real and not a measurement problem.
| Campaign type | Spend | Share | Impr. | Clicks | Conv. | Conv. value | ROAS | CPA | CPC | CVR |
|---|---|---|---|---|---|---|---|---|---|---|
| Performance Max (2) | A$4,540.38 | 93.5% | 209,061 | 3,395 | 76.97 | A$10,702.87 | 2.36× | A$58.99 | A$1.34 | 2.27% |
| Standard Shopping (2) | A$237.28 | 4.9% | 33,284 | 1,178 | 3.98 | A$1,298.69 | 5.47× | A$59.62 | A$0.20 | 0.34% |
| Search (brand, paused) | A$80.23 | 1.7% | 136 | 58 | 0 | A$0 | 0.00× | n/a | A$1.38 | 0% |
| YouTube / Demand Gen / Display | A$0 | 0% | No active campaigns of these types | |||||||
| Account | A$4,857.89 | 100% | 242,481 | 4,631 | 80.95 | A$12,001.57 | 2.47× | A$60.01 | A$1.05 | 1.75% |
| Campaign | Class | Spend | Budget / util. | Conv. | Value | ROAS | CPA | CPC | CTR | CVR | Search IS | Lost: rank | Lost: budget |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Sales | Shopping | Low bidShopping · Max clicks · enabled | Underfunded winner* | A$113.72 | A$2/day · 95% | 3.98 | A$1,298.69 | 11.42× | A$28.55 | A$0.12 | 3.76% | 0.43% | <10% | 71.4% | 20.9% |
| P-max | All ProductsPMax · Max conversions · enabled | Deteriorating | A$4,083.30(A$3,045.05) | A$70/day · 97% | 69.98(48.05) | A$9,644.28(A$13,158.80) | 2.36×(4.32×) | A$58.35(A$63.38) | A$1.37(A$1.15) | 1.69% | 2.35% | 31.1% | 64.1% | 4.8% |
| P-max | Products Custom All productPMax · Max conversions · launched ~28 Sep · "bestseller" label | Not enough data | A$457.08 | A$10/day · ~93% | 6.99 | A$1,058.59 | 2.32× | A$65.41 | A$1.08 | 1.28% | 1.65% | 25.1% | 74.9% | 0% |
| Search | Brand | PEC AustraliaSearch · Max conversions · paused ~22 Sep | Investigate | A$80.23(A$122.77) | A$15/day | 0(4) | A$0(A$970) | 0×(7.90×) | n/a | A$1.38(A$0.73) | 42.6% | 0% | 52.0% | 7.9% | 40.1% |
| Sales | Shopping | All Products | QueenslandShopping · Max clicks · paused ~1 Oct | Loser (already paused) | A$123.56 | A$10/day | 0 | A$0 | 0× | n/a | A$0.51 | 2.88% | 0% | 36.2% | 48.6% | 15.2% |
* Low confidence: 3.98 conversions, one of them a single A$577 order. Budget utilisation = average daily spend ÷ daily budget over the days the campaign ran. 11 paused and 23 removed campaigns had no spend in either window.
None at campaign level. P-max | All Products is not a loser on the data: 2.36× with no stated break-even. It is the campaign whose ROAS fell 45% while its budget went up, and inside it the High Value Purchasers asset group returned only 1.19× on A$808.54 (section 07).
Sales | Shopping | Low bid: 11.42× on a A$2/day budget, losing 20.9% of impressions to budget. It is the only campaign where budget is a real limit. The sample is too small to scale hard (section 10).
P-max | All Products: ROAS 4.32× → 2.36×, CPC +19%, average order value roughly halved. Brand Search: CPC nearly doubled and conversions went to zero before it was paused, while brand terms now show in PMax. P-max | Custom: 7 conversions in 7 weeks is too few to judge. Decide at 30 conversions.
Only one purchase action counts in the Conversions column (Purchase SS, website tag, data-driven attribution, 30-day click window), and all three enabled campaigns bid only on the Purchase goal. The GA4 purchase import agrees within 4% on count. GA4 itself is not connected to this audit tool, so the comparison uses the GA4 purchase event as imported into Google Ads, which is GA4's own attributed count.
| Window | Ads purchases (Purchase SS) | GA4 purchases (import) | Count gap | Ads value | GA4 revenue | Value gap | Ads ROAS | GA4-based ROAS |
|---|---|---|---|---|---|---|---|---|
| Last 60 days | 80.95 | 77.86 | −3.8% | A$12,001.57 | A$11,098.11 | −7.5% | 2.47× | 2.28× |
| Previous 60 days | 52.05 | 47.59 | −8.6% | A$14,128.80 | A$11,916.07 | −15.7% | 4.46× | 3.76× |
| Bucket | Clicks | Spend | Conv. | Value |
|---|---|---|---|---|
| Brand ("pec australia", "pecaustralia", "peconline", "precision engine components australia") | 57 | A$79.66 | 0 | A$0 |
| High-intent part searches ("ba ecu", "vy bcm", "vz ss abs module", "5vzfe injectors"…) | 233 | A$47.47 | 1 | A$577 |
| Off-business / irrelevant | 0 | A$0 | 0 | A$0 |
Only converting term: vr commodore bcm (Shopping Low bid, A$0.09 → A$577). Repair queries such as "abs pump repair" are not waste, because PEC sells repair services.
| Category | Impr. | Clicks | Conv. | Value |
|---|---|---|---|---|
| a6511400360 | 2,466 | 66 | 0 | A$0 |
| ba falcon injectors | 2,946 | 38 | 0 | A$0 |
| rb30 fuel injectors | 1,219 | 28 | 0 | A$0 |
| holden captiva crank angle sensor | 1,543 | 17 | 0.6 | A$65 |
| 80 series speed sensor | 889 | 16 | 2 | A$172 |
| l67 injectors | 895 | 12 | 0 | A$0 |
| vz v6 fuel injectors | 1,335 | 10 | 0 | A$0 |
| r32 gtr cam angle sensor | 142 | 3 | 1 | A$248 |
| map sensor 4jj1 | 149 | 2 | 1 | A$70 |
| Brand terms (pec australia car parts, peconline, precision engine components…) | 65 | 19 | 0 | A$0 |
| Product line | Spend prev 60 | Value prev 60 | ROAS prev | Spend last 60 | Value last 60 | ROAS last | Value change |
|---|---|---|---|---|---|---|---|
| Injector multi-packs (4x / 6x / 8x / 12x) | A$739.95 | A$6,218.46 | 8.40× | A$248.36 | A$1,183.92 | 4.77× | −A$5,034.54 |
| Single injectors & other injector items | A$147.10 | A$832.25 | 5.66× | A$518.62 | A$735.78 | 1.42× | −A$96.47 |
| All fuel injectors | A$887.05 | A$7,050.71 | 7.95× | A$766.98 | A$1,919.70 | 2.50× | −A$5,131.01 |
| Everything else (sensors, valves, modules, brand) | A$2,283.68 | A$7,078.09 | 3.10× | A$4,090.91 | A$10,081.87 | 2.46× | +A$3,003.78 |
| Product | Spend | Conv. | Value |
|---|---|---|---|
| PEC Pre-Cat & Post-Cat O2 Sensors, Ford Territory 4.0L Turbo (Set of 2)8136796668112 | A$6.16 | 3 | A$758.10 |
| PEC Wheel Speed Sensor RR, Kia Cerato BD 2.0L8145717264592 | A$1.00 | 3 | A$699.00 |
| Reman. BCM, Holden Commodore VR (Exchange)shopify_au_7937784185040_… | A$0.31 | 1 | A$577.00 |
| 6x PEC 1300cc E85 Injector Setup, Ford FPV F6shopify_au_8135811596496_… | A$0.35 | 1 | A$497.00 |
| PEC Speed Sensor, Toyota Celica ST185 GT-Four8105528000720 | A$1.34 | 2 | A$340.11 |
| 6x PEC Fuel Injector, Commodore VZ 3.6L V6shopify_au_8135814447312_… | A$0.60 | 1 | A$328.00 |
| PEC EGR Valve, Suzuki Jimny SN413 1.3L8145620041936 | A$6.08 | 1 | A$316.00 |
| 6x PEC Injector + Connector Set, Commodore VK VN6648881578192 | A$1.03 | 1 | A$310.00 |
| 6x PEC Fuel Injectors, Toyota Hilux 4.0L 1GR-FEshopify_au_6645550186704_… | A$0.58 | 0.95 | A$295.92 |
| PEC Turbo Actuator G88, Mercedes/Audi/VW/BMW8136543502544 | A$9.64 | 0.97 | A$252.10 |
| PEC Fuel Injectors, Ford Courier PE 2.6L (Set of 4)6645518860496 | A$4.82 | 1 | A$250.00 |
| PEC Crank Angle Sensor, Nissan Skyline 2.5L8145574592720 | A$7.47 | 1 | A$248.00 |
| PEC Wheel Speed Sensor RR, Hyundai Tucson 2.0L8145717657808 | A$6.74 | 2 | A$235.00 |
| PEC EGT Sensor Mid-DPF, Mazda BT-50 3.2L8145546510544 | A$1.14 | 1 | A$230.00 |
| PEC Crank Angle Sensor, Holden Colorado RG 2.8L8105858466000 | A$25.94 | 2 | A$218.00 |
| PEC Knock Sensor, Toyota 4 Runner 3.0L8105581150416 | A$2.56 | 1 | A$211.98 |
| PEC Throttle Body, Holden Cruze JH 1.6L Turbo8136317534416 | A$3.15 | 1 | A$205.00 |
| PEC Wheel Speed Sensor RH Fr, Landcruiser VDJ78/79shopify_au_8105548546256_… | A$7.00 | 0.5 | A$165.00 |
| PEC Wheel Speed Sensor LH Fr, Land Cruiser VDJ768105552117968 | A$13.29 | 0.5 | A$165.00 |
| PEC ABS Wheel Speed Sensor Rear, Subaru Forester SG8145510924496 | A$1.03 | 1 | A$156.00 |
| Product | Clicks | Spend | Action |
|---|---|---|---|
| PEC MAP Sensor, Nissan Patrol Y61 ZD30DDTI8415854362832 | 7 | A$35.02 | Watch |
| PEC ABS Wheel Speed Sensor, BMW E46 325Ti/330Ci8145511186640 | 5 | A$23.74 | Watch |
| PEC Idle Speed Valve, Mitsubishi Verada KF–KW8105925214416 | 1 | A$23.43 | Watch (A$23 per click) |
| PEC TPS, Toyota Landcruiser FZJ100/1058105416458448 | 8 | A$21.26 | Watch |
| PEC Fuel Injector, Holden Berlina VL RB30E (Set of 6)6648871059664 | 26 | A$20.66 | Check page & price (sold A$406 prior) |
| PEC Cam Angle Sensor, Falcon AU II/III 4.0L8105853812944 | 13 | A$20.47 | Watch (sold A$85 prior) |
| PEC Air Flow Meter, Lancer EVO I–III 4G63T8105900048592 | 6 | A$20.31 | Watch |
| PEC ABS Wheel Speed Sensor Fr, VW Amarok 2H8106297589968 | 4 | A$20.22 | Watch |
| PEC VCT Actuator, Ford Falcon BA6649323946192 | 15 | A$20.00 | Check page & price |
| 4x PEC Fuel Injectors, Subaru Forester EJ2056645521744080 | 5 | A$19.97 | Watch |
| PEC ABS Wheel Speed Sensor, Colorado RC 3.6L8105619161296 | 13 | A$19.82 | Watch |
| PEC Crank Angle Sensor, Pajero NM 3.5L8136572207312 | 6 | A$19.58 | Watch |
| PEC MAP Sensor, Isuzu D-Max TF 3.0L8105620242640 | 11 | A$18.16 | Watch |
| PEC Idle Speed Valve, Galant/Lancer GSR8105907323088 | 13 | A$17.77 | Watch (sold A$222 prior) |
| PEC Air Flow Meter, Nissan Patrol 3.0L Turbo8135892009168 | 9 | A$17.23 | Watch |
| 8x PEC 550cc Injectors, Chrysler 300C 5.7 Hemi6648994037968 | 3 | A$16.65 | Watch |
| PEC EGR Valve, Hilux GUN 2.8L 1GD-FTV8145619550416 | 14 | A$16.60 | Watch |
| PEC EGR Cooler, Mazda BT-50 3.2L8136924004560 | 8 | A$16.34 | Watch |
| PEC Crank Angle Sensor, BMW 325i E36 M50B258145574133968 | 5 | A$16.17 | Watch |
| PEC TPS, Mitsubishi Challenger/Express8106332979408 | 5 | A$16.07 | Watch |
All 20 sit below 0.6× the account CPA, so none qualifies for exclusion yet. Exclude a product only after it passes about A$120 (2× CPA) with 0 purchases. In total, 53 products spent ≥A$10 without converting (A$803.87).
| Product | Spend | Value | ROAS |
|---|---|---|---|
| Reman. BCM, Commodore VR | A$0.31 | A$577.00 | 1,861× |
| 6x 1300cc E85 Injector Setup, FPV F6 | A$0.35 | A$497.00 | 1,420× |
| PEC Throttle Body, Hyundai Elantra XD | A$0.20 | A$148.00 | 740× |
| PEC Wheel Speed Sensor RR, Kia Cerato BD | A$1.00 | A$699.00 | 699× |
| 6x PEC Fuel Injector, Commodore VZ 3.6L | A$0.60 | A$328.00 | 547× |
| 6x Injector + Connector Set, VK VN | A$1.03 | A$310.00 | 301× |
| PEC Speed Sensor, Celica ST185 | A$1.34 | A$340.11 | 254× |
| PEC EGT Sensor Mid-DPF, BT-50 3.2L | A$1.14 | A$230.00 | 202× |
Each of these is a single order credited to a product with almost no spend. PMax often credits the product that was clicked last after a journey through other ads. These ROAS figures cannot be scaled as they stand.
32 products earned ≥4.9× (twice the account ROAS) on under A$5 of spend each: A$5,851 of revenue (49% of the total) from A$68 of spend. Most are single orders on vehicle-specific parts (injector sets, reman modules, wheel-speed sensors), so they cannot be "scaled" one by one. What they share is order value: 9 of the 20 most valuable orders came from products with under A$2 of spend. A value-based bid strategy and a separate high-ticket campaign (section 10) are how the account can lean into them.
6648871059664 RB30E injector set (26 clicks, A$20.66, 0 sales; it sold A$406 in the prior period). 8245669134544 Post-DPF temp sensor, LC200 (20 clicks, 0 sales). 8105529802960 Speed sensor FZJ80 (17 clicks, A$36.72, 0.5 sales). Check price and stock against competitors before cutting.
| Merchant Center 468724259, online channel | Eligible | Not eligible | Not-eligible rate |
|---|---|---|---|
| All products | 54,584 | 8,912 | 14.0% |
| Fuel injector products | 6,710 | 2,579 | 27.8% |
| Local channel (local inventory ads) | About 52,500 more listings are "pending initial review / unverified inventory". Local inventory ads were never completed. | ||
| Current title | Suggested title | Why |
|---|---|---|
| 4x Brand New * PEC * Fuel Injectors For Subaru Forester 2.0L EJ205 | PEC Fuel Injectors for Subaru Forester 2.0L EJ205 (Set of 4) | Removes asterisks and "Brand New" (disapproval risk) and puts the part and vehicle first. Matches the numeric feed's format. |
| 6x New * PEC * 750cc E85 Fuel Injector Set-up For Ford Falcon BA/BF 4.0L Turbo | PEC 750cc E85 Fuel Injector Set for Ford Falcon BA/BF 4.0L Turbo (Set of 6) | Matches the "ba falcon injectors" query (2,946 impressions) |
| 8x New * PEC * 1000CC Fuel Injectors For Holden Calais 6.0L L77 (AFM) 2010-2013 | PEC 1000cc Fuel Injectors for Holden Calais 6.0L L77 (AFM) 2010-2013 (Set of 8) | Engine code stays early, pack size moves to the end |
| New *TOP QUALITY* Throttle Position Sensor For,. Toyota Celica ST184 ST204 2.2L | Throttle Position Sensor for Toyota Celica ST184 ST204 2.2L | "*TOP QUALITY*" is already disapproved as promotional text |
| Asset group (P-max | All Products) | Products targeted | Ad strength | Spend | Conv. | Value | ROAS |
|---|---|---|---|---|---|---|
| GENERAL Shopping | label0 "google ads" × all 5 price bands | Poor | A$1,494.77 | 34.46 | A$5,136.60 | 3.44× |
| Top performing items | 134 hand-picked Shopify-format IDs | Good | A$901.36 | 11.68 | A$1,510.39 | 1.68× |
| High Value Purchasers | label0 "google ads" × all 5 price bands (same as GENERAL) | Good | A$808.54 | 7.93 | A$963.67 | 1.19× |
| REM Shopping | label0 "google ads" × all 5 price bands (same as GENERAL) | Poor | A$786.80 | 14.91 | A$1,954.62 | 2.48× |
| Search themes Shopping (paused) | label1 "pmax" | Poor | A$91.83 | 1 | A$79.00 | 0.86× |
| Campaign | Search IS | Lost to rank | Lost to budget | ROAS | Main limit | Will more budget help? |
|---|---|---|---|---|---|---|
| P-max | All Products | 31.1% | 64.1% | 4.8% | 2.36× | Rank | No Only 4.8% lost to budget. More money mostly buys more of the current low-value mix. |
| P-max | Products Custom | 25.1% | 74.9% | 0% | 2.32× | Rank | No Not budget-limited at all |
| Sales | Shopping | Low bid | <10% | 71.4% | 20.9% | 11.42× | Rank (by design: low CPC) + budget | Small test The only campaign where budget is a real limit |
| Search | Brand (paused) | 52.0% | 7.9% | 40.1% | 0× | Budget, while it ran | Decide on brand first |
| Week starting | Spend | Conv. value | ROAS |
|---|
| Metric | 2–8 Nov | 9–15 Nov | Change |
|---|---|---|---|
| Spend | A$584.93 | A$606.28 | +3.6% |
| Impressions | 29,241 | 30,414 | +4.0% |
| Clicks / CTR | 498 / 1.70% | 539 / 1.77% | +8.2% |
| CPC | A$1.17 | A$1.12 | −4.3% |
| Conversions / CVR | 10.07 / 2.02% | 7.93 / 1.47% | −21% |
| Conversion value | A$962.27 | A$1,543.93 | +60% |
| CPA / ROAS | A$58.09 / 1.65× | A$76.45 / 2.55× | ROAS +55% |
| Driver | Quantified impact | Status |
|---|---|---|
| Multi-pack injector revenue collapsed | −A$5,035 value | ConfirmedThe size of the drop. Its cause (image disapprovals) is a Hypothesis to date-check. |
| Extra spend went to lower-value categories (sensors, MAF, MAP, wheel-speed) | +A$1,807 spend → +A$3,004 value (≈1.66×) | Confirmed |
| Average order value fell | A$271 → A$148 (−45%) | Confirmed |
| PMax CPC rose while CPA held | A$1.15 → A$1.37 (+19%), CPA A$63 → A$58 | Confirmed. CPC is not why ROAS fell, because cost per order did not rise. |
| Budget scaled from about A$53 to A$81 a day | PMax ~A$300/week (Jul–Aug) → ~A$500/week (Sep–Nov), plus a new A$10/day PMax | Confirmed in spend. The budget change dates are older than the 30-day change log. |
| Brand Search paused | −A$970 brand value (prior) vs 0 now | Observation. Some of that revenue probably moved into PMax. |
| Query quality | no measurable shift | Observation (limited by PMax reporting) |
What changed: the account spent more and Google moved that spend from expensive, high-value injector sets to cheaper sensors. Each order now costs about the same (A$60) but is worth about half as much. Tracking, CPC and query quality are not the cause.
| Priority | Issue | Specific action | Expected benefit |
|---|---|---|---|
| P0 | Injector sets disapproved for image overlays (2,579 injector listings, 8,912 total online) | In Shopify and the second feed, replace overlaid images with clean product photos, starting with the injector multi-packs that sold in Jul–Sep (SR20DE 4x, RB30 6x, L67 6x, BA/BF E85 6x, Calais L77 8x, 300ZX 6x). Then request re-review in Merchant Center 468724259. In MC → Diagnostics, note the date the overlay issue first appeared. | Puts back the product line that earned 8.4× ROAS and half of the previous period's revenue. The date check confirms or rules out the cause. |
| P0 | Reporting off All conversions | Agree that ROAS is reported from the Conversions column (Purchase SS) only, and add that to the reporting template. | Stops a 5× overstatement of revenue reaching management. |
| P1 | Bidding ignores order value | About 14 days after the feed fix, switch P-max | All Products from Maximize Conversions to Maximize Conversion Value with no tROAS. It has 35 purchases per 30 days, below the ~50 needed for a stable tROAS. Add a tROAS only after 30 days at ≥50 purchases. Do this as its own step. Do not combine it with budget changes. | PMax bids more for A$300–A$1,500 orders and less for A$40 ones. Average order value should recover first, then ROAS. |
| P1 | Duplicate asset groups, one weak | Pause the High Value Purchasers asset group (A$808.54, 1.19×, same products as GENERAL at 3.44×). Do it in the same week as the feed fix, which does not touch bidding. Raise GENERAL and REM ad strength from Poor. | About A$400 a month moves to asset groups that return 2.5–3.4× on the same products. |
| P2 | Budget-limited Shopping test | Raise Sales | Shopping | Low bid from A$2 to A$5/day for 30 days. At this size the usual 20% step is A$0.40 and too small to show anything. The risk is capped at about A$90. | Tests whether 11× holds at 2.5× volume. Stop if it falls below the account's 2.5×. |
| P2 | Duplicate product sources | Choose one source (Shopify app or the numeric feed) as the one that serves. Exclude the other in Merchant Center. Add product_type to whichever stays, and standardise titles to "PEC [part] for [make model chassis engine years] (Set of N)". | One set of data per product, better relevance (rank), and no conflicting titles. |
| P3 | High-ticket items share a budget with A$40 sensors | Once value bidding is stable, launch PMax – High-ticket (price bands 200–399, 400+ and reman modules) at about A$23/day. P-max | All Products becomes PMax – Core (under A$200) at about A$52/day. Retire the bestseller PMax into the two. Full design in section 11. | Ring-fenced budget for injector sets and reman modules. |
| P3 | Goal and feed clean-up | Set the account default goal to Purchase only. Make the leftover local and call actions secondary. Fix the RAE- repair-service domain mismatch, wrong product types and landing-page errors. Finish local inventory verification or switch off the local channel. | Protects new campaigns and gives cleaner diagnostics. |
| P3 | Brand | Decide on brand coverage. Either relaunch Brand Search at about A$5/day with a brand exclusion on PMax, or leave brand in PMax. The data is too thin (136 impressions in 60 days) to call either way. | Cleaner brand vs non-brand reporting. |
| Campaign | Spend, 60 days | ROAS | Budget now | Recommended | Reason |
|---|---|---|---|---|---|
| P-max | All Products | A$4,083.30 | 2.36× | A$70/day | Hold A$70, then A$52 as PMax – Core | Rank-limited (4.8% lost to budget). Fix eligibility and bidding before adding money. |
| P-max | Products Custom | A$457.08 | 2.32× | A$10/day | Hold A$10, then retire | 7 purchases a month is too thin on its own. Its products move into High-ticket or Core by price. |
| Sales | Shopping | Low bid | A$113.72 | 11.42× | A$2/day | A$5/day (30-day test) | The only campaign where budget is a limit (20.9% lost to budget) |
| New: PMax high-ticket (A$200+) | n/a | n/a | n/a | A$23/day | P3, after value bidding. 5.35× on 17% of product spend in the last 30 days. |
| Shopping Queensland · Brand Search | A$123.56 · A$80.23 | 0× · 0× | paused | Keep paused | Brand decision is in P3 |
| Total daily budget | A$82 | A$80 (A$85 with brand) | Same total. Money moves from Core to High-ticket; the only increase is the A$3 Shopping test. |
Scenario, not a forecast: if the feed fix recovered the A$5,035 of multi-pack revenue lost between the two periods at today's spend, 60-day ROAS would be about 3.5×. There is no evidence yet of how much will come back.
The account gets about 40 purchases a month in total. Value-based bidding needs roughly 30–50 conversions per campaign to learn, so the structure separates only what really behaves differently: high-ticket sets and modules vs. everyday sensors. Splitting further (by state, device, audience or product line) would leave each campaign with too little data.
| Would-be campaign | Price-band label (custom_label_2) | Spend | Share | Purchases | Value | ROAS |
|---|---|---|---|---|---|---|
| High-ticket | 200–399, 400+, reconditioned (reman modules) | A$405.35 | 17% | 7.0 | A$2,167.27 | 5.35× |
| Core | <75, 75–94, 95–199 | A$1,624.51 | 67% | 27.9 | A$2,771.03 | 1.71× |
| Bestseller PMax | none (these products have no price label) | A$352.57 | 14% | 7.0 | A$1,058.59 | 3.00× |
| Other PMax placements (non-product) | n/a | A$51.99 | 2% | 0 | A$0 | 0.00× |
| Account | A$2,434.42 | 100% | 41.9 | A$5,996.89 | 2.46× |
| Campaign | Products | Bidding | Budget/day | Role |
|---|---|---|---|---|
| PMax – High-ticketnew | Price bands 200–399 and 400+, plus all reman modules | Maximize Conversion Value, no target (it will rarely reach 30 purchases a month) | A$23spends ~A$13.50 today | Ring-fenced budget for A$250–A$1,800 orders: injector sets, E85 setups, reman ECU/BCM/ABS |
| PMax – Coretoday's P-max | All Products, renamed | Everything under A$200, including the current bestsellers | Maximize Conversion Value. Add a ROAS target only at ≥50 purchases in 30 days. | A$52~A$68 today across Core, bestsellers and other placements | Most of the order volume and new customers. Keeps the existing campaign's learning history. |
| Shopping – Long-tail sweepertoday's Sales | Shopping | Low bid | All products | Maximize Clicks, CPC capped at about A$0.20 | A$5 | Cheap clicks across the 54,000-product long tail (11× on 4 purchases, low confidence) |
| Search – Brandoptional | "pec australia", "peconline", "precision engine components"; brand excluded from both PMax campaigns | Manual CPC or Maximize Conversions | A$5 | Clean brand vs non-brand reporting. Too little data to say it adds revenue. |
| Retire: P-max | Products Custom (bestseller) | Products move into High-ticket or Core by price | A$10 → 0 | 7 purchases a month is too thin to learn on its own |
Each group gets its own images, headlines and search themes, for example "RB30 injector set" or "Landcruiser wheel speed sensor". Audience signals go inside each group. GENERAL, High Value Purchasers and REM are retired, because they were three copies of the same products.
PMax – High-ticket (every asset group)
All products └─ google ads custom label 0 ● include ├─ 200-399 custom label 2 ● include ├─ 400+ ● include ├─ reconditioned ● include └─ Everything else in "google ads" ⊖ exclude └─ Everything else in "All products" ⊖ exclude
PMax – Core (every asset group)
All products └─ google ads custom label 0 ● include ├─ <75 custom label 2 ● include ├─ 75-94 ● include ├─ 95-199 ● include ├─ 200-399 ⊖ exclude ├─ 400+ ⊖ exclude ├─ reconditioned ⊖ exclude └─ Everything else in "google ads" ⊖ exclude └─ Everything else in "All products" ⊖ exclude
Once the category label is in the feed, split one level further (by product type or custom_label_1) to give each category asset group its own products. Change Core's tree on the same day High-ticket goes live, or the best products have nowhere to show.
| Scenario | Budget moved Core → High-ticket | Assumed return on added High-ticket spend | Assumed Core revenue lost per A$1 cut | High-ticket ROAS | Core ROAS | Account ROAS |
|---|---|---|---|---|---|---|
| Today | n/a | n/a | n/a | 5.35× | 1.71× | 2.46× |
| Conservative | A$300/mo | 2.0× | 1.7× | 3.9× | 1.7× | 2.50× |
| Base | A$300/mo | 3.0× | 1.2× | 4.4× | 1.8× | 2.69× |
| Optimistic | A$300/mo | 4.25× | 1.0× | 4.9× | 1.9× | 2.86× |
| Base, bigger shift | A$600/mo | 2.75× | 1.2× | 3.8× | 2.0× | 2.85× |
| Base + injector image fix | A$300/mo | 50–100% of the ~A$2,400/30 days of lost multi-pack revenue returns | ~3.2–3.7× | |||
The price-band split in the first table is actual data. The scenario rows are models: added spend never earns the average ROAS, and cutting Core removes its weakest spend first, so both marginal returns are assumptions that only live results can confirm. Treat the base case as a planning number and check it after 4–6 weeks.
PEC spent A$4,858 for 81 purchases and A$12,002 in revenue (2.47×), against 4.46× in the previous 60 days. Spend and order count both rose by more than 50%, but each order is now worth A$148 instead of A$271. Performance Max (93.5% of spend) moved away from high-value fuel-injector sets towards cheaper sensors. Tracking is sound, so this is a real loss of revenue.
Other identifiable issue: feed eligibility of the highest-value products. 27.8% of injector listings are disapproved, mostly for promotional overlays on images. That includes the previous period's top seller (18 of 19 SR20DE listings). Multi-pack injector revenue fell by A$5,035, more than the account's whole decline. Maximize Conversions bidding, which treats every order as equal, makes it worse. No budget, query or bid change can fix products that are not allowed to serve.
Remove the promotional overlays from fuel-injector product images and get them re-approved in Merchant Center 468724259. Start with the multi-pack sets that sold in Jul–Sep (SR20DE 4x, RB30 6x, L67 6x, BA/BF E85 6x, Calais L77 8x), then the other 2,579 disapproved injector listings. This product line earned half the account's revenue at 8.4× and is now mostly not allowed to serve.
Monitor weekly: injector not-eligible count (target under 5%), multi-pack injector conversion value, and account average order value (back above A$250). Hold off on judging ROAS until about 14 days after the listings are approved.
The Meta ad account act_2104203616411334 is brand new. Its first spend was on 16 Nov 2026, and it has spent A$35.75 in total. Two days and no purchases is not enough to judge results, so this section reviews the setup to make sure the first weeks of spend are not wasted. Re-audit performance after A$500 of spend or 30 days, whichever comes first.
| Campaign | Daily budget | Spend | Impr. | Link clicks | Landing-page views | Product views | Add-to-carts | Purchases | CPM |
|---|---|---|---|---|---|---|---|---|---|
| TR | Sales | Prospecting | Nov 2026Sales · Purchase · catalog ad "Best Selling Products" | A$28 | A$30.08 | 3,295 | 36 | 21 | 19 | 0 | 0 | A$9.13 |
| TR | Sales | Retargeting | Nov 2026Sales · Purchase · catalog ad "Top Models In-stock" | A$5 | A$5.67 | 723 | 3 | 2 | 4 | 0 | 0 | A$7.84 |
| Account | A$33 | A$35.75 | 4,018 | 39 | 23 | 23 | 0 | 0 | A$8.90 |
Pixel 217063903439069 (last fired 16 Nov 2026). Attribution: 7-day click, 1-day view, 1-day engaged view. Product views carry value (A$4,429 across 23 views, about A$193 each), so the pixel passes product values. No add-to-cart or purchase has fired from Meta traffic yet, so the purchase event is untested.
| Check | Finding | Status | Fix |
|---|---|---|---|
| Website-visitor audience | Both "All Website Visitor | 180 Days" audiences report about 20 people, although Google alone sends about 2,000 clicks a month to the site. The pixel fires, but the visitor pool is almost empty. | Broken | In Events Manager, check that the page-view event fires on every page, set Shopify's Meta data sharing to Maximum and turn on the Conversions API. Re-check the audience size in 7 days. |
| Retargeting ad set | Advantage+ audience is switched on, so Meta can show these ads to anyone, not only engagers and visitors. With an empty visitor pool it is effectively a second prospecting ad set at A$5/day. | Fix | Switch Advantage+ audience off on this ad set, or merge it into Prospecting until the visitor pool passes 1,000 people. |
| Budget vs learning | A$33/day across two campaigns optimising for Purchase is roughly 3–4 purchases a week at Google's cost per purchase (about A$60). Meta needs about 50 a week to finish learning. | Decide | Run one campaign with one ad set. At A$300 spent, if there are fewer than 3 add-to-carts, switch the optimisation event to Add to Cart or pause. |
| Exclusions | Prospecting does not exclude past customers or recent visitors, so it can spend on people who already buy. | Fix | Exclude the customer list and the 180-day visitor audience from Prospecting. |
| Customer list | A customer list of 25,400–29,900 people is uploaded, with 1%, 1–2%, 2–3% and 3% lookalikes. Prospecting uses only the 3% lookalike. | Good | Test the 1% lookalike, which is closest to real buyers. |
| Business Manager | The ad account is not attached to a Meta Business. Account, pixel and catalog ownership sit with individual logins. | Fix | Move the ad account, pixel and catalog into PEC's Business Manager and give agencies partner access. |
| Ad link | Catalog ads fall back to http://peconline.com.au/ (non-secure), the same issue as on Google. | Fix | Change the fallback URL to https://peconline.com.au/. |
| Catalog images | The catalog uses the same overlaid product images that Google disapproves. Meta allows them. | Note | When the images are cleaned for Google, the same feed fix updates Meta. |
| Old audiences | Eight audiences from 2021 are too small to use. | Note | Delete them to keep the list clean. |