Survey Analysis · 2026-09-02

TBF: the real avatars in the survey data

192 pooled post-purchase responses, segmented blind, reveal 8 real customer types — and the current avatar routing predicts none of them.

n=192 pooled4 avatar surveysblind-segmentedquote-verified

The answer

The current avatars are not real buyer segments.

All 192 responses from the four TBF avatar surveys with data (Cursed 49, Overlooked 118, Sceptic 19, Ageing 6) were pooled, shuffled, stripped of labels, and segmented bottom-up by a sandboxed model with no avatar or product context. It found 8 people-shaped segments plus a 15% residual of too-thin answers.

Then the labels were joined back: which avatar funnel a buyer came through does not predict which segment they belong to (permutation chi-square on the 9×4 table: chi2 = 28.5, p = 0.24, Cramér's V = 0.22). Every avatar cohort is roughly the same mix of the same 8 types.

The other 4 avatar surveys (Impoverished, Helper, Ailing, Faithful) have zero responses, so they are untested.

This page is the sequel to the earlier Cursed-vs-Overlooked comparison, which found those two cohorts psychographically identical.

Caveat: this says the avatars don't separate BUYERS psychographically; whether they still earn their keep as ad angles (different hooks, different CTR/ROAS) is a Meta-performance question this survey cannot answer.

Responses pooled192
Real segments found8
Avatar→segment p0.24
Residual thin answers15%

Does the avatar routing predict the segment?

SegmentCursed (n=49)Overlooked (n=118)Sceptic (n=19)Ageing (n=6)
Working-Poor Treadmill12 (24%)33 (28%)4 (21%)1 (17%)
Relationship & Belonging8 (16%)17 (14%)2 (11%)0 (0%)
The Ailing Body4 (8%)16 (14%)1 (5%)0 (0%)
Spiritually-Framed5 (10%)9 (8%)4 (21%)3 (50%)
Debt & Disaster Reset6 (12%)10 (8%)2 (11%)0 (0%)
Sandwiched Caregiver5 (10%)6 (5%)0 (0%)1 (17%)
Fixed-Income Elders0 (0%)7 (6%)1 (5%)1 (17%)
Bereaved & Exposed2 (4%)4 (3%)0 (0%)0 (0%)
Residual7 (14%)16 (14%)5 (26%)0 (0%)

If the avatars were real segments, each column would concentrate in different rows; instead the columns are near-identical mixes, and the shuffle test says a random relabeling produces this much separation 24% of the time. The one eye-catching cell — half the Ageing cohort landing in Spiritually-Framed — is 3 people out of 6.

How bias was kept out

Same blinding discipline as the earlier comparison. Responses were exported from the four surveys.* Supabase tables, deduplicated, product mentions redacted, shuffled into anonymous IDs R1–R192 with the source label held back. The sandboxed model derived segments bottom-up in three passes. Its ID partition was programmatically verified (192 IDs, no duplicates, no omissions) and its quotes checked character-for-character. Only after the segmentation was frozen were the avatar labels joined back for the cross-tab, tested with a 5,000-shuffle permutation chi-square.

How the segments were derived

All 192 responses were read in three passes: (1) a skim to draft candidate clusters, (2) a full forced-assignment pass of every ID to exactly one bucket, (3) a misfit re-read in which borderline cases (mostly "money AND health" and "widow vs caregiver vs elder") were re-checked and boundaries tightened once. The data refused a clean "money vs health vs love" topic split: money appears in a clear majority of answers, but it plays very different roles depending on the person. The natural structure is therefore people-shaped: what ruptured or grinds in this person's life, who they are responsible for, and whether they explain their trouble materially, relationally, bodily, or spiritually. The pool settled into 8 substantive segments + 1 residual thin/off-topic bucket — slightly above the nominal range because the finance majority genuinely splits into four different kinds of people (chronic treadmill, acute catastrophe, fixed-income elder, bereaved survivor), not one.

Assignment audit: the nine segment ID lists were programmatically verified to partition R1–R192 exactly — 192 IDs total, no duplicates, no omissions (details in the Assignment audit card below; the exhaustive ID lists themselves are omitted from this page).

Scope note: this is descriptive only; no targeting, messaging, or product recommendations are made. Where respondents mentioned multiple problem domains, assignment followed the dominant identity (the problem that defines who is speaking), not mere mention-count.

Segment 1 — The Working-Poor Treadmill: “I work, but the math never works”

Portrait. Employed (or recently employed) adults, often parents/providers, whose defining condition is chronic: bills exceed income, paycheck-to-paycheck, nothing saved, effort never converts into advancement. Not a single catastrophe — a permanent headwind. Emotional register: tired frustration and quiet injustice ("no matter how hard I work").

Count: 50 (26.0%)

Psychographics. Desires: stability, a paid-off bill, a house, a buffer, dignity, ability to say yes to kids. Pains: bill triage, grocery money, stagnation. Fears: never getting ahead, failing family, retirement slipping away. Voice: plain, terse, pragmatic, occasionally hopeful. Spirituality: low-to-moderate, incidental ("thank you"). Locus of control: mostly externalized (wages, prices, luck) but still effort-believing.

Segment 2 — Relationship & Belonging: lonely hearts, strained marriages, broken families

Portrait. People whose core wound is relational: seeking a soulmate, a cold or gaslighting spouse, adult children who withdrew, estranged daughters/grandchildren, a deported husband, a family that won't get along. Money often appears but is secondary to attachment. Register: longing, sadness, sometimes tearful.

Count: 27 (14.1%)

Psychographics. Desires: reunion, to be chosen/loved, harmony, a "forever" partner, family completeness. Pains: loneliness, rejection by children, marital coldness, mistrust. Fears: dying alone, permanent estrangement, being gaslit. Voice: confessional, intimate, sometimes capital-letter anguish. Spirituality: moderate (blessings, church). Locus of control: mixed — often feels at the mercy of others' hearts.

Segment 3 — The Ailing Body: illness and pain stealing function, work, time

Portrait. Chronic conditions and injuries — diabetes, cancer, autoimmune disease, back/leg/femur pain, asthma, weight, memory — that limit daily life and often the ability to work, with money trouble downstream of the body. Register: worn, bodily preoccupied, sometimes facing mortality.

Count: 21 (10.9%)

Psychographics. Desires: pain to stop, strength for kids, independence, to work again, more time. Pains: daily pain, hospitalizations, fatigue, diagnoses. Fears: running out of time, becoming dependent, illness bankrupting them. Voice: clinical list of symptoms; matter-of-fact endurance. Spirituality: low, occasional 🙏. Locus of control: strongly external (the body, insurance).

Segment 4 — Spiritually-Framed: curses, luck, awakening, and manifested abundance

Portrait. People who explain their problem energetically or cosmically — at one pole, fear of witchcraft/dark clouds/bad luck/dry spells; at the other, serene awakening, third-eye opening, divine alignment, and "abundance" manifestation. Money and love are desired but through a spiritual mechanism. Register: mystical, fate-oriented, from fearful to transcendent.

Count: 21 (10.9%)

Psychographics. Desires: protection, cleared paths, opened third eye, abundance, alignment, peace. Pains: perceived curses, negative energy, misfortune, stagnation of the soul. Fears: evil people, bad luck that "follows," spiritual attack. Voice: incantatory, affirmation-like, emoji (👁️🙏), grateful/humble. Spirituality: the defining axis — very high. Locus of control: cosmic/spiritual; agency via ritual, blessing, vibration.

Segment 5 — Debt & Disaster Reset: an acute financial wound, wanting a clean slate

Portrait. Distinct from the treadmill: a specific rupture — bankruptcy, identity theft, a scam, a failed business, a lawsuit, a hurricane, divorce debt, medical debt, layoff cascade. The dream is not incremental stability but a reset/windfall that erases the event. Register: besieged, wronged, hoping for one big break.

Count: 18 (9.4%)

Psychographics. Desires: debt forgiven, fresh start, the lawsuit/settlement to land, the house repaired. Pains: collections, lawsuits, scams, storms, hospital bills. Fears: losing home/job/relationship to the cascade, never escaping the balance. Voice: itemized grievance, named villains (scammers, "Car Sharks," State Farm). Spirituality: low-moderate. Locus of control: external and event-driven; luck/legal outcomes loom large.

Segment 6 — The Sandwiched Caregiver: holding up parents, children, siblings, veterans

Portrait. Identity organized around caring for someone else — a mother with dementia, a disabled combat-veteran husband, a blind son, autistic grandchildren, a sister's children or treatments — often while being unpaid, unhelped, or even accused of financial abuse. Register: exhausted, duty-bound, sometimes resentful at siblings who don't help.

Count: 12 (6.3%)

Psychographics. Desires: the cared-for to be safe/housed/healed; peace; one helpful sibling; their own health to hold out. Pains: dementia care, accusations, Medicaid logistics, no respite, financial drain. Fears: the dependent's future after they're gone, being blamed, collapse of their own health. Voice: detailed, logistical, morally indignant. Spirituality: moderate (prayer/devotion as refuge). Locus of control: internal-duty but externally squeezed.

Segment 7 — Fixed-Income Elders: too old to earn, too poor to stop

Portrait. Seniors on SSI/Social Security/pensions — or still working at 73–78 — stretching a fixed check, unable to retire, or forced back to work. The problem is structural: income ended but expenses didn't. Register: weary dignity, matter-of-fact scarcity.

Count: 9 (4.7%)

Psychographics. Desires: to retire with dignity, cover the mortgage, help children a little, not be a burden. Pains: checks that don't stretch, age discrimination, declining health that blocks re-earning. Fears: outliving money, working until death, losing the home. Voice: plain, resigned, grateful. Spirituality: moderate. Locus of control: external (government check, age, health).

Segment 8 — Bereaved & Exposed: loss of the anchor, collapse of the floor

Portrait. Mostly older women who lost the household's anchor — a husband dead (one by suicide), or a mother dead and a husband deported (R7) — and with that person the income, home, and emotional footing, now facing mortgages, funerals, and survival alone. Grief is the identity; money is the emergency it created. Register: raw, pleading, blessing-seeking.

Count: 6 (3.1%)

Psychographics. Desires: breathing room, to keep the home / bury with dignity, a blessing, to get back on their feet. Pains: fresh grief, deportation/suicide, sudden poverty, loneliness. Fears: losing the home, an improper burial, hopelessness. Voice: narrative, devastated but "still trying day by day." Spirituality: moderate-high (blessing language). Locus of control: external — lightning struck; agency is survival.

Segment 9 — Residual: thin, list-only, or off-topic answers

Portrait. Not a customer type — responses too terse ("Health and financial"), all-category checklists, product/logistics questions, or gratitude notes that reveal no person. Kept separate so they don't distort the real segments.

Count: 28 (14.6%)

How distinct the segments really are

The single dimension that explains the most variation

Chronic grind vs acute rupture — i.e., is this a permanent condition to endure or an event to recover from? — explains more variance than the surface topic (money/health/love). It is what splits the huge finance mass into Treadmill (chronic), Debt/Disaster (acute), Bereaved (acute relational rupture with financial fallout), and Fixed-Income Elder (chronic, life-stage-locked). A close secondary axis is explanatory frame: material vs bodily vs relational vs spiritual, which is what separates Ailing Body, Relationship, and Spiritual segments from the money majority even when their requests sound identical. Life stage (working-age provider vs senior) and caregiving load are real but tertiary. In short: money is the near-universal vocabulary of this pool, but temporality (grind vs shock) is the grammar.

Assignment audit

Segment counts: 50 + 27 + 21 + 21 + 18 + 12 + 9 + 6 + 28 = 192. The nine ID lists were checked programmatically: 192 IDs total, 192 unique, missing set empty, duplicate set empty — an exact partition of R1–R192. All verbatim quote fragments were checked character-for-character against the source responses (original typos preserved, e.g. "has good a lot of hospitalizations" R145, "is is" R131).

Reading these as avatars

Limits

  1. One survey question — "biggest problem" — is a thin basis; desires and voice are inferred from a single answer.
  2. Buyers only — this says nothing about who clicks but doesn't buy.
  3. The pool is 61% Overlooked-funnel buyers, so segment shares carry that skew.
  4. Roughly 20–25% of responses were genuine toss-ups between two segments, mostly money braided with health or family.
  5. Segments describe who buyers ARE, not which ad angle converts them — that needs testing.

The dominant structure worth building on is chronic grind vs acute rupture, not the current avatar taxonomy.