Pre-launch validation · Confidential

Stepped Care for Veterans.

A field-validated decision brief on a $52M mental-health benefit launch — with the population it's meant for.

Prepared for
Atlas Federal Health
Product Strategy · MA Veteran Segment
Sample
n = 400
Post-9/11 vets · VA + MA dual-eligible
Method
15–30 min voice
AI-moderated · async
Time in field
5 days
Recruitment to brief delivery

Contents

vtrn.ai
Engagement #2841
Field complete · May 2026
Document version 1.0
Confidential — Atlas Federal Health · Pre-launch validation
01
Executive Summary

The decision in 90 seconds.

TL;DR

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.

Five findings drove this conclusion

01
74%

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%
02
−42

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 dropouts
03
68%

Won'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 weeks
04
4.2×

Frame matters. "Stepped care" outperforms "tier" or "level" framing 4.2× on positive sentiment and 2.8× on intent-to-enroll.

Signal · directional · validated across subgroups
05
+18

Subgroup 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 needed
06
61%

Modeled lift if redesigned. Three changes — single-page enrollment, peer-messenger first touch, & "stepped care" framing — lift Day-90 enrollment from 23% → 61%.

Opportunity · high-conviction

Recommendation

Verdict Launch
w/ redesign

Hold the Q3 launch date. Rebuild the onboarding flow and first-touch messaging before going live in any market.

The 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.

Recommendation confidence 84% · High
Confidential — Atlas Federal Health · Pre-launch validation
02
Methodology

How we got the answer.

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.

Cohort composition

Era of service
100%
Post-9/11 (OEF/OIF/OND)
VA healthcare
100%
Active enrollee, Priority Group 1–4
MA dual-eligible
100%
Atlas-eligible, 4 launch markets
Branch mix
35/27/22/16
Army / Marines / Navy / AF
Disability rating
62%
≥70% service-connected
Gender
73/27
Men / Women (oversampled)
Geography
68/32
Urban-suburban / Rural
Mean age
38.4
SD 7.1 · range 25–58

Cohort distribution by branch and disability rating

N = 400 · stacked by SC%

Interview design

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.

Analytical approach

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).

Confidential — Atlas Federal Health · Pre-launch validation
03
Headline Findings

Five findings that decide it.

1

The concept itself resonates — and harder than competing benefits Atlas has tested.

74%
Concept interest
95% CI ±4.3
+27
Net appeal score
vs. +12 prior benefit avg.
62%
"Solves a real problem"
unprompted, n = 248

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.

Concept appeal by subgroup

Net appeal score · 95% CI bars · n = 400
Decision implication

The product earns its launch slot. Don't kill the program — fix the design.

2

A −42 net trust score on existing communications is the single largest enrollment risk.

−42
Net trust · current letter
"distrustful" minus "trusting"
3.4×
Higher trust w/ vet messenger
vs. corporate-letter baseline
41%
Of attrition is trust-driven
cited unprompted

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.

Trust signal by communication channel

Net trust score (-100 to +100) · n = 400
Decision implication

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.

Confidential — Atlas Federal Health · Pre-launch validation
3

Onboarding length is the highest-leverage redesign — and the cheapest to fix.

68%
Won't enroll as drafted
multi-step current flow
81%
Will enroll if <10 min
single-page hypothetical
2.6×
Friction-driven dropout
vs. trust-driven (controlling for both)

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.

Modeled enrollment by onboarding length

Cumulative Day-90 enrollment · projected from stated intent · n = 400
I quit twice and started over. If the first screen had been simpler, I'd have finished it the first night. SFC, Retired · Post-9/11 · Texas · Topic: enrollment_friction · Echoed by 41 other respondents
Decision implication

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.

Confidential — Atlas Federal Health · Pre-launch validation
4

"Stepped care" outperforms every other framing tested.

4.2×
Sentiment vs. "tier"
positive-mentions ratio
2.8×
Intent-to-enroll vs. "level"
post-exposure shift
+19
Net comprehension
"I understood that" rate

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."

Frame test — sentiment by exposure framing

Net positive sentiment · n = 100 per frame, randomized exposure
Decision implication

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.

5

Subgroup variance is real and cannot be ignored.

+18
Women vs. men · peer-led
NPS delta · p < .01
−22
Rural vs. urban · telehealth-only
NPS delta · p < .001
+31
100% SC vs. 0–30% SC
overall appeal delta

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.

Sentiment heatmap by subgroup × topic

Net sentiment scores · darker = stronger ·n = 400
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
Decision implication

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.

Confidential — Atlas Federal Health · Pre-launch validation
04
Voice of the Veteran

What they actually said.

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.

Theme: Trust friction (n = 164 mentions)

It feels like they don't trust me to speak for myself. I've been doing this paperwork dance for fifteen years. If you can't put a vet's name on the letter, I'm not opening it.
SGT, Separated 100% SC Florida · trust_in_va
Sentiment −0.78 Hesitation 0 · High conviction Echoed in 38% of trust-themed responses
If a buddy of mine sent me this — same exact letter, just signed by another vet — I'd read it. Same letter from corporate? Trash.
CPL, Veteran 70% SC Ohio · peer_messenger
Sentiment −0.34 Conviction +0.91

Theme: Onboarding friction (n = 188 mentions)

The form alone took me twenty minutes. I quit twice and started over. If the first screen had been simpler, I'd have finished it the first night.
SFC, Retired Post-9/11 Texas · enrollment_friction
Sentiment −0.62 Disfluency cluster on "form" — 4 hesitations Echoed by 41 other respondents
Every system asks me to prove I served. I served. The DD-214 is in your computer. Why are we doing this dance again?
SSG, Veteran 50% SC North Carolina · verification_burden
Sentiment −0.71 High conviction · sustained

Theme: Stepped-care framing (n = 92 mentions)

I like that. "Stepped" sounds like… I get to start where I am. I'm not the worst case. I just need someone to talk to once a week.
PO2, Veteran 30% SC Washington · framing_resonance
Sentiment +0.68 Hesitation low · slow tempo (reflective)
"Tier" makes me think of insurance plans. Or like, who's more important. I don't want to be a tier-three patient. "Stepped" doesn't feel like that.
CPT, Veteran 10% SC Virginia · framing_resistance
Sentiment −0.18 Conviction +0.74

Theme: Subgroup-specific signal (women veterans, n = 109)

If it's a woman vet on the other end of that first call, I'll engage. If it's anybody else — I won't pick up. Twenty years of being the only woman in the room means something.
CPT, Veteran Woman Post-9/11 California · peer_match · gender_specific
Sentiment +0.42 Conviction +0.96 · representative of segment Echoed in 47% of women-veteran responses
Confidential — Atlas Federal Health · Pre-launch validation
05
Decision Implications

Three options. One recommendation.

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.

Option A · Launch as drafted, Q3 target $52M · 4 markets · Day-90 enroll: 23% (modeled)

Proceed with the current onboarding flow and existing communication framing. Take the timing win. Accept the design risk.

Upside
  • Q3 launch hits planned earnings cycle
  • Engineering team avoids 4–6 week delay
  • First-mover advantage in 2 of 4 markets
Risk
  • 23% Day-90 enrollment ≈ 9,200 members vs. plan of 24,000
  • ~$8.4M acquisition cost wasted on attriting cohort
  • Brand damage in tight veteran community is durable
Option C · Defer launch, broader research before commitment 2026 forfeit · re-validate Q1 2027 · cost: $1.2M sunk

Pause the launch. Conduct broader, larger-N research across all four target markets and additional cohort dimensions (Vietnam-era, family caregivers, etc.) before committing.

Upside
  • Lowest design risk — most data before commitment
  • Surfaces edge cases not visible in n=400
Risk
  • Forfeits 2026 launch entirely; competitor catches up
  • $1.2M sunk research cost without proportional information gain — n=400 is already adequate
  • Sends signal of weakness to internal stakeholders

What happens next under Option B

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
Confidential — Atlas Federal Health · Pre-launch validation
06
Appendix · Quant Tables

The full numbers.

A1 · Concept appeal by subgroup

Subgroup n Interest % Net appeal vs. cohort avg Sig.
All40074.0+27
Men29271.6+23−4n.s.
Women10880.6+38+11p < .05
Urban / suburban27275.0+29+2n.s.
Rural12871.9+22−5n.s.
0–30% SC15262.5+10−17p < .01
40–60% SC9673.0+24−3n.s.
≥70% SC15284.2+41+14p < .001
Army14075.7+30+3n.s.
Marines10876.9+33+6n.s.
Navy8871.6+23−4n.s.
Air Force6468.8+18−9n.s.

A2 · Trust signal by communication source

Source / framing Net trust % would open % would act Δ vs. baseline
Atlas corporate letter (current draft)−4231%11%— (baseline)
Atlas letter, signed by clinical lead−1844%19%+24
Atlas letter, introduced by veteran ambassador+2279%52%+64
Same-era peer veteran (gender-matched)+3886%61%+80
VSO co-branded (DAV / IAVA / VFW)+2781%54%+69
SMS from peer (vs. letter)+1468%42%+56

A3 · Modeled enrollment by design configuration

Design configuration Day-30 Day-60 Day-90 Δ vs. baseline
Current draft (7-step + corporate letter + "tier")14%19%23%— (baseline)
+ Single-page enrollment28%38%44%+21
+ Peer-led first touch31%44%52%+29
+ "Stepped care" framing36%51%57%+34
All three combined (Option B)42%55%61%+38

Limitations & caveats

About this study

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.

Confidential — Atlas Federal Health · Pre-launch validation
Closing note

You weren't buying surveys.

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.

vtrn.ai

Decision-grade access to verified veteran cohorts

Engagement #2841
Field complete · May 17, 2026
Brief delivered · May 17, 2026
Document version 1.0