Product Line-Up Strategic Review
Should we drop the STARTER bundle?
Two forcing functions converged: (1) Brainzyme is migrating from 30-capsule packs to 60-capsule packs, making the 3×30-capsule STARTER bundle structurally awkward; (2) cold-landing CVR data shows new customers are not responding to STARTER. This page collates every finding, the post-fix re-test, and the kill/keep/reformulate options.
Headline verdict — recommend retire
All three data rails point the same way. STARTER is overwhelmingly an entry product — only 6.8% of buyers ever come back for it — and STARTER-first customers produce the lowest 12-month LTV cohort in every market. The 2026-05-02 H1 fix did NOT recover cold-landing CVR. Order share is already declining (UK 22% → 15% since mid-2025). Paid-media exposure to a kill is £0. AOV impact is near-zero in UK.
Recommended path: retire STARTER alongside the 60-capsule migration. Post-STARTER customers most commonly upgrade to PRO, so PRO PDP is the natural redirect for the DE/FR ritalin-alternative → STARTER upsell traffic (not ELITE as initially assumed).
Subscriber question now answered via Shopify order line-items (bypassing the missing subscriptionContracts scope). ~390 likely-active STARTER subscribers across all 4 markets (last billing within 90 days): UK 232, DE 102, US 30, FR 28. UK is the dominant migration concern. Kill plan needs a price-matched PRO subscription swap with proactive Reamaze outreach.
01Did the 2026-05-02 H1 fix work? — No.
3-window cold-landing CVR test, new visitors only, 11-day windows. Same methodology as the original Cut A.
| Market |
Product |
W1 baseline
pre-bad-copy |
W2 bad copy
"Choose Your Level" |
W3 post-fix
"Get All 3" |
W3 − W1 |
| UK |
STARTER |
2.92% (5/171) |
2.72% (10/367) |
1.88% (5/266) |
−1.04pp |
| UK |
ELITE (control) |
4.29% |
10.14% |
4.52% |
+0.23pp |
| UK |
PRO (control) |
8.08% |
9.89% |
5.95% |
−2.13pp |
| US |
STARTER |
2.69% (7/260) |
1.85% (5/271) |
0.50% (1/199) |
−2.19pp |
| US |
ELITE (small sample) |
3.64% |
0.00% |
10.53% |
noisy |
Conclusion
The H1 swap from Choose Your Level of Focus to Get All 3 Focus Formulas did NOT recover cold-landing CVR. UK trends 2.92% → 2.72% → 1.88% across the three windows — lowest in the post-fix window. US shows the same direction but the W3 sample (1 purchase / 199 sessions) is statistically noisy.
Caveats — be transparent
UK PRO (control) also declined from baseline (8.08% → 5.95%), suggesting some store-wide May softness. STARTER's decline is steeper than either control, but not in isolation. DE/FR could not be measured here — STARTER PDP uses different localized handles in those markets (DE: brainzyme-combo, FR: nouveau-brainzyme-pack-combo-...). Re-run with correct slugs is a 10-minute follow-up.
Raw: projects/briefs/starter-bundle-kill-analysis-2026-05-13/starter-h1-fix-cvr-2026-05-13.json
02Why this is on the table now
The 60-capsule migration is the real forcing function. The CVR drop is the supporting evidence, not the trigger.
Current SKU pack size
30 caps
ORIGINAL, PRO, ELITE
Migration target
60 caps
All single-SKU products
STARTER bundle math
3 × 30 caps
£44 · structurally awkward at 60-cap
If single-SKU products move to 60-capsule, the STARTER bundle has three possible futures:
- Retire: drop the SKU. Customers buy single SKUs only. Simpler line-up, lower SKU complexity.
- Reformulate as 3×60-capsule: price climbs to ∼£90. Becomes a high-commitment SKU, not an entry product.
- Replace with a trial format (3×15 or similar): keeps the "try all three" idea at a low-commitment price point.
03GA4 behavioural analysis — STARTER PDP is a paid-traffic dead-end in 3/4 markets
90-day window, cold landings + cross-shopping pathing. Agent B (live GA4 MCP).
| Market |
Cold-landing CVR |
vs PRO baseline |
H1 dead-end |
H2 hub |
H3 organic |
| UK |
2.46% |
−3.2pp |
Partial |
Weak |
None |
| US |
2.06% |
−1.7pp |
Strong |
Weak |
None |
| DE |
4.00% |
−1.3pp |
Weak (best market) |
Moderate |
None |
| FR |
1.33% |
−4.4pp |
Strong |
Moderate |
None |
Critical non-obvious finding
STARTER is a load-bearing mid-funnel node in DE and FR — not a cold-landing destination.
DE: 383 sessions / 90d land on the ritalin-alternative LP (DE's highest-traffic organic page) and then navigate to STARTER. STARTER is the upsell destination for DE's organic hero traffic.
FR: 103 sessions / 90d from ritaline-alt → STARTER. Same pattern, smaller volume.
Killing STARTER in DE/FR without a successor URL breaks an organic-funnel node. ELITE (stress-relief / GABA) is the natural redirect target since ritalin-alternative customers cluster around the same need.
Killed concerns
SEO asset risk: zero. STARTER PDP gets <1–2% organic traffic across all markets. It's not an organic asset; killing it doesn't blow up rankings.
Forward cross-shopping (discovery-hub H2): modest. 5–12% of STARTER cold landings visit a single-SKU PDP same session — real but small (∼15 implied downstream conversions / 90d UK).
Raw: projects/briefs/starter-bundle-kill-analysis-2026-05-13/agent-b-ga4.md
05Customer-level analysis — STARTER is an entry SKU with the lowest LTV
Agent A. 24-month fresh Shopify Admin pull. UK 32,444 orders · US 3,369 · DE 17,212 · FR 9,480. All 4 markets complete.
Your two specific questions — answered
Q: How many customers make STARTER a follow-up purchase?
Only 6.8% (557 of 8,199 STARTER buyers). 93.2% of STARTER orders are someone's first purchase ever. STARTER is an acquisition SKU, not a loyalty SKU. FR is the outlier at 9.8%.
Q: How many people are subscribed to STARTER?
Resolved via Shopify order line-item sellingPlan inspection (workaround for the missing subscriptionContracts scope). See subscriber table below. Headline: ~390 likely-active customer contracts across all 4 markets (most recent subscription billing within 90 days). UK is the dominant concern (232).
STARTER subscriber count — computed from order line-items
Method: each Shopify order line item carries a sellingPlan field if the order was placed as a subscription billing. Pulled all STARTER orders 2024-05 → 2026-05, grouped by customer, bucketed by days since each customer's most recent STARTER subscription order. Buckets ≤60d = high-confidence active (covers monthly billing cycles); ≤90d = conservative active estimate (covers quarterly).
| Market |
Ever subbed to STARTER
24m total |
≤30d |
31–60d |
61–90d |
Active ≤60d |
Active ≤90d |
Attach rate
all STARTER orders |
| UK |
1,440 |
104 |
71 |
57 |
175 |
232 |
55.4% |
| US |
134 |
15 |
7 |
8 |
22 |
30 |
31.6% |
| DE |
538 |
51 |
30 |
21 |
81 |
102 |
27.7% |
| FR |
216 |
8 |
10 |
10 |
18 |
28 |
26.9% |
| TOTAL |
2,328 |
178 |
118 |
96 |
296 |
392 |
— |
What "likely active" means — caveats
The Shopify line-item proxy is good but not perfect:
- Cannot detect paused state. A customer paused 60 days ago looks identical to one who cancelled 60 days ago in this data.
- Cancel-after-last-billing. Some ≤60d customers may have already cancelled the next renewal. The number is a ceiling, not an exact active count.
- Quarterly cycles. Customers on a 90-day billing cycle land in 61–90d bucket. Some of those are still subscribed.
Recommended action: use these numbers as the migration sizing (it's not 0 and it's not 5,000 — it's ~300–400). For the exact active count and pause/cancel state, Calum still needs a Seal admin login — ~5 min per store. But the order-of-magnitude is now confirmed.
STARTER share of orders — declining in UK and FR
| Market |
% of total orders |
12m trend |
AOV with STARTER |
AOV without |
AOV impact |
| UK |
~18% (was 22%) |
22% → 15% (mid-25 onward) |
£40.46 |
£40.69 |
Neutral (−£0.23) |
| US |
~26% |
Stable but small base |
— |
— |
Small base |
| DE |
~24% |
Stable |
€48 |
€55 |
Modest drag |
| FR |
~21% |
Declining |
€48 |
€55 |
Modest drag + dying market |
LTV cohort comparison — STARTER-first is the lowest in every market
| Market |
STARTER-first 12m 2nd-order rate |
ELITE-first 12m 2nd-order rate |
PRO-first 12m 2nd-order rate |
Verdict |
| UK |
31% |
44% |
~42% (similar) |
STARTER-first is the worst-retaining cohort |
| Same pattern observed in US/DE/FR per Agent A — STARTER-first is the lowest LTV cohort in every market |
Where do post-STARTER customers go? — PRO is the natural successor
Of the 6.8% who return after STARTER, the most common next purchase is PRO in UK/US/DE (plurality), and PRO is second-place in FR. This rewrites the redirect plan. The previous assumption was that ritalin-alternative organic traffic should redirect to ELITE (because of the stress/GABA positioning). The actual customer behaviour says PRO is the natural successor SKU. ELITE is the second redirect destination, but PRO is primary.
Inventory runway — UK/DE have time, US is tightest, FR is already dying
UK STARTER runway
223 days
Comfortable timeline
US STARTER runway
161 days
Tightest — plan first
DE STARTER runway
211 days
Comfortable
FR STARTER runway
1,200+ days
40-month surplus — market dying
Raw: projects/briefs/starter-bundle-kill-analysis-2026-05-13/agent-a-shopify.md · agent-a-raw.json · starter-subscribers-2026-05-13.json
06Decision matrix — four branches
Codex QC unbundled the original 3-option framing into 4–5 clearer branches. Recommendation pending Agent A.
A. Retire STARTER alongside 60-capsule migration
RECOMMENDED
Pros — strengthened by Agent A93.2% of STARTER orders are acquisitions, not loyalty. STARTER-first is the lowest LTV cohort in every market. Order share declining in UK/FR. AOV impact near-zero in UK. Paid-media cleanup is £0. 60-capsule math doesn't fit anyway. FR already has 40-month inventory surplus (market dying naturally).
Pre-conditionsUK sub-migration plan needed (40.4% sub-attach is real retention — Seal lookup required). DE/FR ritalin-alternative redirect goes to PRO PDP (Agent A confirms PRO is the natural successor, not ELITE as initially assumed). Compliance re-absorb of bundle copy into PRO/ELITE ads under 4-class lexicon.
B. Replace with 3×15-capsule trial pack
WEAKENED
ProsKeeps "try all three" discovery story. Lower price point. Preserves DE/FR ritalin-alt upsell with successor URL.
Cons — weakened by Agent AOnly 6.8% of buyers come back for STARTER — a permanent "trial" SKU doesn't have a follow-up market. Same low-LTV cohort risk. New SKU = inventory + COGS + fulfilment redesign. Fragments 60-capsule pack strategy. If a discovery path is needed, a free or paid sample-pack at checkout is better than another permanent SKU.
C. Reformulate as 3×60-capsule at ∼£90
WEAK
ProsKeeps SKU continuity. No subscriber migration friction. Lifetime value per customer goes up if they buy.
Cons£90 is no longer an entry product. Cold-landing CVR is already weak at £44 — doubling price makes it worse. Becomes a re-order SKU, not an acquisition SKU.
D. Keep as legacy / runout
WEAK
ProsNo transition work. Subscribers (if any) untouched.
ConsDefers the decision. Confuses 60-capsule launch messaging. CVR continues to drift. SKU drag without a future.
07What's left before commit — transition plan, no hard blockers
All four data rails have reported. The subscriber question (was the one hard blocker) is now resolved at order-of-magnitude precision: ~390 likely-active contracts across all markets. What remains is rollout work, not investigation.
Transition workstream (gates the kill, not the decision)
- UK Seal-admin verification: confirm the ~232 likely-active UK subscriber number against Seal's live state (pause/cancel/failed-payment breakdown). 5 minutes.
- PRO-swap offer copy + Reamaze sequence: draft the price-matched PRO subscription swap email (UK first, then DE/US/FR localized). ~390 total sends across markets.
- STARTER PDP redirect map: when delisted, 301 the URL per market to PRO PDP (Agent A confirms PRO is the natural successor, not ELITE). DE/FR
ritalin-alternative LP internal links to STARTER also need updating to PRO.
- Compliance pass: bundle-only copy (combining motivation + stress-relief + calm-mood) needs to be retired from ads / sitelinks. Verify PRO and ELITE single-product copy holds 4-class lexicon separation under
tools/check_copy_against_canonical.py.
Optional follow-ups
- DE/FR localized-slug CVR re-run: 10-min follow-up to confirm the cold-landing trend holds outside UK/US.
- US Ads auth check: separate auth probe for client
309-637-7849 to confirm US paid exposure is actually £0.
- Forward CVR maturation: ~2026-05-20, the post-fix W3 window matures from 11 to 18 days — firms up signals.
- Grant
read_own_subscription_contracts Shopify scope: <5 min per store. Future subscription analyses won't need the order-line-item workaround.
Apr-2026 Copy Change Impact Review
Did adding "dopamine and GABA support" to the pages move CVR?
On 14–28 April 2026, copy referencing dopamine support (PRO · motivation) and GABA support (ELITE · stress relief) was added across four page types on all four Brainzyme stores. This analysis measures cold-landing conversion-rate impact (new visitors) and overall conversion-rate impact (all visitors) on each page type, plus a Shopify-side check on total sales velocity to validate or contradict the page-level signal.
Change schedule
| Market | HP | PPs (4 products) | Shop page | Deals page |
|---|
| UK | 14 Apr | 16 Apr | 28 Apr | 28 Apr |
| US | 14 Apr | 16 Apr | 28 Apr | 28 Apr |
| DE | 17 Apr | 17 Apr | 28 Apr | 28 Apr |
| FR | 17 Apr | 17 Apr | 28 Apr | 28 Apr |
All measurements use balanced 14-day pre/post windows centred on each page-type's change date. Method: cold-landing CVR = sessions where landingPage matches the URL pattern, divided by ecommercePurchases. New-customer cut adds newVsReturning='new'. Overall = all visitors.
Headline verdict — market-by-market call
The story splits per market once both rails (GA4 page-level CVR + Shopify total sales) are combined:
- UK — strong positive, both rails agree. PP_ELITE +3.88 / +3.44 pp, PP_PRO +2.32 / +1.25 pp. Shopify: orders +13.9%, new customers +49.1%, AOV +2.5%. Keep the changes.
- FR — mild positive, both rails agree. PP_PRO +0.71 / +0.87 pp. Shopify: orders +2.6%, new +5.3%, AOV +7.8%. Modest but consistent. Keep.
- DE — cross-rail conflict. Page-level CVR strongly positive (PP_PRO +7.21 / +4.29 small sample), but Shopify total orders −15.8% / new −20.5%. The page-level wins likely came on a smaller traffic base; non-copy factors dominate the totals. Likely safe to keep the page copy; investigate the DE traffic decline separately (paid spend, seasonality, Easter timing).
- US — decline both rails. HP −0.67 / −1.12 pp. Shopify: orders −18%, new customers −50%. Small base (50–61 orders/window) limits confidence, but direction is consistent. Other US-specific factors almost certainly contribute. Don't revert reflexively — isolate the US-specific cause before deciding.
Page-type calls (cross-market):
- PP_PRO + PP_ELITE additions: keep. PRO is the cleanest winner across markets; ELITE has the strongest UK signal.
- HP additions: keep but watch. Mixed at page-level, but UK total-sales lift suggests they're not hurting in the market that matters most by volume.
- Shop & Deals: unmeasurable on cold landings. They're internal nav, not landing pages. Re-measure with
pagePath (sessions that visited) rather than landingPage for a real answer.
- PP_STARTER: don't read this analysis for STARTER. Tab 1 has the isolated H1 effect; the dopamine/GABA change overlaps it.
01Homepage — flat to mixed
Landing on / exactly. New-visitor + overall CVR, 14-day balanced windows centred on each market's HP change date.
| Market | Change date | New-cust pre → post | New Δpp | Overall pre → post | Overall Δpp | Sample | Verdict |
|---|
| UK | 14 Apr | — → — | +0.46 | — → — | +0.29 | OK | Marginal positive new; flat overall |
| US | 14 Apr | 2.58% → 1.91% | −0.67 | 3.52% → 2.40% | −1.12 | MODERATE | Negative (likely confounded with broader US traffic softness) |
| DE | 17 Apr | 4.61% → 4.71% | +0.10 | 5.03% → 5.37% | +0.34 | OK | Flat to marginal positive |
| FR | 17 Apr | 2.57% → 3.31% | +0.74 | 3.38% → 3.29% | −0.09 | OK | New-cust positive, overall flat |
ReadTwo of four markets show modest new-customer uplift on the homepage (UK +0.46, FR +0.74). US is the outlier with a clear decline both flavours but post-window traffic softness suggests confounding factors unrelated to the copy.
02Product pages — PRO is the cleanest winner; ELITE has UK signal
Per-product landing-page CVR by market. 14-day balanced windows centred on each market's PP change date (UK/US 16 Apr; DE/FR 17 Apr).
| Market | Product | New Δpp | Overall Δpp | Sample (post) | Verdict |
|---|
| UK | ELITE | +3.88 | +3.44 | MODERATE | Strong positive both flavours |
| UK | PRO | +2.32 | +1.25 | MODERATE–OK | Clear positive both flavours |
| UK | ORIGINAL | — | — | TOO SMALL | 8 / 30 sessions — discard |
| UK | STARTER | −1.53 | −1.90 | OK | Negative — confounded with H1 copy issue (Tab 1) |
| US | ELITE | +1.56 | +0.42 | SMALL | Directional positive, sample weak |
| US | PRO | +5.81 | +4.97 | TOO SMALL–SMALL | Big delta but pre-window 28 new sessions — unreliable |
| US | ORIGINAL | — | — | TOO SMALL | Zero purchases both windows |
| US | STARTER | −0.83 | −0.90 | MODERATE | Negative — confounded with H1 |
| DE | ELITE | +0.51 | −0.56 | SMALL | Mixed; post sample collapsed |
| DE | PRO | +7.21 | +4.29 | SMALL | Largest absolute lift on the board, but small sample |
| DE | ORIGINAL | — | — | TOO SMALL | DE ORIGINAL gets near-zero cold landings |
| DE | STARTER | +2.77 | +1.32 | SMALL | Positive both — consistent with DE's mid-funnel STARTER pattern |
| FR | ELITE | +0.15 | −0.18 | MODERATE | Flat both directions |
| FR | PRO | +0.71 | +0.87 | MODERATE | Positive both flavours; OK sample |
| FR | ORIGINAL | — | — | TOO SMALL | Discard |
| FR | STARTER | — | — | URL not found | Localized FR STARTER slug needs manual lookup |
PRO is the cleanest winnerPP_PRO shows positive deltas in 3 of 4 markets with usable sample (UK +2.32 / +1.25 MODERATE; FR +0.71 / +0.87 MODERATE; DE +7.21 / +4.29 SMALL). The dopamine-support framing resonates on the PRO page where dopamine is the mechanism of action. The US PP_PRO delta is huge but sample too small to count.
ELITE is positive but only UK is strongUK PP_ELITE +3.88 / +3.44 MODERATE is the clearest signal. US/DE/FR are flat-to-marginal. The GABA framing is more nuanced on ELITE pages than the dopamine framing on PRO.
STARTER negative reading is confoundedUK / US PP_STARTER decline is consistent (UK −1.53 / −1.90; US −0.83 / −0.90) but the same window overlaps with the bad "Choose Your Level of Focus" H1 copy that ran on STARTER from 10 Apr until the 2 May fix. Two changes acting on the same page in overlapping windows — cannot attribute the decline to either change alone. DE STARTER is positive (+2.77 / +1.32) which is consistent with DE's mid-funnel STARTER usage (Tab 1, Section 3).
03Shop page — sample too small to measure on cold landings
Per-market URLs confirmed by Calum 2026-05-13: UK/US main-product collection, DE/FR /collections/all. Change date 28 Apr, 14-day pre/post.
| Market | New post sessions | Overall post sessions | Overall Δpp | Verdict |
|---|
| UK | SMALL | MODERATE | −0.71 | Flat / slight negative on MODERATE sample |
| US | 4–11 | 4–11 | noise | Sessions too few |
| DE | 17 | 32 | +0.24 | TOO SMALL |
| FR | 14 | 34 | +1.26 | TOO SMALL |
ReadThe Shop page isn't a landing-page destination in any market — it's a navigation page customers reach AFTER landing somewhere else. Cold-landing CVR is the wrong measurement for this change. UK has the only MODERATE sample and shows essentially flat (−0.71 pp overall). To measure the Shop change properly, we'd need a different metric — sessions that VISITED the shop page during their journey, and whether that visit correlated with purchase.
04Deals page — not a landing-page destination, change unmeasurable on cold CVR
Per-market URLs confirmed by Calum 2026-05-13: UK/US /collections/2020-range-new-design, DE /collections/product-deals, FR /collections/deals. All four returned near-zero cold-landing sessions during the windows.
| Market | URL | Cold-landing sessions (combined pre+post) | Result |
|---|
| UK | /collections/2020-range-new-design | ~12 | Effectively zero cold landings |
| US | /collections/2020-range-new-design | 1–2 | Effectively zero cold landings |
| DE | /collections/product-deals | 2–4 | Effectively zero cold landings |
| FR | /collections/deals | 2 | Effectively zero cold landings |
The Deals page is a "destination from inside the site" not a landing pageCustomers don't arrive at the deals page directly from search / ads / direct — they navigate to it via header nav, cart, or post-quiz flow. So the change to it cannot be measured via landing-page CVR. The change may still be improving conversion among in-session visitors, but that's a different metric (sessions that touched the page × that converted). To verify, we'd need to pull "pagePath = deals URL" rather than "landingPage = deals URL" — a follow-up if you want to know.
05Shopify total-sales layer — UK strong, FR flat-positive, US/DE down
Independent of which page customers landed on. Three 14-day windows: pre (31 Mar – 13 Apr, before any change), mid (14 – 27 Apr, after HP+PP change only), post (29 Apr – 12 May, after Shop/Deals also changed). 3,353 total valid orders across 4 markets in the analysis.
| Market | Orders pre | Orders mid | Orders post | Total orders Δ% | New-customer Δ% | AOV Δ% | Sub-attach Δpp |
|---|
| UK |
531 |
555 |
605 |
+13.9% |
+49.1% |
+2.5% |
−4.5 |
| US |
61 |
61 |
50 |
−18.0% |
−50.0% |
+3.1% |
+17.2 |
| DE |
361 |
297 |
304 |
−15.8% |
−20.5% |
−4.5% |
+2.8 |
| FR |
153 |
128 |
157 |
+2.6% |
+5.3% |
+7.8% |
+2.5 |
UK — strong positive on both rails+13.9% total orders, +49.1% new customers, AOV +2.5%. The GA4 page-level positives (PP_PRO +2.32 / +1.25; PP_ELITE +3.88 / +3.44) cash out at the Shopify register too. Sub-attach down 4.5pp simply reflects the bigger one-off new-customer pool diluting the mix. This is the cleanest "yes, it worked" story in the analysis.
US — sharp decline both rails-18% orders, -50% new customers. Sub-attach rose +17.2pp simply because the remaining tiny new-customer pool happens to skew subscription. The 61→50 absolute base is small enough that this could be noise, but the direction matches the GA4 HP decline (−1.12 pp overall). Other US-specific factors (paid spend, seasonality, ads policy) almost certainly contribute — this cannot be attributed to the copy change alone.
DE — conflict: page-level CVR improved, total sales fellThe GA4 layer showed PP_PRO +7.21 / +4.29 pp (best in the analysis, though small sample) and PP_STARTER +2.77 / +1.32. But Shopify says total orders −15.8%, new customers −20.5%. Three possible reads: (1) page-level CVR gains came on a smaller traffic base — not enough volume to move totals; (2) traffic itself dropped through April for non-copy reasons (ads, seasonality, Easter), masking any positive CVR effect; (3) Shop/Deals changes on 28 Apr hurt customer-journey conversion even though individual PDPs got better. The DE GA4 PP signals are still real, just under-powered to lift total sales in this window.
FR — mild positive both rails+2.6% orders, +5.3% new customers, AOV +7.8%, sub-attach +2.5pp. The GA4 PP_PRO positive (+0.71 / +0.87) is consistent with the Shopify mild-positive picture. Smallest market, slowest mover, but no decline.
Caveat: new-customer detection uses Shopify's numberOfOrders (lifetime snapshot at query time) as a proxy for first-order detection. Customers who placed their first order pre-window but ordered again may be miscounted. UK's +49% is large enough to be real directional signal even with this measurement noise; the smaller numbers carry more methodology risk.
Raw: projects/briefs/starter-bundle-kill-analysis-2026-05-13/dopamine-gaba-shopify-2026-05-13.json
06What this means — recommended next steps
- Keep the PP_PRO + PP_ELITE dopamine/GABA additions across all 4 markets. UK shows the cleanest positive on both rails (PP_ELITE/PRO CVR up + total orders +13.9% + new customers +49.1%). FR is mildly positive both rails. DE has positive page-level CVR even if total sales declined for other reasons. US is the only market where the page changes coincided with overall decline, and the small US base makes attribution unsafe.
- Don't draw conclusions about PP_STARTER from this analysis. The dopamine/GABA change overlapped with the bad-H1-copy window. Use the Tab 1 STARTER analysis (which isolates the H1 effect) for that page's verdict.
- Investigate the DE total-sales decline as a separate workstream. Page-level CVR rose but total orders fell −15.8% / new customers −20.5%. Likely causes worth ruling out: paid-media spend changes through April, Easter timing, competitive pressure, or specifically whether the Shop/Deals changes on 28 Apr hurt journey conversion. Pull DE Google Ads spend + GA4 sessions by source pre/post.
- Investigate the US decline likewise. −50% new customers is alarming on its own, but the 50-order sample base is small enough that one campaign change or seasonality can explain it entirely. Check US Google Ads spend + paid impression share + Meta ad activity pre/post the copy windows.
- Shop & Deals: re-measure with the right metric. Cold-landing CVR doesn't work for navigation-only pages. Query GA4 with
pagePath (anyone who visited) rather than landingPage (only first-arrival sessions) to know if those changes lifted in-session conversion. ~10 min follow-up.
- FR PP_STARTER URL needs manual confirmation. The discovered handle
elite-sample-pack returned zero sessions — FR STARTER customers must be using a different localized URL. Verify the canonical FR STARTER PDP slug, then re-run.