Who our insurance customers are — controlled for cohort maturity.

Two things that contaminate headline numbers if not handled. Both need to be reflected in how we read every chart below.

The typical live insurance customer is a Female, 35-44, on BCBS, in a top-10 state. Nearly half attend only one session and don't return. This describes the center of the customer base, not a narrow slice.

Top quartile of the mature cohort (90+ days tenure) — 131 clients averaging $448 mLTV over 141 mean days of tenure. This is the persona paid media should try to acquire more of.

Three observations about this persona that matter for targeting:

• She's not defined by age — the top quartile is evenly distributed across 25-64. Age-based targeting is a mistake; the psychographic signals are much sharper than the demographic ones.
• She's already attempted weight loss — 91% "Yes" on past weight loss is the strongest single biographical signal we have. Creative that reflects "you've tried this before, this time is different" should outperform generic weight-loss positioning.
• She's covered by BCBS/Aetna/Anthem 83% of the time — UHC is dramatically underrepresented (9% vs 19% in the full cohort). This is the "customer insurer" signal cleanly confirming that UHC customers don't become high-mLTV customers, regardless of tenure.

Top quartile ($448 mean) vs bottom quartile (−$30 mean) of the mature cohort, n=131 each. All comparisons are tenure-matched (everyone has ≥90 days to accumulate sessions), so the differences here reflect real customer attributes, not enrollment recency.

Reimbursement rate looks like a strong mLTV signal on paper — but it's not actionable at acquisition because we can't predict it from anything the customer tells us. Here's what actually drives it, in order of magnitude.

1. Plan tier within payer — the biggest factor (not in our control)

Within a single customer insurer, reimbursement varies 15-25 percentage points based on the employer's specific plan:

Example of state-level variance within a single insurer: Aetna customers in NC reimburse at 73%, Aetna customers in OH at 35%. Same insurer, different employer plans, 38-point spread. UHC is the one payer that reimburses consistently low across all states — that's a UHC-specific terms issue, not a plan-tier issue.

2. Deductible status — the "Deductible / Denial" archetype

75% of the low-reimb archetype (230 of 308 clients) got $0 reimbursed. Mean billed: $232, mean copay: $30, mean outstanding (written off): $143. That outstanding balance is the signature of an unmet deductible — the insurer processed the claim but said the client still owes it.

These aren't bad customers. They're customers whose plan left them holding the bag on a $200 bill. Many would become profitable once their deductible is met — but they churn before that happens.

3. Provider documentation variance — smaller effect, worth investigating

Most providers fall in the 4-17% zero-reimb range. Two outliers:

• Michelle Rodgers — 131 clients, 24% zero-reimb, 40% mean reimbursement rate. Could be documentation quality, could be case mix (she has the highest sessions-per-client which might indicate she's the "stick with it" provider for denied claims).
• Justin Klein — 70 clients, 26% zero-reimb, 44% mean. Similar pattern worth investigating.

4. What does NOT drive reimbursement

Reimbursement rate is not predicted by customer attributes we can target on: age, gender, state (after controlling for insurer), goals, stress level, past weight-loss, anxiety flag, session count. It's primarily a plan-tier and deductible-status outcome — neither of which the customer tells us at quiz time.

Client mLTV builds with tenure — clients who enrolled in September have had 6 months to accumulate sessions; clients who enrolled in March have had 3 weeks. Raw cohort means are a lower bound on the mature-state economics. The pattern is consistent and should be expected to continue.

Mean mLTV at different tenure thresholds (live cohort)

The plateau at ~$177-$182 between 90 and 150 days suggests most mLTV accumulation happens in the first 90 days, then slows. A reasonable estimate of "real" per-client insurance mLTV in the current business is $170-$190, not $143.

Per-cohort mLTV — raw numbers (not tenure-adjusted)

Green bars are mature (90+ days). Red bars are immature — their mLTV will continue to build. Reimbursement rate is stable at 45-52% across all post-fix months — the per-client economics aren't degrading, the cohorts just have less tenure.

Session velocity — a leading indicator that's cohort-independent

Instead of comparing raw mLTV across cohorts, sessions-per-30-days-of-tenure is cohort-independent and predicts mLTV cleanly. Applied to the mature cohort (≥60 days tenure):

Each velocity step up roughly doubles mLTV. This is the cleanest leading indicator for retention intervention — if a client is in the <0.6 bucket at 30-day tenure, they're unlikely to recover without intervention. Worth instrumenting as an early-warning trigger for the retention team.

Four tiers, strongest to weakest. Tier 1 = ship. Tier 2 = act with monitoring. Tier 3 = directional only. Tier 4 = not reliable.

Confidence markers

Providers control sessions; they don't control reimbursement routing. So the right lens for evaluating providers is session-based — attendance, cancellation, no-show, session depth — not raw mLTV. This section also shows each provider's cohort mix, because comparing a provider who mostly got Feb/March clients against one who mostly got September clients is meaningless without tenure context.

Session metrics by provider, using first session attended as the denominator

Once a client has met the provider in session 1, everything that follows is on the practice. This table measures what happens after that first meeting. Post-fix cohort (n=1,698) used for tenure-sensitive metrics; combined cohort (n=3,818) used for attendance/cancel where tenure is irrelevant. Sorted by return rate — the % of clients who came back for a second session.

Benchmark: overall post-fix cohort return rate is 53%. Providers above 65% are meaningfully outperforming the mean after meeting their clients. Return rate measures "did they want to come back at all." Subsequent sessions / 30d measures "how quickly do returning clients come back." Both matter — some providers like Heather Lupkey have lower return rates but high velocity on returners, while Tracy Betz and Val Frank have both.

What the first-session-denominator ranking reveals

• Tracy Betz (74% return) and Val Frank (73%) are the return-rate leaders. Nearly three-quarters of their clients come back after the first session. Their subsequent velocity (0.76-0.80) is also top-tier. This is the pattern to study.
• Lakelyn Lumpkin (67% return, 0.99 subseq/30d) is the velocity leader on a young book. Her clients are coming back fast — but her book is only 38 days post-first-session on average. Will the pattern hold as her book matures? Worth watching closely.
• Heather Lupkey (44% return, 0.75 subseq/30d) has an unusual split. Less than half her clients come back at all, but the ones who do come back quickly. Possible pattern: first session is either a great fit or a bad one, with limited middle ground. Worth discussing with her directly to understand.
• Justin Klein (34% return, 0.27 subseq/30d) is the clear underperformer after controlling for cohort maturity. Book is mature (86 days post-first), clients aren't coming back, and the ones who do come back slowly. Not a tenure artifact. Direct intervention needed.
• Heidi Barbey (55% return on 190 post-fix clients) is at the cohort median. Lifting her return rate to ~65% (the Destini Moody / Andrea Jones tier) on the largest book in the business is the biggest scale opportunity. Her 16.7% cancellation rate is the specific failure point.
• Taylor Larsen (62% return, 0.72 subseq/30d, 90% attendance, 7.9% cancel) is the most balanced high performer — near the top on every axis on a mature book. Closest thing to a reference practice.

Each provider's post-fix cohort mix — for tenure context

Different providers have books skewed to different enrollment periods. Providers whose books are mostly Feb/March 2026 clients will look worse on total-session metrics because their clients have had less time.

Heather Lupkey (88%), Emily Hammon (92%), Candace Sorden (82%), and Lakelyn Lumpkin (68%) have books that are mostly Feb/March 2026 enrollees. Their low session counts are predominantly a tenure effect, not a practice effect. Conversely, Michelle Rodgers (56% mature) and Tracy Betz (67% mature) have relatively old books.

What this means for provider interventions

• Justin Klein (34% return, 0.27 subseq/30d on a mature 86-day book) is the clearest underperformer. Only a third of his clients come back after meeting him, and the ones who do come back slowly. Not a tenure artifact. Priority for direct practice review.
• Tracy Betz and Val Frank (73-74% return rate) are the post-first-session retention leaders. Study what's happening in their first sessions that produces a return rate 20 points above the cohort median.
• Heidi Barbey's 190-client book at 55% return rate — moving her to the 65% tier that Destini Moody and Andrea Jones run would convert ~19 additional post-fix clients into returners. Her 16.7% cancellation rate is the specific failure point. Coaching/practice support, not reassignment.
• Heather Lupkey's 44% return / 0.75 velocity split is unusual enough to be worth a conversation. Low first-session conversion but strong ongoing engagement for returners suggests either a fit/misfit pattern in how clients are being routed to her, or something specific about her first-session style that polarizes. Understand before intervening.
• Candace Sorden (40% return) and Emily Hammon (43%) on young books — still early to judge definitively, but the return rate is meaningful regardless of tenure (it's a post-first measure). Worth flagging for Q2 review.
• Taylor Larsen is the reference practice — 62% return, 0.72 subseq velocity, 90% attendance, 7.9% cancellation on a mature 74-day book. The most balanced high-performer in the data.

Self-pay clients look dramatically more valuable on cash mLTV (+75% per client). Most of that advantage is subscription billing timing, not real per-session value. Once you normalize for revenue matched to sessions actually delivered, OOP and insurance have roughly comparable per-client economics. The cohort is still strategically interesting — different demographics, higher retention, subscription-oriented product — but the size of the advantage was overstated.

OOP vs insurance — cash mLTV and session-matched mLTV side by side

Cash mLTV = revenue collected − clinical cost delivered (what's in the bank). Session-matched mLTV = (median revenue per session × sessions delivered) − clinical cost (what's been earned on services actually provided). The second is the fair per-customer-value comparison.

Session-matched mLTV uses median rev/session ($119 OOP, $127 insurance) as the per-session economic rate, multiplied by sessions delivered. This is conservative — it assumes any revenue above the median rate (e.g., subscription prepayment) doesn't represent realized value yet. At the mature level, session-matched mLTV is within 3% between the two cohorts. At the top quartile, they're essentially tied.

What IS genuinely different about OOP (after the subscription-timing correction)

• Demographics diverge sharply. 22% of OOP clients are 65+ vs 2.5% of insurance (Medicare doesn't cover general nutrition counseling outside diabetes/kidney disease). 24% male vs 14%. This is a genuinely different customer.
• Retention is higher. 61% return rate vs 53% — an 8-point gap that is NOT explained by subscription inflation. These customers are self-selected for commitment.
• Geography tilts different. CA, NY, CT overrepresented vs insurance. Self-pay is a coastal-premium market pattern.
• Subscription model is the value driver, not per-session premium. OOP customers pay a roughly similar per-session rate to what insurance reimburses ($119 vs $127 median). The mLTV advantage comes from subscription commitment, not higher session pricing.

Top OOP persona (mature top quartile, n=22) — on session-matched economics

Provider comparison (self-pay, n ≥ 15)

Only two providers have meaningful self-pay books in the post-fix cohort:

Janet Lau is self-pay-specialized; Lakelyn Lumpkin appears in both books. Both return rates well above the insurance-provider median of 55%.

• This is a bivariate analysis — each cut looks at one variable at a time. Feature analysis is the next step and will replace the Tier 1/2/3/4 framework with a proper feature-importance ranking.
• Mature cohort analysis uses n=524. Adequate for the top-line numbers but too small for some deeper cuts. The mature cohort grows by ~170 clients per month as 60-day-old clients cross the 90-day threshold.
• Quiz capture rates. The Feb-Mar tracking bug affects 42% of the live cohort. Quiz-based signals are measured on the captured subset. Once the bug is fixed, re-run the analysis with ~40% more coverage.
• Multiple comparisons. 30+ significance tests run. Tier 1 findings (99%+ confidence, n>100) mostly survive Bonferroni correction; lower-tier findings should be treated as suggestive pending validation.
• Provider mLTV is not payer-controlled. Session metrics (attendance, no-show, cancel) are what providers control. mLTV differences between providers could be partly payer routing.
• Session velocity (sessions/30d tenure) tapers with tenure. Early clients do more sessions per month than they will at 6-months tenure — that's normal. Use at matched tenure points (e.g. at 60-day mark) for cross-client comparison.
• Clinical cost is flat $60/hr. Contribution margin after labor, not fully-loaded.
• Pre-September data is used only in Section 08 for behavioral metrics (attendance). Never combined for mLTV math.

Top Nutrition Coaching · v6 · 1,698 live insurance clients · Sep 1 2025 – Mar 31 2026 · Cohort-maturity-aware · Feature analysis next