Skinnify CMO Case · Interactive

01 · Performance by channel

The business scaled 2.4x at the same MER, but growth now runs through one AppLovin campaign and efficiency stepped down over the year.

Last 30 days (10 Oct – 8 Nov 2026) vs the previous 30, then the 12-month trend and each channel.

12 months, month by month

Sep 2025 – Nov 2026. November is 1–8 Nov only (hatched). Pick a view.

Channel ROAS depends on who you ask

Last 30 days. Switch the attribution model; the dashed lines are the breakeven range (~1.5 today, ~1.65–1.7 once shipping costs are loaded).

Repeat buying stops at day 90

Cohorts by first order TW. 90-day repeat rate tripled, but 180-day adds only ~1 point.

Measurement to fix first

These change how every number above should be read.

Buyers say AppLovin really finds them

Post-purchase survey, "where did you first hear from us?" TW. Share of answers per month.

Read: "while playing a game" went from 13% to 52% of buyers. That roughly matches AppLovin's 60% of attributed new-customer orders, so AppLovin is credited about right: the question is its cost, not whether it works. Facebook/Instagram is named first by 32% vs 26% attributed (under-credited), Google by only 4–6% (a harvester). Word of mouth is ~2%.

BOGO adds revenue, not profit per customer

Customers acquired Jul 2025 – Apr 2026, by first-order type, 180-day view TW

Read: BOGO buyers spend more but leave $3 less gross profit than single-pair buyers. A non-legging item in the first basket is the LTV lever: up to $128 gross profit and 22% repeat. With nCAC at ~$70–75, there's no LTV cushion; the plan has to win on CAC and first-basket margin.

Klaviyo closes acquisition; the retention flows barely earn

Revenue by stream, last 90 days, Klaviyo-reported via TW TW

Read: ~85% of flow revenue comes from people who haven't bought yet (welcome + abandon). Flows for existing customers make ~$50k in 90 days; win-back earns $0.10 per recipient. Net list growth was negative in Aug (−1.0k) and Sep (−1.4k).

02 · Audience

The buyer is a woman 45+, and the older she is, the cheaper she converts.

Meta last 30 days Platform, 7-day click, $151.6k spend. On-site behaviour from the TW pixel TW.

Age: share of spend vs share of purchases

Women, Meta Platform. Toggle the metric.

Read: 55+ is 47% of spend and 53% of purchases; 65+ has the best ROAS (2.19) and lowest CPA ($54). 24% of spend still goes to women under 45 at 1.5–1.6, under breakeven.

Who Meta actually reaches

Advantage+ audience segments, every campaign over $1k Platform

Read: warm audiences convert at half the CPA. The CBO "testing" campaign spends 65% on them, so its 2.88 ROAS doesn't carry into ASC (1.35).

Placements: Facebook Feed is best and cheapest

Bubble size = spend. Platform

US states: coastal metros return the least

Meta CPM Platform vs store revenue per Meta dollar TW. A directional proxy: Meta doesn't report purchases by region.

Mobile is 74% of revenue and converts worst

TW pixel sessions TW

Read: a CRO problem for a 55+ buyer on a phone (type size, sizing, trust above the fold), not a targeting problem. See the simulator in Growth levers.

Countries

Meta Platform · store revenue TW

Read: Canada has a third of the US CPM and the best ROAS on only $7k/month: the cheapest incremental volume.

The leak is mobile checkout

Funnel by device, last 30 days TW

Read: mobile adds to cart as well as desktop, but only 37% of mobile carts become orders vs 47% on desktop. Mobile LCP is 3.46s (good ≤ 2.5s). Closing half the gap is worth ~+10% of total orders at the same spend.

Landing pages: revenue per session

Top landing pages, last 30 days TW

Read: the mom-gift advertorial earns $0.38 per session, ~10x worse than the PDP, on 11k sessions a month: fix or cut it. The 2-pair-offer page beats the PDP and both BOGO tests.

03 · Creative formula

Skinnify has the best mechanism story. The winners add a real messenger, early proof and a short cut.

The 3 ATIF ads next to Skinnify's own human winners and the competitors' longest survivors, beat by beat. Click any frame to play the ad from that moment.

Storyboard

Pick a comparison, or toggle ads (up to 5). Border colour = how well the beat is executed.

StrongOKWeak / lateMissing / off-screen

When each beat lands

Every ad on one clock. The dashed line is the 3-second hook window. Toggle seconds vs % of runtime.

Read: in all three ATIF ads the offer starts after 0:47 and reviews only appear on the end card. On Google Demand Gen only 5.8% of A007's impressions reach 25% of the video Platform, so the proof and offer reach almost nobody. Week 50 says everything in 13 seconds.

The formula, element by element

Click a row to see the opportunity for Skinnify. Colour = strength of that element.

Where the trust comes from

Hook gets attention; hold predicts ROAS

Meta videos YTD, $15k+ spend Platform. Bubble = spend, colour = 7-day-click ROAS. Grey = A027/A033 tiny tests.

Read: the strongest hooks (Mum vs. Twin 65%, A028 54%) have the weakest ROAS. Winners pair a 40–50% hook with a 40%+ hold. A033's 27% hook is far below the 45–54% of top videos.

A007 hook test: the hook is the bottleneck, not the body

Google Demand Gen, 3 hook versions, ~$500 each Platform. % of impressions reaching each point.

Read: h1 wins (21.4% view rate, 2.4% completions). The curve flattens after 25%: people who survive the first ~15s mostly finish. AppLovin picked the same h1 (86% of spend).

AI ads pay ~2x the CPM of human winners

AppLovin creative sets YTD, $20k+ Platform. D0 ROAS in the label.

A007 is saturating its audience, not fatiguing

Weekly, AppLovin Platform CPM and CTR; pixel ROAS TW

Read: CPM went $71 → $171 while CTR rose and pixel ROAS held at 2.23: the people who see it still buy. Fix with new persona hooks on the same body, not a new concept.

Which hook AppLovin chose

Share of each concept's spend by hook version Platform

The A027 file in the storyboard is h1, which got 4% of spend. The body is shared, so the rest of the breakdown holds; the winning h3 opening should be reviewed before scaling variants.

AppLovin keeps adding ~20% after day 0

ROAS by attribution window, creative sets YTD Platform

Meta's budget sits on its weakest formats

Meta video assets by narrative format, last 30 days. Bar = spend Platform; label = nCPA TW

Read: founder-story and Nurses-listicle videos take 62% of Meta video spend at $90–97 nCPA. Testimonials and comparisons do ~$62 on a fraction of the money. Statics: ~5% of spend at 2.93 ROAS and $50 nCPA.

Leading with BOGO in the ad costs more per new customer

Meta, last 30 days, by offer shown in the creative. nCPA TW

Read: ads with no explicit offer: 2.57 ROAS, $59 nCPA. BOGO-led ads: 1.63 and $89, on 67% of spend. Every ATIF ad closes on BOGO. Test story-first versions that reveal the offer on the page.

Improvement opportunities, ranked

P1 = brief into the next ATIF batch this week.

04 · Competitors

Fanka is the scale threat, Sweetflexx the closest product. Both beat Skinnify on proof and risk reversal.

Sites and Meta Ad Library Public, Similarweb May–Oct 2026 (panel estimates, relative only).

Offer & risk reversal

Brand sites, 9 Nov Public

Traffic & engagement

Similarweb Public

Where the traffic comes from

Share of 6-month visits by channel, Similarweb Public, approximate

Read: Skinnify is display-led (41.5%, mostly AppLovin). Paid social is 10% vs Fanka's 16%: Meta looks under-used. Fanka buys search 7x harder.

How long the survivors have run

Days live as of 9 Nov 2026, Meta Ad Library US Public. Run length is the best proxy for a winner: the US library shows no spend.

05 · Growth levers

Fix what's leaking, rebuild the creative pipeline, then diversify for New Year.

17 levers from the 90-day plan and the Triple Whale deep dive. Impact and effort are my judgement; the evidence behind each is tagged.

Impact vs effort

Click a bubble or a row. Filter by phase.

What four levers are worth

A simple, transparent model on a non-BFCM base: 12-month average revenue of $856k/month TW. Outputs are Estimate, not forecasts.

90-day roadmap

Day-90 scorecard

06 · How to read this

Two levels of data, two jobs.

Top-down business view

Triple Whale and Shopify: revenue, MER, nCAC, margin, new vs returning, cohorts. Real orders. Decides total spend and the split between channels (Linear and First Click). Triple Attribution is shown but not used for decisions: it credits 1.6x the real orders.

Platform

Bottom-up optimization view

Meta Ads Manager (7-day click), AppLovin Axon (D0 incl. views), Google Ads (incl. engaged views). What the algorithms bid on, so it's the right lens for creative, ad set and campaign decisions. Not comparable across channels.

Test

Incrementality

Geo holdouts and lift tests sit above both and settle each channel's budget. None has been run yet; two are in the plan (AppLovin, Meta).

Sources and caveats

Store skinnify-usa, USD, America/New_York. "Last 30d" = 10 Oct – 8 Nov 2026; 12-month = Sep 2025 – 8 Nov 2026. EU and AU stores are not in Triple Whale.
Triple Whale summary metrics, orders and pixel TW; platform fields in TW's ads table are the platforms' own numbers Platform.
Meta via the Meta Ads API and Supermetrics; Google via Supermetrics; AppLovin via Axon CSV exports (creative sets, assets, daily video files) Platform.
Competitors: brand sites, Meta Ad Library (US active ads; no spend or views shown for US commercial ads), Similarweb panel estimates Public.
Creative beat quality ratings and lever impact/effort scores are my judgement, based on the evidence shown next to them.
The last 30 days include pre-BFCM demand; Nov 2026 is month-to-date.