House of Spoils · Internal · Not for external sharing

Traffic Strategy Review
August 2026

What we changed in June, what it earned, and what to fix next.
All numbers: Shopify net revenue (ex-tax, paid orders) + GA4 traffic + Meta spend.

The problem — June 2026

Clickbait pages ate our traffic.

9 "clickbait" product pages pulled 10,820 pageviews a day — and sold almost nothing.

Ads sent buyers to eye-catching pages, not to prints people buy.
Best sellers got only 6.8% of product-page traffic.
Result: lots of visits, weak revenue per visit — $1.03 per session. "$ per session" = net revenue ÷ site visits. Our core quality measure.

The fix — June 23–24

All roads now lead to the 48 best sellers.

Catalog ads (Meta + Google) re-pointed at the top-48 best-seller collection.
Site traffic rerouted the same way.
One collection = one source of truth: Shopify collection of 48 prints, mirrored to ad product sets.

We did not cut spend. We redirected where the spend lands.

Result — 61 days after the change

Yes. Revenue per visit doubled.

$ per session

$1.03 → $2.09

+103% — a doubling, and still rising

Revenue per day

+32%

$17,595 → $23,154

Sessions per day

−35%

Fewer visits, far better visits

Conversion rate

+92%

Orders ÷ sessions nearly doubled

Orders per day

+24%

18.4 → 22.8

Trajectory

$1.94 → $2.24

First 30 days vs next 31 — improving, not decaying

Mechanism

We redirected traffic. We did not cut it.

Clickbait pages

−95%

Daily pageviews: 10,820 → 516. The 9 named pages lost 91–99% of their traffic.

Best-seller pages

+331%

Daily pageviews: 1,323 → 5,701. Best sellers now get 69.9% of product-page traffic (was 6.8%).

Meta spend per day actually rose +25%. Session quality did the work — not budget cuts.

Case study — Horsepower (the print)

Dedicated campaigns don't beat the tide.

We ran a dedicated UGC ad campaign for one print for 4 weeks. It returned 0.68× — breakeven is 1.76×.

Print	Dedicated campaign?	Revenue lift (4 wk)
Horsepower	Yes — $17,455 spend	+72.1%
Porsche Surf	No	+79.1%
Sitewide baseline	—	+14.7%

The comparator with no campaign grew faster. A seasonal tide lifted all boats — the campaign added roughly nothing. Meta itself attributed only 3 purchases to it. Verdict: pause it; no new dedicated print campaigns.

Watch-out

Junk carousel traffic distorts our numbers.

Aug 16–17: the catalog carousel flooded the site with 43,163 near-zero-value sessions in 2 days.
For Horsepower alone, that flood = 29.9% of its entire 60-day pageview count.
Everyday carousel traffic bounces at 78%; every other source bounces at 13–28%. "Bounce" = visitor leaves without doing anything.
Root cause: the carousel's hero card is a favorite-variant issue on Horsepower. Fix is manual, in Meta Commerce Manager (Clay/Kylie) — still outstanding.

The roster — all 48 best-seller slots, last 60 days

$652.6k earned on $192.1k Meta spend.

Keep — clearly earning the slot

7

$130,892 · 20.1% of revenue

Watch — holding, something off

31

$420,043 · 64.4% of revenue

Weak — swap-out candidates

10

$101,673 · 15.6% of revenue

Top earners: Porsche Surf $62.6k · Horsepower $51.4k · Risky Business $32.5k.

Key finding

The problem is not weak prints.
It is misallocated ad spend.

The weak 10 earn

15.6%

of roster revenue ($101,673)

…but consume

36.5%

of tracked ad spend ($70,208)

Print	L60 ad spend	Gross ROAS
Birdies — worst slot on the roster	$23,753	0.68×
Letting Go	$15,108	1.56×
Escape Riva	$11,220	1.62×

All three sit below the 1.76× breakeven. That is $50,081 of spend returning $57,978 — blended 1.16×. ROAS = revenue ÷ ad spend. Below 1.76× we lose money on the sale.

Proposed actions

Pending approval — nothing changed yet

Swap 3. Cut spend on 2. Protect 1.

Swap out → in: Birdies → High Class · Sunset Dip → Get In · French Riviera → Land Of The Free. Replacements all earn $11–12k L60 at 4–6× the roster's median $ per pageview.
Cut spend, keep the prints: Letting Go + Escape Riva — healthy prints being over-fed ($26,328 at ~1.6×). Clay conversation, not a collection change.
Do NOT remove Horsepower: the dashboard flags it, but its metric is polluted by junk carousel traffic — on real Shopify revenue it is the #2 print (4.29× ROAS).
No new dedicated print campaigns — the Horsepower test failed (0.68× vs 1.76× breakeven).

How to judge campaigns · Fairing post-purchase survey

How we should measure success.

Fairing survey, all-time (May 2024–Aug 2026) · 2,128 responses · 56% response rate
Q: "When was the first time you heard about House of Spoils?"

Heard of us >1 year before buying

44%

Nearly half of buyers knew us over a year before they purchased

Knew us >2 weeks before buying

~62%

Most buyers had a long consideration window before converting

Bought the week they discovered us

~17%

Only 1 in 6 buyers converts in the platform's 7-day attribution window

AOV — long-consideration buyers

$1,042–$1,089

Buyers who waited months or years spend more per order

AOV — same-week buyers

$921

Impulse buyers spend less

Takeaway: 7-day platform ROAS can't see most of our buyers. Judge campaigns on months, not weeks — and re-measure Horsepower in October before killing anything.

Caveats — read before deciding

Know the limits of these numbers.

Meta-only spend. Google Ads spend is not counted (no API access). All ROAS figures overstate true efficiency.
4-week window on the Horsepower campaign test. Directionally strong, but short for a high-price, slow-decision product.
Attribution limits. Meta's dedicated adset is invisible in GA4; incremental figures credit the campaign generously — reality is likely worse, not better.
Unequal windows. Pre = 34 days, post = 61 days; all comparisons are per-day. Summer seasonality is not controlled for.

Next steps

Four decisions on the table.

1. Max approves / rejects the 3-print swap.
2. Clay: reduce spend on Letting Go + Escape Riva; pause the Horsepower UGC adset.
3. Clay/Kylie: fix the Horsepower carousel hero variant in Commerce Manager (outstanding since Aug 18).
4. Review again at the next swap window — cadence stays 1–2× per month.

Sources: bestsellers-deepdive-2026-08-27.md · roster-review-2026-08-27.md — full evidence trails inside.