This brand sells daily-essential supplements — multivitamins, protein, creatine and greens. The profit sits with customers who reorder, which makes Subscribe & Save the part of the account that matters most.
The problem was not retention
We audited the account before we won the business. Sales were tracking behind the prior year and both Subscribe & Save revenue and subscription count had gone flat. The brand was signing up about as many subscribers each month as it lost.
Retention was not the issue. Thirty-day sat at 89.3% and ninety-day at 70.6%, well ahead of the 50 to 60% most brands here see. Not enough new people were starting. Ad spend was already at the top of a healthy range, with 37% in upper-funnel video and only 39% in the lower-funnel campaigns that close sales.
The subscriber flywheel
Growth in active subscriptions since July 2025
Both lines start at 0%. Percent growth vs. July 2025.
| Period | Active subscriptions | Prior-year baseline |
|---|---|---|
| Jul | +0% | +0% |
| Aug | +6% | +9% |
| Sep | +8% | +8% |
| Oct | +10% | +8% |
| Nov | +22% | +4% |
| Dec | +27% | +3% |
| Jan | +28% | +3% |
| Feb | +36% | +3% |
| Mar | +36% | +5% |
| Apr | +42% | +1% |
| May | +50% | +1% |
| Jun | +61% | +1% |
- Active subscriptions
- Prior-year baseline
What we changed
Change the scoreboard
We made subscriber acquisition the number we report first and moved ROAS to second. A True Portfolio CAC model separates what it actually costs to win a customer from cheap branded-search orders. A 1.0 ROAS floor and a stop-loss if CAC gets near LTV keep the spending honest.
Rebuild the funnel underneath it
We moved budget out of upper-funnel video into lower-funnel manual campaigns, then added AMC audiences, dayparting and retention targeting. In July we took the funnel off Amazon with a three-month DSP flight that starts on reach and ends on retargeting.
The audit became the playbook
Every problem we flagged in the audit turned into a piece of work with a number attached to it. Nine of them ran at the same time over nine months.
- 01
Flat subscriber trendline
Put subscriber growth at the top of the weekly report and judged spend on LTV against CAC instead of campaign ROAS.
Base at an all-time high, +59% YoY
- 02
Budget skewed upper-funnel
Shifted budget into lower-funnel manual campaigns, took over the old Sponsored Brands and Display setup, and added dayparting and AMC audiences.
Manual campaigns drove ~50% of monthly orders
- 03
Ad creative under-converting
Swapped the talking-head videos for cuts with text on screen, then rolled Sponsored Products Video out across the catalog.
Video now pulls its weight on acquisition
- 04
Acquisition matching attrition
Ran market-basket analysis every month and built combination offers out of the products people already buy together.
Q1 repeat customers +16% QoQ
- 05
Stockouts breaking momentum
Took over demand forecasting and the FBA supply chain. Rolling forecasts, AWD buffer stock, and moving subscriptions to a sister SKU whenever one ran thin.
Subscriber base held through every stockout
- 06
Damage-driven negative reviews
Reported reviews that broke Amazon policy, added an insert that routes support to the brand, and built an email flow asking happy customers to review.
Ratings climbed across every core SKU
- 07
Owned audiences untapped
Rebuilt Brand-Tailored Promotions around cart abandoners and lapsed buyers, and restarted Creator Connections with creators we vetted and sent product to.
New demand from outside paid search
- 08
Listings under-converting
Merged the Essential 30s and 60s under one parent, A/B tested pooling reviews, and rewrote the catalog for how Amazon AI search reads listings.
Post-merge conversion rate ~30–35%
- 09
No off-Amazon halo measurement
Started tracking branded search, awareness pool and search ratio every week so halo was visible, then launched DSP to feed it.
Awareness audience +60% in month one
Proof points
Prime Day
Prime Day did what it needed to
Event window vs. the pre-event lead-in period
| Period | Value |
|---|---|
| Pre-event CVR | 12.65% |
| Event CVR | 22.02% |
Amazon DSP
The DSP flight moves from reach to retargeting
Share of DSP spend by month across the three-month flight
- Month 1 · live
- Month 2
- Month 3
The DSP flight was built to pay off later, not immediately. Reach fills the audience pool in month one; retargeting harvests it by month three. Retargeting was already returning 1.32× ROAS on 23% of the spend.
Where it landed
Subscribe & Save now brings in about 40% of revenue, ahead of the 30 to 35% most consumable brands manage, and it comes from a subscriber base that grows every month instead of holding flat.
The growth is coming from the subscriber base, not from better ad efficiency. We gave up some ad efficiency on purpose. It was the trade the audit called for.
An established subscriber is worth about six times what a one-time buyer is worth, and that figure is up roughly 25% on the audit baseline because subscribers are staying longer and ordering more. That is why we are willing to pay more than most to bring in a new customer — the cost comes back several times over across the life of a subscription, so judging this account on a single month of ROAS gives the wrong answer.