Sunny Within — Product & Acquisition Dashboard

Weekly net sales by product — what drove growth

Toggle products, switch stacked (composition) vs lines (trajectories), and smooth weekly noise. Vertical markers flag the moments that changed the trajectory.

The moments that moved the line

Product detail

Meta spend vs Longevity net sales — the takeoff

Weekly Meta ad spend (bars, left axis) against Longevity Watermelon net sales and the store total (lines, right axis). The dashed markers show when the ignition creative launched and when store sales inflected.

Which campaigns drove Longevity (Feb–Apr 2025 takeoff window)

"Pre" = total spend on that campaign across the prior 3 months (Nov 2024–Jan 2025). The two Longevity campaigns went from near-zero to ~$214k — 85% of spend.

The winning hook × creator × landing-page combinations

Top Longevity ads by spend in the takeoff window. The pattern that unlocked scale: whitelisted-creator UGC video (Ken Lawson) framing Longevity as a smarter, better-absorbed NAD+ alternative, with discount/value-price copy, pointed at an offer PDP.

Creative leaderboard — spend & first-order ROAS

The highest-spend Longevity creatives, grouped by creator, angle, or landing page. Bars = Meta spend; the number on each bar is first-order ROAS (Meta pixel revenue ÷ spend; ~1.0 is breakeven on the first order, with profit coming from subscription rebills).

Every captured creative

Discount depth vs net sales over time

Bars = monthly net sales (left). Line = discount rate — total discounts as a % of gross sales (right). Markers flag the promotional milestones. Final month partial.

What the discounting is telling us

1. Price wasn't the growth lever — creative was. The steepest growth (Mar–Aug 2025) happened while discount depth fell to ~7%. The Ken "smarter NAD+" creative + a modest Welcome Offer converted without deep price cuts.
2. The ladder (T1/T2/T3) was a defensive build. It was constructed in Aug–Sep 2025 — exactly when paid ROAS cracked — and rolled out through early 2026. Deep tiers (T2 = 36% off) are a lever to hold CAC when the creative stops carrying it.
3. 2026 is discount-dependent. Discount rate is back to 11–14% while sales drift down — i.e., paying more margin per order just to hold volume. Feb 2026's 14.3% spike (deep-tier rollout + Strawberry) is the clearest signal of the shift from creative-led to discount-led acquisition.
4. The model tolerates it because of subscription LTV — first orders can run ~28–36% off (near/below breakeven) because the profit is in rebills. But rising discount depth on a plateauing base compresses contribution margin, which is the quiet risk heading into H2 2026.

LTV curves by acquisition cohort

Each line is a monthly cohort — cumulative net sales per customer as they age (month 0 = first order). Higher and steeper = more valuable customers. Lines are colored by acquisition era; the bold black line is the blended average. Newer cohorts are short because they haven't aged yet.

Every cohort — value & how it was acquired

What this means — a plan moving forward

1. The model works — LTV:CAC ~2.5–3× with ~1–2 month payback. First-order AOV (~$90–115) roughly equals CAC (~$75–105), and cumulative LTV reaches $210–260 by month 12. This is a fundamentally healthy subscription business; the issue is acquisition quality drift, not the model.
2. Discipline beats volume — the data proves it. The best cohorts by quality are Sep–Nov 2025 (repurchase 48–50%, LTV:CAC ~3.0), acquired right after the Aug→Sep spend pullback. Cutting the marginal, expensive spend bought higher-intent, higher-retaining customers. Chasing volume did the opposite.
3. Creative — not discounting — built the valuable cohorts. The Feb 2025 Ken "smarter NAD+" cohort jumped first-order AOV from ~$76 to ~$106 and set the LTV ceiling ($224 @12mo). The 2026 pivot to deeper discount tiers coincides with falling first-order AOV ($112→$90) and softening early LTV — deep discounts are buying weaker customers.
4. 2026 cohort quality is eroding — act now. CAC has climbed $78→$118, repurchase is sliding, and same-age LTV (M2/M3) has dropped since April 2026. Provisional LTV:CAC on the newest cohorts is near ~1.5 and falling. The June 2026 cohort ($118 CAC, 21% repurchase) is the clearest warning.
5. The playbook to reverse it: (1) Re-invest in creator-UGC "smarter NAD+"-style creative to restore first-order AOV and ROAS — the constraint is winning creative, not budget. (2) Cap CAC (~$85–90) and re-scale only against efficient demand rather than buying $118 customers. (3) Pull back deep acquisition discounting (T2/T3) that dilutes cohort quality; protect first-order AOV. (4) Double down on retention/dunning to defend the 45–50% repurchase rate — the flywheel that makes ~1× first-order ROAS profitable. (5) Watch Mar–May 2026 cohorts mature: if M6 LTV lands below ~$200 (vs $210–238 norm), treat it as structural and prioritize retention over acquisition.

Methodology — cohorts & LTV

Cohort = acquisition month. A customer belongs to the month of their first order. "New customers" that month = first-ever order in it; the Month 0…N columns = cumulative sales per customer at N months after that first order. Source: Lifetimely (keyed on the customer's Shopify/DTC identity).

Maturity caveat. Recent cohorts are young, so their to-date LTV understates final LTV (the curves extend as they age), and the earliest months' "new" can be over-counted until full prior history is loaded. CAC, first-order AOV, repurchase rate, and same-age (M0–M3) LTV are the reliable early signals. This is DTC only — Amazon customers are not in these cohorts.

LTV by acquisition creator — 3, 6 & 12-month

Lifetimely first-touch cohorts, sliced by the creator in the acquiring Meta ad. Equal-maturity LTV (sales per customer) so it's apples-to-apples. Dashed line = store-average 6-month LTV. Sorted by 6-month LTV.

Retention by creative dimension (first-touch)

What this means — creative as an LTV lever

1. The acquisition mix is skewed toward the lowest-LTV creator. Ken drove ~40% of Meta-acquired customers (7,306) but has the lowest LTV of the majors ($197 at 6mo, $217 at 12mo). Sharone ($238/$267) and Anne ($216/$265) retain 10–23% better — yet are acquired at roughly half Ken's volume.
2. Reallocation is the single cheapest LTV win. Shifting even 20–30% of Ken's budget toward Sharone- and Anne-style creators (and scale-testing Emily's age-defiance/beauty angle, the highest 6-mo LTV at $240) lifts blended LTV with no change in total spend — and improves the LTV:CAC that Tab 6 showed is eroding.
3. Pick creative by downstream retention, not just first-order ROAS. Ken and the "NAD+" hook win on volume/ROAS, but Sharone/Anne win on LTV. Optimizing acquisition purely on day-0 ROAS systematically over-buys the lower-LTV creator. Add a 6-month-LTV guardrail to creative decisions.
4. Retire low-retention angles from acquisition. "Pills suck" ($151 6-mo) and Erewhon/PR social-proof ($168) bring the weakest-retaining customers (small samples, but directionally clear). Keep them for awareness; don't scale them for new-customer acquisition.
5. Point winning creative at the offer PDP. "New PDP with offer" first-touch customers retain slightly better than the "5 reasons" PDP ($188 vs $183 at 6mo) — consistent with its higher ROAS in Tab 4. Default new creative to the offer PDP.
6. Build the creator bench. The business is dangerously dependent on one face (Ken). Sharone and Anne are proven higher-LTV; formalize a roster and a monthly cadence of new creators in their style to reduce fatigue risk and raise the LTV floor.

legacy

MER = combined revenue ÷ total ad spend. aMER = total ad spend ÷ total new customers (Shopify + Amazon) = true blended customer-acquisition cost. Shopify new customers are direct. Amazon new customers are identifiable from the AFN order data (customer id/email) — a customer is "new" in a month if it's their first-ever Amazon order. The full launch-to-now Amazon history (Sep 2025–Jul 2026) is warehoused, so every month uses TRUE de-duplicated counts (customer email, first-seen logic). Amazon settles at ~50% new, so blended CAC runs meaningfully below DTC-only.

1. The business is more efficient than it looks — and more than it used to be. Blended MER rose from ~1.5–2.0× during the 2025 scale-up to ~4.3× today. The revenue decline is largely a deliberate shift from growth (spending $300k/mo at ~1.6× MER) to harvest (spending ~$120k/mo at ~4× MER), not a broken model.
2. DTC-only reporting understates efficiency by ~20–30%. Blended MER runs ~0.7–0.9× above the DTC-only number every month, and the gap widened through 2026 as Amazon grew. Judging Meta on Shopify ROAS alone has been penalizing it for demand that simply checks out on Amazon.
3. There is real re-investment headroom. With blended MER ~4× and LTV:CAC ~2.5–3× (Tab 6), the business is arguably under-spending. The binding constraint is winning creative (Tab 7 — scale Sharone/Anne beyond Ken), not budget. Combined revenue peaked at ~$708k (Jan/Mar 2026); re-accelerating it while holding MER ≥3× is the prize.
4. But re-scaling must fix the quality drift first. Tabs 5–7 show 2026 cohorts are eroding (rising CAC, deeper discounts, lower LTV). Pour spend back in only alongside: higher-LTV creative, less deep discounting, and blended-MER measurement — otherwise you buy volume at falling LTV.
5. Adopt blended MER + combined revenue as the top-line KPIs. Target: combined revenue back toward $700k/mo at blended MER ≥3.5×, with subscription-weighted DTC growth (to protect LTV) and Amazon as the marketplace-capture channel.

Methodology — how "new vs returning" is defined

Shopify (exact). New customers = customers − returning_customers (ShopifyQL). Shopify flags a buyer as returning if they have any prior order in the store's full history, so a "new" customer's first-ever order is in that month. Keyed on the real customer email; exact back to launch (Aug 2024).

Amazon (from AFN order data). Keyed on the stable pseudonymous Buyer Email alias. A customer is new in month M if their first order across all warehoused AFN history falls in M (first-seen logic). The full launch-to-now history is warehoused — Sep 2025 through Jul 2026 — so the first-order-ever test is exact for every month (no baseline gap; Sep 2025 is the launch month). Oct 2024–Aug 2025 predate Amazon (Amazon new = $0 before the late-Aug 2025 launch; an order-count proxy covers the launch sliver).

Cross-channel caveat. Shopify (real email) and Amazon (alias) keys can't be joined, so a person who buys on both is counted once per channel — and Shopify's lifetime test is slightly stricter than Amazon's warehoused-window test, making blended figures marginally optimistic in the earliest AFN months. aMER / nCAC divides ad spend by these new-customer counts.

DTC vs Amazon — monthly net/ordered sales

Stacked bars: Shopify DTC net sales (green) + Amazon ordered product sales (gold); the black line is combined revenue. Amazon temporarily more than filled the DTC dip — combined hit an all-time high (~$708k) in Jan & Mar 2026, above the Aug-2025 DTC peak — then both channels fell together in Q2 2026, so the offset broke down. The marker flags Amazon's launch, which lands exactly on the DTC peak and the Meta spend cut.

Amazon by product (Seller Central, lifetime)

What this means — channel shift, leakage, and how to measure it

1. Amazon launched at the exact inflection. First Amazon sales were the week of Aug 31, 2025 — the same moment DTC peaked ($615k) and Meta spend was cut 57%. Amazon then scaled to ~$100–150k/month (24–28% of DTC) through 2026.
2. Half the "DTC decline" is channel shift, not lost demand. DTC-only fell −31% from the August peak to June; but DTC + Amazon combined fell only −15%. Amazon is absorbing a large share of the demand that used to land on Shopify.
3. Amazon is mostly organic — and that's the leakage tell. Amazon ad spend is ~$8k/month against ~$120k/month in Amazon sales; demand is brand/search-driven. The same Meta creative DTC pays for ("Sunny Within / liposomal NAD") almost certainly seeds Amazon searches that convert on Amazon — which mechanically depresses the DTC-pixel ROAS we watched fall from 0.93 to 0.73 in 2026, even though Meta may still be driving the total demand.
4. Stop judging Meta on DTC ROAS alone. Switch to a blended MER = (DTC net + Amazon sales) ÷ (Meta + Amazon ad spend). On that basis Meta is meaningfully more efficient than the Shopify-only view implies, and the "efficiency collapse" is partly a measurement artifact of a second channel opening.
5. But mind the LTV cost of the shift. DTC is where the 76%-subscription flywheel and first-party data live; Amazon customers are largely one-time (Subscribe & Save aside) and Amazon owns the relationship. Every sale that shifts DTC→Amazon likely trades recurring LTV for a one-time order — so this channel shift quietly pressures the LTV we saw eroding in Tab 6.
6. Plan: (1) adopt blended DTC+Amazon MER reporting immediately; (2) treat Meta as demand-gen for both channels, not a DTC-only line item; (3) defend/resource Amazon (~$1.3M annualized, strong 98% buy box, ~2% refunds) with ratings, inventory and the missing SKUs (NAD+, Creatine+); (4) use DTC-exclusive offers/subscription incentives to steer high-LTV, repeat-intent demand back to Shopify while letting Amazon capture the marketplace-native shopper.

DTC e-commerce aMER — actual vs plan

Solid = actual (Jan–Jun), dotted = forecast (Jul–Dec); dashed grey = the locked plan. aMER = new-customer revenue ÷ media (the Growth Hub definition). The plan is a seasonal curve — it steps down to ~1.02 for the summer lull — so the real gap opens in June, at ~17% below plan.

Month-by-month scorecard — what's driving it

Amazon aMER (context)