- Blended (why the totals moved):
The headline ROAS pop (1.36x → 2.75x) is a spend-mix artefact, not a performance win: total spend collapsed 60.5% ($2,410.40 → $952.05) almost entirely because Demand Gen was cut by $1,288.87 (from $1,358.18 to $69.31) and PMax was zeroed out (from $159.42 to $0) — together that's $1,448 of the $1,458 total spend decline. Orders fell 52 → 40 (-23%) and revenue fell $3,278 → $2,619 (-20%), so the account got smaller and less productive even as the ratio improved. Brand Search IS rose to 91.9% (from 90.3%) and Reviews IS to 76.5% (from 74.3%), so the shrinkage wasn't a coverage problem — it's a deliberate/accidental budget reallocation away from upper-funnel toward brand, and since 1dc undercounts PMax/Demand Gen, the real damage to top-of-funnel assisted revenue is likely invisible here.
- Search Brand:
Orders fell 36 → 30 (-17%) and revenue $2,467 → $1,895 (-23%) despite spend rising 16.5% ($377.05 → $439.45), pushing blended ROAS down 6.54x → 4.31x. CPC_Core, the workhorse, dropped 31 → 25 orders on slightly higher spend ($304 → $324), cutting ROAS 7.12x → 5.18x (-27%); Products was worse, spend up 61% ($67 → $108) while orders slipped 5 → 4 and ROAS cratered 4.52x → 1.58x (-65%). Modifiers is a rounding error (1 order/$7, 6.51x) and Core remains dormant at $0 both weeks. IS actually improved to 91.9%, so this is a CPC/conversion-efficiency problem, not an impression-share problem — likely rising brand CPCs or a softer landing-page/offer week.
- Search Non Brand (Reviews):
Small but genuinely better: orders 3 → 5 (+67%), spend down 15% ($217.68 → $185.34), and ROAS more than doubled 0.55x → 1.41x, all on the single SR_US_NB_SpaDr_Reviews campaign. IS climbed to 76.5% (from 74.3%), so the gain is coming from cheaper, more efficient auctions rather than a spend increase. Still, at 1.41x ROAS and $37.07 CPA against a $52.18 AOV this segment remains marginal — improvement, but not yet a scale candidate.
- Shopping Brand:
This is a real deterioration, not noise: the sole campaign (SHP_TSD_US_PPT_ACQ_VOLUME_CBO_WK45) lost more than half its orders (12 → 5, -58%) while spend actually rose 8% ($238 → $258), driving ROAS down 2.73x → 1.79x (-34%) and revenue down 33% ($685.87 → $462.78). With CPA jumping to $51.59 against AOV $92.56, the campaign is spending more to convert far fewer buyers — check for feed/price/stock issues or rising Shopping CPCs before adding budget back.
- Pmax Non Brand (Serum):
PMax went from $159.42 spend / 0 tracked orders to fully paused at $0 this week — a 100% spend cut on a campaign that was already showing zero last-click conversions. Because 1dc structurally undercounts PMax, this campaign should not be judged as "dead weight" purely on this feed; the zero-order readout could be masking real assisted revenue that's now been lost entirely by turning spend off. Pull GA4 assisted conversions/revenue for this campaign before deciding whether the pause was the right call.
- Demand Gen (Discovery):
Spend was slashed 95% ($1,358.18 → $69.31) on DGEN_US_NB_TSD_HairSerum_Discovery, with 0 tracked orders in both periods — this single cut accounts for the majority of the blended spend collapse and thus most of the ROAS "improvement." Zero direct orders on $1,358 of prior spend is a legitimate red flag on efficiency, but Demand Gen is upper-funnel and 1dc won't capture view-through/assisted impact, so the true cost of this cut can only be assessed via GA4 assisted conversions, not this report alone.