01 — Strategic thesis
Proven message, unproven delivery
Coco’s broadly-targeted autoimmune ads already convert at roughly $3–4 per result. A separate organic UGC push reached 10M+ views but almost no conversions. The signal is clear: the message works; what’s been missing is directed creative and a funnel built to convert.
The throughline. Across every audience we plan to test, one emotional wound recurs — being dismissed by the healthcare system. Roughly half of women report experiencing medical gaslighting, and women are ~1.5× more likely than men to have health concerns dismissed. No competitor in this category is explicitly marketing to that wound. It is our single strongest, most defensible angle.
Our approach has three moves:
- Double down on what already works — broad autoimmune / invisible-illness women — as the always-on base.
- Expand into specific condition communities (endometriosis, POTS, PCOS) that share the same emotional core, each routed to its own matching App Store page.
- Measure the truth in our analytics, not in Ads Manager — a deliberate workaround for a platform constraint that reshapes how health apps are allowed to optimize (see §06).
We start narrow, prove the funnel beats the old economics, then scale by community — rather than spreading budget thin across unproven audiences on day one.
02 — Audience map
Same customer. We just get more specific.
Demographics are constant — women 30–60, ~95% female user base. What changes is the community and the identity. We segment by condition-identity and emotional truth, because that is what advertorials and listicles convert on.
| Stage | Audience | Why it’s here | |
|---|---|---|---|
| Double down | Autoimmune & invisible illness | Where your $3–4 ads already win. ~80% female; onset 30s–50s — the best age fit. The cross-condition “spoonie” community is the retargeting hub. | Proven |
| Expand | Endometriosis | ~6.5M US women; diagnoses peak ages 35–49. ~90% report being disbelieved. The richest “it’s all in your head” language of any segment. | Test |
| Expand | POTS | 83% first misdiagnosed as a psychological condition — the strongest “it’s not just anxiety” hook. Already showed organic traction from a standing start. | Test |
| Expand | PCOS | Affects ~1 in 10 women; up to ~75% undiagnosed. The largest, most vocal social community in the set. | Test |
| Explore | Perimenopause 35+ | 30M+ US women 40–55; the strongest demonstrated out-of-pocket spend. Needs its own funnel and branding, so it follows the core. | Explore |
| Explore | Fibromyalgia / long COVID | Strong female skew and the same fight-to-be-believed wound; fuzzier sizing, so a later test. | Explore |
Not pursued initially: general health-anxiety (not seen in the current user base) and broad wellness (off-niche). We cite prevalence, community size, and clinical data — not dated social view counts.
03 — Messaging angles
Four ways to say one thing
Every angle points at the same wound from the audience map. “Finally believed” is the anchor — it works across all segments; the others are leverage for specific communities.
Finally believed Anchor
The validation a 10-minute visit never gives her.
Hits hardest · all segments
Always in your corner
Coco checks in daily — she’s not doing this alone.
Hits hardest · autoimmune · fibro
The broken system
Hundreds for ten minutes every few months — Coco’s there daily, for the price of a coffee.
Hits hardest · perimenopause
The answers she was denied
“Coco helped me push for the test that finally got me diagnosed.”
Hits hardest · endo · POTS
The compliance spine. Across all four, the system is the obstacle and Coco is the ally that helps her get care — never a replacement for it. This framing keeps us inside medical-claim rules and Meta policy while still carrying real emotional charge. The “answers she was denied” angle does double duty: it satisfies Apple’s requirement to encourage professional care, which in turn makes the harder-hitting angles defensible.
04 — The data behind the audiences
Why these communities
Each segment is chosen on size, age-fit, and the strength of the dismissal narrative. Confidence noted where data is dated or estimated.
| Segment | US scale | Diagnosis delay | The wound (data) |
|---|---|---|---|
| Autoimmune | ~23.5M+ Americans; ~80% women | Long, varies by disease | Onset 30s–50s. “Spoon theory” identity, invisibility (“but you don’t look sick”). |
| Endometriosis | ~6.5M US women | ~4–8 yrs | ~90% report symptoms disbelieved. Peak diagnosis ages 35–49 — best age fit. |
| POTS | ~0.5–3M (wide range) | Median ~2 yrs | 83% first misdiagnosed as psychological; >50% told “it’s all in your head.” |
| PCOS | ~1 in 10 women; ~75% undiagnosed | ~4.3 yrs | Largest, most vocal online community; “just lose weight” dismissal. |
| Perimenopause | 30M+ US women 40–55 | Information void | Best willingness-to-pay: 21% delay care over cost; only 26% have full coverage. |
| Fibro / ME-CFS | Millions; ~90% CFS female | Long | No biomarker → constant battle for legitimacy; strong spoonie overlap. |
Cross-cutting. ~50% of women report medical gaslighting at some point; women ~1.5× more likely than men to be dismissed. The recurring exact phrases — “it’s all in your head,” “it’s just anxiety,” “all women go through this” — are our creative raw material.
Confidence flags: POTS prevalence spans a ~6× range across sources; autoimmune totals are NIH-era and somewhat dated; the endometriosis delay figure depends on geography. We cite the defensible figures and avoid dated social-view counts.
05 — Competitive landscape
What’s working — and the white space
We mapped the AI-health and companion category to find where Coco can win rather than fight.
| Player | Positioning | Funnel / motion |
|---|---|---|
| Juno | Closest comparable. “24/7 AI health assistant for chronic illness” — clinical tracker. Oxford/UCL credibility anchor; ~125K patients. | Condition-led creator UGC → App Store. Owns the clinical story. |
| Tolan | AI companion (Gen-Z). Not health, but the UGC growth benchmark to copy. | Fresh creator accounts, multiple videos/day → App Store. 3M+ downloads. |
| Doctronic | “AI doctor.” Heavy paid volume; carries safety criticism. | Free chat → paid visits, via a network of condition landing pages. |
| Bevel | AI health coach — longevity / quantified-self. Wellness, not illness. | High-volume creator UGC → trial. Different buyer. |
| Lotus | Free primary care. Funding-led launch. | No meaningful paid motion yet — not a near-term competitor. |
White space for Coco:
- Companion warmth for chronic illness. Juno owns the clinical tracker; nobody pairs it with genuine companion warmth that “gets” flares, fatigue, and being dismissed. That gap is Coco’s exact positioning.
- The “finally believed” angle is unowned. No competitor markets to the medical-gaslighting wound — despite it being the most shareable hook in the category.
- Trust contrast. Doctronic’s safety baggage leaves room for a “supportive, never replaces your doctor, built with the community” position.
- Borrow the engine, skip the fight. Tolan and Bevel validate the creator-UGC mechanics without competing for our buyer.
06 — Measurement & compliance
The constraint that shapes everything
A platform rule change — not a Coco-specific issue — determines how we’re allowed to optimize and measure. Building the plan around it from day one is what keeps the numbers honest.
Meta’s health & wellness restriction (Jan 2025). An app for women with chronic illness is almost certainly classified as a restricted health advertiser. In practice, this limits or blocks sending trial/subscribe events to Meta — so we likely cannot optimize campaigns directly on trial-starts. We optimize on upper-funnel signals (landing-page views, installs, engagement) and reconcile true cost-per-acquisition in analytics.
Our workaround — and why the condition pages matter. Each condition-specific App Store page carries its own unique URL. Routing each audience to its matching page gives us per-condition conversion data on the back end — a clean read on what’s actually converting, in your analytics rather than Ads Manager. The condition pages aren’t just better creative; they’re our measurement layer.
Copy guardrails (built into every template):
- No copy that asserts a personal attribute — never “You have [condition]?” Advertorials let the reader self-identify, which is both higher-converting and compliant.
- Safe: “health coach,” “companion,” “guidance, plans, check-ins.” Off-limits: replacing medical advice, implying diagnosis, “AI doctor/therapist,” unsubstantiated outcome claims.
- Apple requires substantiated claims and that we encourage consulting a professional — which our “get-you-to-care” angle does by design.
07 — Funnel architecture
Three paths, matched to intent
For a skeptical, problem-aware, considered purchase, the research is unambiguous: a pre-sell page beats sending cold traffic straight to the store — the format change alone is the single biggest conversion lever.
| Funnel | Audience stage | Role |
|---|---|---|
| Advertorial | Cold / problem-aware | Primary. First-person story arc; scales Coco’s proven “quote-story” format. |
| Listicle | Solution-aware, comparing | Secondary / retargeting. Educational, scannable, save-worthy. |
| Straight-to-app | Warm / branded | Retargeting and branded traffic only. |
All three route to the matching condition App Store page, so the message stays consistent from ad to app page.
Platform: Meta only to start (Facebook + Instagram) — the single channel we’re advertising on while we prove the funnel. Note: as we hit scale, we’ll slowly begin testing native, Google Ads, TikTok, and other platforms — once the core economics are working on Meta.
The production system. A modular engine: fill one messaging matrix per segment, run it through reusable format templates (advertorial / listicle / ad copy), and output versioned variants per funnel. One source of truth, scalable on-brand output — the directed creative the previous UGC efforts lacked.
08 — Benchmarks & targets
What good looks like
Modeled from RevenueCat’s 2025 subscription dataset (115K+ apps) and category Meta benchmarks.
- Cost per install: model $13–18 (US health skews high and is volatile). CPM: $16–22.
- Trial-start rate: median 6.2% → top-decile 20.3% — the widest lever; the paywall and onboarding are where we win.
- Trial→paid (health & fitness): median ~40%.
- Economics hinge on annual mix: annual retains ~44% at year one vs. ~3% for weekly. Weekly is an acquisition wedge; route hard to the annual plan. Hold LTV:CAC ≥ 3:1.
09 — How we’ll work creative
What’s already winning in this market
We pulled every competitor’s live ads from the Meta Ad Library and sorted by how long each has been running. Since brands kill losers fast, a long-running ad is a proven winner — the most reliable signal available (Meta doesn’t publish impressions for commercial ads). Here are the workhorses, and what we take from each.
Captured directly from the Meta Ad Library, June 17, 2026. “Days running” = time since each ad’s first appearance. These are real, currently-active ads.
DoctronicRunning 192 days
“Smarter healthcare — answers in seconds, even from your couch.”
The category’s single longest-running ad. Leans on 24/7 convenience, runs through a whitelisted creator page into a condition-specific landing page. We borrow the advertorial funnel and creator-whitelisting; we skip the “AI doctor” claim (compliance + their safety baggage).
DoctronicRunning 119 days
“Is this normal? I finally stopped wondering and just asked.”
A proven scroll-stopper — the exact 2am-Googling, self-doubt moment our audience lives in. We run this hook, pointed at finally being believed rather than diagnosed.
JunoRunning 27 days
“Finally an app that understands chronic illness.”
Our closest comp validates our anchor angle word-for-word, with spoonie creator UGC. But they frame it as a clinical companion. White space: own the medical-gaslighting wound with more emotional depth + condition-level message match.
TolanHigh-velocity UGC
“I wish someone told me there’s an app that just listens.”
Pure creator UGC on fresh accounts, multiple videos a day. Different audience (Gen-Z), but the growth engine we copy mechanically: creators who live the problem, problem→solution, high volume.
BevelRunning 139 days
“The everything health app — free.”
Feature- and utility-led ads have staying power, but speak to the longevity / quantified-self buyer — not ours. Confirms our edge is emotional, not a feature list.
LotusRunning 32 days
“If nothing’s worked for your [pain], you haven’t tried this.”
A classic problem-led direct-response opener — easy to adapt per condition as a listicle or advertorial hook (“if nothing’s worked for your flares…”).
The playbook we’ll run. Borrow: Doctronic’s relatable-question hooks + advertorial/condition-landing funnel + creator-whitelisting, and Tolan’s high-velocity UGC engine on fresh accounts. Own (white space): the “finally believed / not gaslit” wound + companion warmth + condition-level message match (ad → matching App Store page) — nobody is running this. Avoid: “AI doctor/therapist” and diagnosis claims, and feature-list ads — both for compliance and because emotion out-pulls features in this market.
10 — Sources
Selected references
Primary and industry sources behind the sizing, competitive, benchmark, and compliance figures in this plan.
Audience & clinical
Dysautonomia International (POTS diagnostic delay) · NP Journal (POTS missed diagnosis) · PMC (autoimmune prevalence in women) · Allara Health (PCOS & endometriosis statistics) · Epic Research (endometriosis by age) · PMC (“take me seriously” endometriosis study) · KFF / Northwell (medical gaslighting) · Join Midi / GoodRx (menopause population & cost survey)
Competitive
Juno (juno-chat.com, Y Combinator, Product Hunt) · Tolan (GeekWire, TikTok UGC engine analysis) · Doctronic (doctronic.ai, STAT News, eMarketer) · Bevel (App Store)
Benchmarks & funnel
RevenueCat — State of Subscription Apps 2025 · Superads (wellness CPI) · Mesha / Triple Whale / Affect Group (Meta health CPM) · Pigeon Digital / DTC Funnels (advertorial vs. listicle by awareness)
Compliance
Foley Hoag (Meta 2025 health & wellness rules) · Meta (personal attributes & personal health policies) · Apple (App Store Review Guidelines §1.4.1, §5.1.3)
Figures are drawn from the sources above as of June 2026. Where sources diverge (e.g. health CPM, POTS prevalence), we model ranges rather than point estimates and flag confidence in-line.