NANOBAG

Are more people buying with the new bag picker?TEST RUNNING — DAY 2 OF 7

Half our visitors currently see a new card-style picker on the product page (big buttons naming each bag and what it's for). The other half see the current design. Everything else is identical. Updated 12 Aug 2026.

The answer so far

Just as many people are buying. Both versions have produced exactly 112 orders each.
But they're buying different bags. With the new picker, shoppers pick a third more Daypacks — our most premium bag — and far fewer Slings. They also put slightly fewer bags in each order, so the average order is worth about 5% less right now. More of them go on to accept the post-purchase offer, which wins some of that back.
No decision yet — and no cause for concern. Two days of data is too little to call a winner. The test is healthy, the split is fair, and the decision date is 17 August.

People buying

Even

112 orders each side

Value per order

−5%

$64.58 vs $61.08

Daypacks sold

+32%

share of bags: 22% → 32%

Test health

Good

fair 50/50 · all checks passed

What people buy, old picker vs new

Share of bags sold on each side. The new picker names every style and what it's for — and that visibly moves choices toward backpacks and away from the cheaper cross-body Sling.

Current pickerNew picker

Standard tote · $16–19

29%
28%

Daypack zip backpack · $26–30

22%
32%

XL large tote · $18–21

21%
18%

Sling cross-body · $16–19

19%
12%

Micro small tote · $14–19

7%
5%

Pack drawstring · $16–19

3%
5%

644 bags across 224 orders, free gifts excluded. Two days of data — treat the pattern as an early signal, not a conclusion.

Why this matters: if the pattern holds, the new picker trades basket size for bag quality — customers buy our best bag more often, but fewer bags per order. Whether that's a win depends on margins per style, which is exactly what the final report will weigh.

What happens next

13 AugSafety checkpoint — stop early only if the new picker is clearly losing sales. On current numbers it isn't; the test continues.
17 AugScheduled end. If the result is clear, we decide; if it's still close, we extend a few days rather than guess.
AfterFinal report with the recommendation: roll out, roll back, or iterate on the picker's design.

Why you can trust these numbers

Visitors are split 50/50 automatically by Shopify. Every order is invisibly labelled with which picker the buyer saw, so the sales figures above come from real orders, not estimates — and we cross-check three independent systems (Shopify's test dashboard, the order records, and site analytics) daily. All fidelity checks passed: each market shows the right version, and the label on every one of the 308 orders checked matched.

Appendix for the analytics team — full tables ▾

Shopify Rollouts (verdict layer), read 12 Aug

Metric	Control	Treatment	Δ
Sessions	3,014	2,945	—
Conversion rate	4.68%	4.14%	−0.54pp (CIs overlap)
Reached checkout	8.76%	7.20%	−1.56pp
Added to cart	16.09%	14.13%	−1.97pp
Bounce	29.00%	29.68%	≈ even

Audience includes EU where the treatment is off (English-only labels) — both EU arms see identical pages, so dashboard rates are diluted toward zero; the true test-market effect is larger in either direction.

Order tags (money layer), 11 Aug 00:00 → 12 Aug ≈15:00 HKT

	Control	Treatment
Orders (day 1 / day 2)	70 / 42	71 / 41
Revenue	$7,233	$6,841
AOV / median	$64.58 / $58.85	$61.08 / $50.21
Paid bags per order	2.99	2.76
Post-purchase upsell taken	16.1%	19.6%

GA4 (behavior layer), six test markets

	Control	Treatment
Users engaging the picker	1,609	1,522
Style switches per engaged user	6.4	6.2
ATC-click users	487	468
ATC users ÷ picker users	30.3%	30.7%

Hygiene: set GA4 exploration range to 11–17 Aug (current exports include 10 Aug, the previous test's last day); any Events-report view must carry the six-market country filter — EU always stamps control by design. Attribution audit: 308 orders, 0 tag/stamp conflicts, 1 mixed-arm cart (0.3%). New-vs-returning split unavailable (API scope). Rollouts has no API — dashboard figures transcribed manually.

NB05 "Only Switcher" · nanobag.com · 50/50 split, 18 markets (EU sees the current picker on both sides by design) · Test window 11–17 Aug 2026 · Interim report, day 2 of 7 — final verdict after 17 Aug.