The first production run of the Pause Rulebook, re read on 8 September under the revised rules. Every advertised product type in both accounts, sorted into the lists. Nothing on this page has been switched off yet. This page is the review.
The 1 September page had a Schedule list: nineteen types, $11.8K, reduced "because it is the off season." That list is retired. Ten of the nineteen were being reduced on evidence the rulebook would never accept for a switch off (laundry supplies ran at 38% last year and 5% the year before), and "off peak" was often an average or even a peak month: electronics was at 1.96x its average in July while cell phone accessories sat on the reduce list.Seasonality index. Built on the parent category (level 2), 2022 to 2025. For each year, the category's orders in a month are divided by that year's total and multiplied by 12, so 1.0 is an average month; the years are then averaged with equal weight. The peak is the highest month.index(month) = mean over years of (month orders ÷ year orders × 12)
peak = the month with the highest indexSince 8 Sep the index is only ever a shield or a push: it holds a cut back (peak guard) or flags a category to lean into (ramp). It is never a reason to reduce anything. Now every losing type faces the same test in every month of the year, and the index does two things only: it holds a switch off back when a peak is close, and it flags a category for a push when its peak is next month.
The peak guard used to look only at strongly seasonal categories (1.6x or more) and only two months out. Toy sets at Wholesale Sock Deals, a chronic loser whose category peaks in November at 1.31x, was excluded on 1 September for 90 days, straight through Black Friday. The guard now applies to every category at any amplitude: no exclusion may be live during a peak month or the month before it.Peak guard. An exclusion is live for 90 days: the run month plus the next two. If the category's indexed peak month, or the month before it (the restore ramp), lands inside those 90 days, the chronic verdict is held at Watch instead. It applies at any amplitude.months until peak = (peak month − run month) mod 12
held if months until peak ≤ 3toy sets: peak Nov, run Sep → 2 → held. linens: peak Apr, run Sep → 7 → clear to exclude. Toy sets stays on at about $3 a day and is judged again in December on its own season.
Switch off, for 90 days: mens apparel, animal toys, outdoor toys, boys apparel, linens, kitchen & dining and winter gear at All Time Trading; lamps and lanterns at Wholesale Sock Deals. $8,191 of 90 day spend, of which $2,051 is carved out and keeps serving. Each lost money in this same window two years running.Chronic loser. Lost money in this same window two years running. Last year must have real spend (at least $50) at ACoS over 50%; the year before must be over 50% too, or have no judgeable spend. Zero orders on $50+ counts as over 50%.LY spend ≥ $50 and LY ACoS > 50%
and (LY2 ACoS > 50% or LY2 spend < $50)This is the only route to a switch off, and it is the same test in every month of the year. Then the peak guard is checked before anything is applied.
None of the eight has a category peak inside the next 90 days. Every exclusion is re tested on the December run and reverts automatically unless re confirmed.Exclusion lifetime. Every exclusion expires after 90 days. To stay off, the type must fail the chronic test again on the next run's fresh window and pass the peak guard again. Nothing stays off by inertia.
Five of the seven All Time Trading exclusions peak in June and were on the old Schedule list as "seasonal, off peak." June was inside the window we judged: they ran through their own peak and lost money, for the second year running, which is the definition of chronic. The type level judgment also works the other way: kitchen & dining is excluded at All Time Trading, where it has never sold in this window, and left alone at Wholesale Sock Deals, where it converts.
Variant check passed on all eight: nothing advertised under these types has a sibling size or color listed under a type that stays on, so no traffic can slide sideways to a surviving variant.
Your sign off before we touch it: mens apparel at All Time Trading, $5,867 this window, 31% of the account's shopping spend.Share of account.$5,867 ÷ $19,166 window shopping spend = 30.6% 52% ACoS last year and 53% the year before, through its own June peak both times. Nine individually profitable products, $2,051, are carved out and keep serving.Carve out. Before an Exclude is applied, every product inside the type is checked, with variants pooled into one parent (by item group id where the feed has one, otherwise by title with the size and color words stripped). A parent with at least $50 of window spend at 35% ACoS or under is exempted and keeps serving.parent spend ≥ $50 and parent cost ÷ parent value ≤ 0.35 → keeps servingProduct data only ever keeps things on; it never adds a pause.
It is the largest single line on any list, so we want your agreement before acting on it. The exclusion covers the other $3,816 at 6% to 34% ACoS, expires in 90 days, and cannot be renewed into May, when the category's ramp begins.
Push now: personal care at Wholesale Sock Deals peaks in October, so September is its ramp month. Sixteen types, $10.2K, most already profitable at 11% to 32% ACoS. Bids and budget attention go here this month.Ramp timing. The month before a category's indexed peak is the ramp: anything held back comes back on, bids go up on the profitable types, and nothing in the category is reduced, so it enters the peak with recent data Google can bid on.ramp month = the month before the peak monthPersonal care at Wholesale Sock Deals peaks in October, so September is its ramp.
Deodorant, hair products and hand sanitizer, which the August run had reduced as off peak, are back on and inside this push. At All Time Trading, party supplies is in its peak month now; party center pieces is losing money but is held by the peak guard and stays on through it. Nothing on a ramp list is reduced, whatever its current numbers; the losing ones are re read after the peak.
Held by the peak guard: six Wholesale Sock Deals types that peak in October or November stay on and are judged in December.Peak guard. An exclusion is live for 90 days: the run month plus the next two. If the category's indexed peak month, or the month before it (the restore ramp), lands inside those 90 days, the chronic verdict is held at Watch instead. It applies at any amplitude.months until peak = (peak month − run month) mod 12
held if months until peak ≤ 3toy sets: peak Nov, run Sep → 2 → held. linens: peak Apr, run Sep → 7 → clear to exclude.
Toy sets, hanging decorations, balloons, sock pallet deals, winter sets scarves and hand sanitizer are all chronic on the numbers; the guard holds them because a 90 day switch off would run through their peak. A note on scissors: it has no orders on meaningful spend this year, but its prior year spend lands just under the $50 evidence bar in this exact window, so the rules hold it at Watch rather than stretch the evidence.Evidence bar, $50. A prior year counts as evidence only if the type spent at least $50 in that same window. Under $50 a year says nothing either way: a $19 year with no orders is not a failure. With under $50 in both prior years there is nothing to compare against, so the type waits at Watch.
Every group that clears the spend floor walks this exact path, in every month of the year. The first exit that matches wins. Seasonality never appears as a question: it only holds a cut back at the end (the peak guard) or flags a push (the ramp list, which runs alongside).
Each account's flagged groups, in the order the questions route them, plus the ramp list. ACoS is ad spend divided by the revenue it produced; at or under 35% is left alone and does not appear in the verdict lists.ACoS is ad spend divided by the conversion value those ads produced, both from Google Ads shopping performance for the product type, click date attribution.ACoS = cost ÷ conversion value
toy sets: $278 ÷ $760 = 36.6%At or under 35% a type is left alone and never enters the lists. Inside Performance Max the product level cost Google reports is a floor, so a type's true ACoS can only be higher than shown. "Last year" and "year before" are the same calendar window in 2025 and 2024. A middot means the group had under $50 of spend that year, which is too little to judge.Evidence bar, $50. A prior year counts as evidence only if the type spent at least $50 in that same window. Under $50 a year says nothing either way: a $19 year with no orders is not a failure. With under $50 in both prior years there is nothing to compare against, so the type waits at Watch. Hover or tap any Like this. Every icon on the page opens the definition or the arithmetic behind the number next to it. for the calculation behind a number.
Window spend $105,687 at 24.3% ACoS, against $81,849 at 18.3% in the same window last year.
Lost money in this same window in consecutive years, and its category is not about to peak. Switched off for 90 days, minus individually profitable products (carved out, they keep serving), then re tested.Chronic loser. Lost money in this same window two years running. Last year must have real spend (at least $50) at ACoS over 50%; the year before must be over 50% too, or have no judgeable spend. Zero orders on $50+ counts as over 50%.LY spend ≥ $50 and LY ACoS > 50%
and (LY2 ACoS > 50% or LY2 spend < $50)This is the only route to a switch off, and it is the same test in every month of the year. Then the peak guard is checked before anything is applied.
| Product type | 90d spend90d spend is the type's ad cost across the window, Shopping and Performance Max together.sum of cost for every product in the type, Jun 3 to Aug 31PMax reports product level cost as a floor, not a ceiling, so the true figure can be higher. | ACoSACoS is ad spend divided by the conversion value those ads produced, both from Google Ads shopping performance for the product type, click date attribution.ACoS = cost ÷ conversion value toy sets: $278 ÷ $760 = 36.6%At or under 35% a type is left alone and never enters the lists. Inside Performance Max the product level cost Google reports is a floor, so a type's true ACoS can only be higher than shown. | Last yearLast year / year before is the type's ACoS over the same calendar dates in 2025 and 2024, so a type is always compared to itself in season, never to the account average.LY ACoS = LY cost ÷ LY conversion value, same datesShown only where that year had at least $50 of spend; a middot means less. "no orders" means $50+ was spent with no conversion value, which counts as over any bar. | Year before | PeakPeak is the strongest month of the parent category's order index, 2022 to 2025 (the seasonality index chip under the tree shows the formula). It is listed for every type so the calendar is visible, but it only ever does two things: hold a switch off back (peak guard, any amplitude) or put the type on the ramp list. It is never a reason to reduce. |
|---|---|---|---|---|---|
| lamps and lanterns · electronics | $198 | 50% | 50% | · | Jul |
Worked in this window last year, does not now. Gets a cause before anything is switched off.Fine last year. Same dates in 2025, at least $50 of spend, ACoS at or under 35%. If it cleared the bar a year ago, something changed, and pausing would hide the symptom.LY spend ≥ $50 and LY ACoS ≤ 35% → Investigate
| Product type | 90d spend90d spend is the type's ad cost across the window, Shopping and Performance Max together.sum of cost for every product in the type, Jun 3 to Aug 31PMax reports product level cost as a floor, not a ceiling, so the true figure can be higher. | ACoSACoS is ad spend divided by the conversion value those ads produced, both from Google Ads shopping performance for the product type, click date attribution.ACoS = cost ÷ conversion value toy sets: $278 ÷ $760 = 36.6%At or under 35% a type is left alone and never enters the lists. Inside Performance Max the product level cost Google reports is a floor, so a type's true ACoS can only be higher than shown. | Last yearLast year / year before is the type's ACoS over the same calendar dates in 2025 and 2024, so a type is always compared to itself in season, never to the account average.LY ACoS = LY cost ÷ LY conversion value, same datesShown only where that year had at least $50 of spend; a middot means less. "no orders" means $50+ was spent with no conversion value, which counts as over any bar. | Year before | PeakPeak is the strongest month of the parent category's order index, 2022 to 2025 (the seasonality index chip under the tree shows the formula). It is listed for every type so the calendar is visible, but it only ever does two things: hold a switch off back (peak guard, any amplitude) or put the type on the ramp list. It is never a reason to reduce. |
|---|---|---|---|---|---|
| mens apparel · apparel | $9,961 | 39% | 30% | 20% | Apr |
| novelty toys · toys & games | $5,151 | 42% | 15% | · | Nov |
| backpacks · backpacks & luggage | $3,261 | 59% | 18% | 16% | Jul |
| school supply kits · school supplies | $1,579 | 84% | 28% | 2% | Dec |
| womens apparel · apparel | $1,375 | 61% | 10% | 24% | Apr |
| educational toys · toys & games | $1,320 | 64% | 29% | · | Nov |
| home decor · home & kitchen | $732 | 75% | 18% | 23% | Nov |
| women's footwear · footwear & shoes | $717 | 44% | 26% | 14% | Apr |
| hygiene gear · non-profit / charity | $681 | 46% | 33% | 42% | Nov |
| seating · furniture | $525 | 38% | 13% | 6% | Jul |
| christmas · seasonal items | $381 | 55% | 34% | 46% | Nov |
| cowboy & boonie hat · caps & headwear | $357 | 47% | 28% | 26% | Apr |
| sports toys · toys & games | $296 | 36% | 28% | · | Nov |
| halloween · seasonal items | $285 | 91% | 17% | no orders | Nov |
| baby diapers · baby care | $273 | no orders | 32% | · | Dec |
| party paper goods · party supplies | $260 | 127% | 17% | · | Nov |
| pens · school supplies | $259 | 141% | 11% | · | Dec |
| pencils · school supplies | $250 | 203% | 32% | · | Dec |
| poster & foam boards · school supplies | $235 | 57% | 11% | · | Dec |
| pet accessories · pet supplies | $213 | 39% | 10% | 16% | Aug |
| baby accessories · baby care | $212 | 37% | 12% | · | Dec |
| electric fans · electronics | $205 | no orders | 11% | · | Jul |
| diabetic socks · socks & hosiery | $183 | 60% | 17% | 102% | Nov |
| glue · office supplies | $164 | 44% | 2% | · | Jul |
Losing money now but not two years running, not enough history yet, or chronic but held by the peak guard. Keeps serving, checked again next month.Watch holds three kinds of type: no judgeable history in either prior year (under $50 both years); bad now but not chronic, which since 8 Sep includes every off season type that used to be reduced on the old Schedule list; and chronic but held by the peak guard. All three keep serving. It costs little to wait a month for a clean reading, and it costs real sales to cut demand on weak evidence.
| Product type | 90d spend90d spend is the type's ad cost across the window, Shopping and Performance Max together.sum of cost for every product in the type, Jun 3 to Aug 31PMax reports product level cost as a floor, not a ceiling, so the true figure can be higher. | ACoSACoS is ad spend divided by the conversion value those ads produced, both from Google Ads shopping performance for the product type, click date attribution.ACoS = cost ÷ conversion value toy sets: $278 ÷ $760 = 36.6%At or under 35% a type is left alone and never enters the lists. Inside Performance Max the product level cost Google reports is a floor, so a type's true ACoS can only be higher than shown. | Last yearLast year / year before is the type's ACoS over the same calendar dates in 2025 and 2024, so a type is always compared to itself in season, never to the account average.LY ACoS = LY cost ÷ LY conversion value, same datesShown only where that year had at least $50 of spend; a middot means less. "no orders" means $50+ was spent with no conversion value, which counts as over any bar. | Year before | PeakPeak is the strongest month of the parent category's order index, 2022 to 2025 (the seasonality index chip under the tree shows the formula). It is listed for every type so the calendar is visible, but it only ever does two things: hold a switch off back (peak guard, any amplitude) or put the type on the ramp list. It is never a reason to reduce. |
|---|---|---|---|---|---|
| laundry supplies · home & kitchen | $785 | 41% | 38% | 5% | Nov |
| deodorant · personal care | $650 | 52% | 40% | 12% | Oct |
| boys socks · socks & hosiery | $604 | 59% | 77% | 5% | Nov |
| outdoor recreation · sporting and outdoors | $559 | 81% | 49% | 112% | Dec |
| bundle care sets · non-profit / charity | $546 | 40% | · | · | Nov |
| hair products · personal care | $453 | 99% | 38% | 34% | Oct |
| cell phone accessories · electronics | $360 | 73% | 71% | 4% | Jul |
| bandanas · caps & headwear | $324 | 59% | 36% | 16% | Apr |
| hanging decorations & cut out · party supplies peak guard | $280 | 560% | 111% | · | Nov |
| toy sets · toys & games peak guard | $278 | 37% | 90% | 320% | Nov |
| hand sanitizer · personal care peak guard | $275 | 41% | no orders | · | Oct |
| event planning gear · non-profit / charity | $233 | 218% | · | 35% | Nov |
| balloons & balloon holder · party supplies peak guard | $202 | 53% | 65% | · | Nov |
| led party supplies · party supplies | $194 | 43% | · | · | Nov |
| sock pallet deals · socks & hosiery peak guard | $183 | no orders | 150% | no orders | Nov |
| baby apparel · baby care | $182 | 61% | 92% | 10% | Dec |
| liquidation center · non-profit / charity | $174 | no orders | · | · | Nov |
| pest control · home & kitchen | $173 | 36% | 46% | 41% | Nov |
| scissors · school supplies | $168 | 203% | · | no orders | Dec |
| winter sets scarves · winter wear peak guard | $167 | 44% | 221% | 150% | Nov |
| flag · office supplies | $159 | 134% | · | · | Jul |
Category peaks next month, or is in its peak now. Push, not pause: lift bids on the profitable types, make sure the rest are serving, so the category meets its demand with recent data. Runs alongside the verdicts, so a Watch item can also be here.Ramp list. Every type whose category peaks next month, or is in its peak month now, with at least $50 of window spend and a peak that repeats: the category was above its average in that month in all, or all but one, of the index years, so a single year's fluke does not get budget thrown at it.months until peak = (peak month − run month) mod 12
on the list if ≤ 1, and above average at the peak in ≥ (years − 1) of the index yearsThe action is a push, not a return to neutral: lift bids and check budget share on the profitable ones, make sure the losing ones are at least serving, so the category enters its peak with recent data. Types on this list are re read after the peak, never cut into it.
| Product type | 90d spend90d spend is the type's ad cost across the window, Shopping and Performance Max together.sum of cost for every product in the type, Jun 3 to Aug 31PMax reports product level cost as a floor, not a ceiling, so the true figure can be higher. | ACoSACoS is ad spend divided by the conversion value those ads produced, both from Google Ads shopping performance for the product type, click date attribution.ACoS = cost ÷ conversion value toy sets: $278 ÷ $760 = 36.6%At or under 35% a type is left alone and never enters the lists. Inside Performance Max the product level cost Google reports is a floor, so a type's true ACoS can only be higher than shown. | PeakSeasonality index. Built on the parent category (level 2), 2022 to 2025. For each year, the category's orders in a month are divided by that year's total and multiplied by 12, so 1.0 is an average month; the years are then averaged with equal weight. The peak is the highest month.index(month) = mean over years of (month orders ÷ year orders × 12) peak = the month with the highest indexSince 8 Sep the index is only ever a shield or a push: it holds a cut back (peak guard) or flags a category to lean into (ramp). It is never a reason to reduce anything. | Also on |
|---|---|---|---|---|
| toothbrushes and toothpaste · personal care | $1,319 | 11% | Oct · next month | left alone |
| medical supply · personal care | $1,312 | 32% | Oct · next month | left alone |
| soap & body wash · personal care | $1,131 | 21% | Oct · next month | left alone |
| skin care · personal care | $1,115 | 18% | Oct · next month | left alone |
| personal care items · personal care | $1,097 | 14% | Oct · next month | left alone |
| bath & body · personal care | $902 | 14% | Oct · next month | left alone |
| cosmetics · personal care | $861 | 24% | Oct · next month | left alone |
| deodorant · personal care | $650 | 52% | Oct · next month | watch |
| hair products · personal care | $453 | 99% | Oct · next month | watch |
| shampoo & conditioner · personal care | $396 | 26% | Oct · next month | left alone |
| hand sanitizer · personal care | $275 | 41% | Oct · next month | watch |
| shaving razors · personal care | $211 | 23% | Oct · next month | left alone |
| hygiene kits · personal care | $142 | no orders | Oct · next month | below floor |
| feminine care · personal care | $120 | 18% | Oct · next month | below floor |
| manicure and pedicure items · personal care | $105 | 18677% | Oct · next month | below floor |
| cotton balls & swabs · personal care | $70 | 20% | Oct · next month | below floor |
Window spend $19,166 at 25.0% ACoS, against $33,742 at 37.7% in the same window last year.
Lost money in this same window in consecutive years, and its category is not about to peak. Switched off for 90 days, minus individually profitable products (carved out, they keep serving), then re tested.Chronic loser. Lost money in this same window two years running. Last year must have real spend (at least $50) at ACoS over 50%; the year before must be over 50% too, or have no judgeable spend. Zero orders on $50+ counts as over 50%.LY spend ≥ $50 and LY ACoS > 50%
and (LY2 ACoS > 50% or LY2 spend < $50)This is the only route to a switch off, and it is the same test in every month of the year. Then the peak guard is checked before anything is applied.
| Product type | 90d spend90d spend is the type's ad cost across the window, Shopping and Performance Max together.sum of cost for every product in the type, Jun 3 to Aug 31PMax reports product level cost as a floor, not a ceiling, so the true figure can be higher. | ACoSACoS is ad spend divided by the conversion value those ads produced, both from Google Ads shopping performance for the product type, click date attribution.ACoS = cost ÷ conversion value toy sets: $278 ÷ $760 = 36.6%At or under 35% a type is left alone and never enters the lists. Inside Performance Max the product level cost Google reports is a floor, so a type's true ACoS can only be higher than shown. | Last yearLast year / year before is the type's ACoS over the same calendar dates in 2025 and 2024, so a type is always compared to itself in season, never to the account average.LY ACoS = LY cost ÷ LY conversion value, same datesShown only where that year had at least $50 of spend; a middot means less. "no orders" means $50+ was spent with no conversion value, which counts as over any bar. | Year before | PeakPeak is the strongest month of the parent category's order index, 2022 to 2025 (the seasonality index chip under the tree shows the formula). It is listed for every type so the calendar is visible, but it only ever does two things: hold a switch off back (peak guard, any amplitude) or put the type on the ramp list. It is never a reason to reduce. |
|---|---|---|---|---|---|
| mens apparel · apparel | $5,867 | 40% | 52% | 52% | Jun |
| animal toys · toys & games | $673 | 38% | no orders | · | Jun |
| outdoor toys · toys & games | $474 | 79% | 113% | · | Jun |
| linens · home & kitchen | $440 | 64% | 116% | no orders | Apr |
| boys apparel · apparel | $263 | 53% | 99% | 1635% | Jun |
| kitchen & dining · home & kitchen | $170 | 67% | 103% | no orders | Apr |
| winter gear · non-profit / charity | $106 | no orders | no orders | · | Jan |
Carved out, keeps serving ($2,051): individually profitable products inside these types are exempt from the action.Carve out. Before an Exclude is applied, every product inside the type is checked, with variants pooled into one parent (by item group id where the feed has one, otherwise by title with the size and color words stripped). A parent with at least $50 of window spend at 35% ACoS or under is exempted and keeps serving.parent spend ≥ $50 and parent cost ÷ parent value ≤ 0.35 → keeps servingProduct data only ever keeps things on; it never adds a pause.
wholesale bulk men's cotton t-shirts - mixed colors size lar · in mens apparel · $1,202 at 34% ACoS
12 pieces men's cotton short sleeve t-shirt size large, asso · in mens apparel · $197 at 30% ACoS
wholesale bulk men's t-shirts - fruit of the loom assorted s · in mens apparel · $169 at 30% ACoS
72 pieces fruit of the loom mens assorted t shirts, assorted · in mens apparel · $120 at 33% ACoS
1200 pieces mens underwear pallet deal - mens underwear · in mens apparel · $99 at 18% ACoS
48 pieces yacht smith mens cotton short sleeve t shirts mix · in mens apparel · $72 at 6% ACoS
12 pieces men's cotton short sleeve t-shirt size small, asso · in mens apparel · $67 at 18% ACoS
72 pieces fruit of the loom mens assorted t shirts, assorted · in mens apparel · $65 at 20% ACoS
72 pieces mens boxer briefs, assorted sizes and colors - men · in mens apparel · $61 at 22% ACoS
Worked in this window last year, does not now. Gets a cause before anything is switched off.Fine last year. Same dates in 2025, at least $50 of spend, ACoS at or under 35%. If it cleared the bar a year ago, something changed, and pausing would hide the symptom.LY spend ≥ $50 and LY ACoS ≤ 35% → Investigate
| Product type | 90d spend90d spend is the type's ad cost across the window, Shopping and Performance Max together.sum of cost for every product in the type, Jun 3 to Aug 31PMax reports product level cost as a floor, not a ceiling, so the true figure can be higher. | ACoSACoS is ad spend divided by the conversion value those ads produced, both from Google Ads shopping performance for the product type, click date attribution.ACoS = cost ÷ conversion value toy sets: $278 ÷ $760 = 36.6%At or under 35% a type is left alone and never enters the lists. Inside Performance Max the product level cost Google reports is a floor, so a type's true ACoS can only be higher than shown. | Last yearLast year / year before is the type's ACoS over the same calendar dates in 2025 and 2024, so a type is always compared to itself in season, never to the account average.LY ACoS = LY cost ÷ LY conversion value, same datesShown only where that year had at least $50 of spend; a middot means less. "no orders" means $50+ was spent with no conversion value, which counts as over any bar. | Year before | PeakPeak is the strongest month of the parent category's order index, 2022 to 2025 (the seasonality index chip under the tree shows the formula). It is listed for every type so the calendar is visible, but it only ever does two things: hold a switch off back (peak guard, any amplitude) or put the type on the ramp list. It is never a reason to reduce. |
|---|---|---|---|---|---|
| school supply kits · school supplies | $600 | 116% | 21% | 29% | Aug |
| soap & body wash · personal care | $224 | 133% | 22% | 60% | Aug |
| cleaning supplies · home & kitchen | $212 | 63% | 18% | · | Apr |
| books · school supplies | $182 | 75% | 28% | · | Aug |
| personal care items · personal care | $126 | 78% | 9% | · | Aug |
| medical supply · personal care | $107 | 37% | 4% | 18% | Aug |
Losing money now but not two years running, not enough history yet, or chronic but held by the peak guard. Keeps serving, checked again next month.Watch holds three kinds of type: no judgeable history in either prior year (under $50 both years); bad now but not chronic, which since 8 Sep includes every off season type that used to be reduced on the old Schedule list; and chronic but held by the peak guard. All three keep serving. It costs little to wait a month for a clean reading, and it costs real sales to cut demand on weak evidence.
| Product type | 90d spend90d spend is the type's ad cost across the window, Shopping and Performance Max together.sum of cost for every product in the type, Jun 3 to Aug 31PMax reports product level cost as a floor, not a ceiling, so the true figure can be higher. | ACoSACoS is ad spend divided by the conversion value those ads produced, both from Google Ads shopping performance for the product type, click date attribution.ACoS = cost ÷ conversion value toy sets: $278 ÷ $760 = 36.6%At or under 35% a type is left alone and never enters the lists. Inside Performance Max the product level cost Google reports is a floor, so a type's true ACoS can only be higher than shown. | Last yearLast year / year before is the type's ACoS over the same calendar dates in 2025 and 2024, so a type is always compared to itself in season, never to the account average.LY ACoS = LY cost ÷ LY conversion value, same datesShown only where that year had at least $50 of spend; a middot means less. "no orders" means $50+ was spent with no conversion value, which counts as over any bar. | Year before | PeakPeak is the strongest month of the parent category's order index, 2022 to 2025 (the seasonality index chip under the tree shows the formula). It is listed for every type so the calendar is visible, but it only ever does two things: hold a switch off back (peak guard, any amplitude) or put the type on the ramp list. It is never a reason to reduce. |
|---|---|---|---|---|---|
| balls · toys & games | $330 | 39% | · | · | Jun |
| womens apparel · apparel | $160 | no orders | no orders | 8% | Jun |
| educational toys · toys & games | $158 | 162% | · | · | Jun |
| toy sets · toys & games | $143 | 44% | · | · | Jun |
| party center pieces · party supplies peak guard | $142 | 40% | 59% | · | Sep |
| umbrellas & rain gear · home & kitchen | $113 | 35% | · | no orders | Apr |
Category peaks next month, or is in its peak now. Push, not pause: lift bids on the profitable types, make sure the rest are serving, so the category meets its demand with recent data. Runs alongside the verdicts, so a Watch item can also be here.Ramp list. Every type whose category peaks next month, or is in its peak month now, with at least $50 of window spend and a peak that repeats: the category was above its average in that month in all, or all but one, of the index years, so a single year's fluke does not get budget thrown at it.months until peak = (peak month − run month) mod 12
on the list if ≤ 1, and above average at the peak in ≥ (years − 1) of the index yearsThe action is a push, not a return to neutral: lift bids and check budget share on the profitable ones, make sure the losing ones are at least serving, so the category enters its peak with recent data. Types on this list are re read after the peak, never cut into it.
| Product type | 90d spend90d spend is the type's ad cost across the window, Shopping and Performance Max together.sum of cost for every product in the type, Jun 3 to Aug 31PMax reports product level cost as a floor, not a ceiling, so the true figure can be higher. | ACoSACoS is ad spend divided by the conversion value those ads produced, both from Google Ads shopping performance for the product type, click date attribution.ACoS = cost ÷ conversion value toy sets: $278 ÷ $760 = 36.6%At or under 35% a type is left alone and never enters the lists. Inside Performance Max the product level cost Google reports is a floor, so a type's true ACoS can only be higher than shown. | PeakSeasonality index. Built on the parent category (level 2), 2022 to 2025. For each year, the category's orders in a month are divided by that year's total and multiplied by 12, so 1.0 is an average month; the years are then averaged with equal weight. The peak is the highest month.index(month) = mean over years of (month orders ÷ year orders × 12) peak = the month with the highest indexSince 8 Sep the index is only ever a shield or a push: it holds a cut back (peak guard) or flags a category to lean into (ramp). It is never a reason to reduce anything. | Also on |
|---|---|---|---|---|
| party center pieces · party supplies | $142 | 40% | Sep · peak month | watch |
| watches · jewelry & accessories | $102 | 4% | Oct · next month | left alone |
| costumes & accessories · party supplies | $72 | 20% | Sep · peak month | below floor |
| party favors · party supplies | $62 | no orders | Sep · peak month | below floor |
The investigate list is still the story, and it is still one story rather than 30. Wholesale Sock Deals bought 75% more clicks than the same window last year and got 13% more orders, with ACoS moving from 18.3% to 24.3%. Most of the types on the investigate list did not independently break. The account is buying more traffic that converts less, which is the bid strategy and traffic quality problem identified in the rulebook. Fixing that is worth more than every pause on this page combined, and it is where the investigate work goes next.
The eight exclusions are applied as feed label exclusions with their carve outs, the personal care push goes in as bid and budget changes, and the mens apparel decision is yours. Nothing moves before then.
The script runs again. Every exclusion is re confirmed or reverted, watch items get their next reading, and the ramp list turns to the November categories: socks, laundry, party supplies and winter wear.