A field-validated decision brief on a $52M mental-health benefit launch — with the population it's meant for.
Launch — but not as currently designed. Concept appeal is strong (74% interest, +27 net), but 68% of post-9/11 veterans would not enroll in the program as drafted. The blocker is not the benefit — it's the onboarding flow and the trust framing of first-touch communications. With three targeted redesigns, modeled enrollment lifts from 23% to 61% in the first 90 days. Timing the launch to Q3 holds; the design does not.
Concept resonates strongly. Stepped-care framing scores +27 net appeal — strongest among post-9/11 vets with 70–100% disability ratings.
Signal · positive · 95% CI ±4.3%Trust gap on existing comms. Net trust score on the current outreach channel is starkly negative. "Atlas" branding tested 3.4× higher when paired with a vet-led messenger.
Risk · acute · drives 41% of dropoutsWon't use as designed. The drafted enrollment flow is the single largest barrier. 81% would enroll if completion takes <10 minutes.
Risk · structural · solvable in 6 weeksFrame matters. "Stepped care" outperforms "tier" or "level" framing 4.2× on positive sentiment and 2.8× on intent-to-enroll.
Signal · directional · validated across subgroupsSubgroup variance is real. Women veterans score peer-led delivery +18 NPS above men. Rural −22 vs. urban on telehealth-only models.
Watch · segment-specific design neededModeled lift if redesigned. Three changes — single-page enrollment, peer-messenger first touch, & "stepped care" framing — lift Day-90 enrollment from 23% → 61%.
Opportunity · high-convictionThe concept earns its launch slot. The current design does not. Atlas should redirect 4–6 weeks of product engineering toward (1) a single-page enrollment, (2) a peer-messenger-led first communication, and (3) "stepped care" framing throughout the member experience. Soft-launch in 2 markets first; full rollout to 4 markets at Day 60 once the redesigned flow clears a 50% Day-30 enrollment threshold.
Four hundred post-9/11 veterans with active VA healthcare and Medicare Advantage dual eligibility participated in a structured, AI-moderated voice interview between May 12–17, 2026. Each interview ran 17–28 minutes (median 22). Recruitment was identity-validated against service metadata at intake; no self-reported eligibility was accepted. Asynchronous fielding meant every participant completed on their own time within a 5-day window. Verbatim depth was captured at survey-scale economics. Interviews were coded by a hybrid model — automated theme extraction with human spot-check on 12% of transcripts — and quantified into the metrics presented in this brief.
Conversations followed an open-prompt framework — three primary discussion areas, with the AI moderator probing follow-ups based on participant responses rather than reading from a script. The same five core questions were asked of every participant; everything else was emergent.
All transcripts were embedded into a vector index for semantic search, auto-coded for topic + sentiment + tone markers (hesitation, conviction, emotional inflection), and quantified at the cohort and subgroup level. Every quoted verbatim in this brief is timestamped and linkable to the source recording. Confidence intervals are reported at 95% throughout. Subgroup deltas are flagged where they cross statistical significance (p<.05).
Three out of every four veterans we spoke with affirmed the concept of stepped-care mental health support without prompting. Strongest resonance was among 70–100% service-connected veterans (84% interest, +41 net appeal). The framing of graduated intensity — meeting the veteran where they are rather than gating access behind clinical thresholds — repeatedly surfaced as what made it different from VA-only or commercial-only options.
The product earns its launch slot. Don't kill the program — fix the design.
The Atlas brand is not the problem; the messenger is. Four out of ten participants cited an inability to trust the source as the reason they would not act on the current outreach — even when they affirmed the concept itself. Substituting a peer-led first touch (a fellow veteran, ideally same-era) lifted message acceptance 3.4× without altering the offer itself. Letters signed by clinicians or executives consistently underperformed; letters introduced by veterans consistently overperformed.
Restructure first-touch comms before launch. The letter as designed will be discarded by the population it's targeting. Peer-led first communication is non-negotiable for Day-1 success.
The current 7-step enrollment flow is the largest barrier to launch success — bigger than trust, frame, or pricing. When participants walked through the proposed flow verbally, dropout intent doubled at step 3 (third-party verification) and tripled at step 5 (clinical pre-screening). Conversely, when shown a single-page hypothetical with attestation-based eligibility, enrollment intent climbed from 23% to 81% — the largest single behavior shift we observed in the entire study.
The economic case is overwhelming. The single-page redesign is roughly 200 engineering hours; the friction it removes drives a modeled ~38-point lift in Day-90 enrollment.
Single-page enrollment with attestation-based eligibility is the single highest-ROI change Atlas can make pre-launch. Engineering ROI is approximately 200× over the first 90 days.
Four framings were tested against the identical underlying benefit description, in counterbalanced order. The differences were stark. "Tier" and "level" framings consistently triggered hierarchical-sorting language ("am I a low-tier patient?") and resistance from participants with prior negative VA priority-group experiences. "Matched care" tested neutrally. "Stepped care" was clearly preferred — described by participants as "starting where I am" and "not having to prove I'm sick enough."
Adopt "stepped care" as the public-facing brand language across all touchpoints — letter, web, app, in-clinic materials. The cost of relabeling is rounding error against the conversion lift.
Three subgroup deltas are statistically significant and operationally consequential. Women veterans favor peer-led delivery materially more than men (+18 NPS). Rural veterans reject telehealth-only models (−22 NPS) and prefer hybrid in-person options. Highly disabled veterans (≥70% SC) score appeal +31 above lower-rated peers. None of these can be addressed by averaging.
| Stepped Care | Peer-led | Telehealth-only | Atlas brand | Single-page enroll | |
|---|---|---|---|---|---|
| Men | +24 | +8 | +22 | −18 | +58 |
| Women | +30 | +34 | +1 | −21 | +62 |
| Rural | +27 | +20 | −24 | −19 | +56 |
| Urban | +25 | +10 | +15 | −16 | +60 |
| ≥70% SC | +41 | +22 | +5 | −12 | +64 |
| 0–30% SC | +10 | +9 | +19 | −22 | +38 |
Build segment-aware delivery from Day 1. Specifically: peer-led pathway for women veterans; hybrid in-person + telehealth for rural members; disability-rating-aware care matching, not gating.
Every claim in this brief traces back to a verbatim. Below: the most representative quotes per theme, with attribution chips and signal strength. The full recording corpus is searchable by topic in the Atlas vtrn.ai workspace.
Atlas faces a clean three-way choice. Each option carries a different risk profile, time-to-revenue, and member-experience trajectory. Quantified estimates below assume Atlas's standard MA acquisition CAC of $312 and a 3.2-year veteran member lifetime.
Proceed with the current onboarding flow and existing communication framing. Take the timing win. Accept the design risk.
Use the 6 weeks before Q3 to rebuild onboarding (single-page, attestation-based), recast first-touch comms (peer-led), and adopt "stepped care" language throughout. Soft-launch in 2 markets first, full 4-market rollout at Day 60 once metrics clear thresholds.
Pause the launch. Conduct broader, larger-N research across all four target markets and additional cohort dimensions (Vietnam-era, family caregivers, etc.) before committing.
| Week | Workstream | Owner | Gate to advance |
|---|---|---|---|
| W1–2 | Single-page enrollment build · attestation flow | Product engineering | Internal usability tests > 85% completion |
| W2–4 | Peer-messenger first-touch redesign · recruit 12 vet ambassadors | Member experience | Letter A/B test ≥ +25 net trust score |
| W3–5 | "Stepped care" rebrand across 14 surfaces | Marketing | Comprehension audit > 80% |
| W6 | Soft-launch markets 1 & 2 · 25% rollout | GTM | Day-30 enrollment > 50% |
| W10 | Full rollout markets 3 & 4 | GTM | — |
| W14 | Day-90 enrollment review · second n=200 validation study | Product strategy + vtrn.ai | — |
| Subgroup | n | Interest % | Net appeal | vs. cohort avg | Sig. |
|---|---|---|---|---|---|
| All | 400 | 74.0 | +27 | — | — |
| Men | 292 | 71.6 | +23 | −4 | n.s. |
| Women | 108 | 80.6 | +38 | +11 | p < .05 |
| Urban / suburban | 272 | 75.0 | +29 | +2 | n.s. |
| Rural | 128 | 71.9 | +22 | −5 | n.s. |
| 0–30% SC | 152 | 62.5 | +10 | −17 | p < .01 |
| 40–60% SC | 96 | 73.0 | +24 | −3 | n.s. |
| ≥70% SC | 152 | 84.2 | +41 | +14 | p < .001 |
| Army | 140 | 75.7 | +30 | +3 | n.s. |
| Marines | 108 | 76.9 | +33 | +6 | n.s. |
| Navy | 88 | 71.6 | +23 | −4 | n.s. |
| Air Force | 64 | 68.8 | +18 | −9 | n.s. |
| Source / framing | Net trust | % would open | % would act | Δ vs. baseline |
|---|---|---|---|---|
| Atlas corporate letter (current draft) | −42 | 31% | 11% | — (baseline) |
| Atlas letter, signed by clinical lead | −18 | 44% | 19% | +24 |
| Atlas letter, introduced by veteran ambassador | +22 | 79% | 52% | +64 |
| Same-era peer veteran (gender-matched) | +38 | 86% | 61% | +80 |
| VSO co-branded (DAV / IAVA / VFW) | +27 | 81% | 54% | +69 |
| SMS from peer (vs. letter) | +14 | 68% | 42% | +56 |
| Design configuration | Day-30 | Day-60 | Day-90 | Δ vs. baseline |
|---|---|---|---|---|
| Current draft (7-step + corporate letter + "tier") | 14% | 19% | 23% | — (baseline) |
| + Single-page enrollment | 28% | 38% | 44% | +21 |
| + Peer-led first touch | 31% | 44% | 52% | +29 |
| + "Stepped care" framing | 36% | 51% | 57% | +34 |
| All three combined (Option B) | 42% | 55% | 61% | +38 |
Field study commissioned by Atlas Federal Health Product Strategy team. Conducted by vtrn.ai between May 12–17, 2026. All audio recordings retained 14 days post-fielding per vtrn.ai data retention policy; transcripts retained 90 days; structured findings retained indefinitely as anonymized signal. Engagement ID: 2841. Brief prepared by the vtrn.ai research operations team. Recipient: Sarah Chen, VP Product Strategy, Atlas Federal Health.
You were buying reduced decision risk on a $52 million launch — answers from the population affected, when being wrong is expensive.
Five days of fielding. Four hundred verified veterans. One brief that tells you what to ship, what to fix, and what the modeled impact is. The recordings, full transcripts, and search interface live in the Atlas vtrn.ai workspace and are accessible to your team for the life of the engagement.
If a board member, regulator, or auditor wants to verify any claim in this brief, every signal traces back to a verbatim recording with timestamp. There is no number on these pages we cannot produce the source for.
Decision-grade access to verified veteran cohorts