TCF Research Initiative — TC Navigator Findings, 2026 YTD
TCF Research Initiative

Understanding the patient experience through data

TCF is building one of the only longitudinal datasets on the testicular cancer patient journey — from first symptom to survivorship. Our research draws from AI-assisted patient interactions, web behavior, and community-sourced data to surface what clinical studies miss. This report covers the first six months of the TC Navigator, January 25 – July 24, 2026.

379
Patient conversations analyzed
3,988
Individual messages
38
Countries represented
01

TC Navigator: First Six Months

Total Conversations
379
Jan 25 – Jul 24, 2026 · 181 days
Avg. Messages per Session
10.5
Median 4 — a small group goes very deep
Deep Engagement Sessions
30
20+ message conversations (7.9%)
Longest Single Session
338
Messages — a newly diagnosed patient in Italy
Conversations Showing Distress
247
65.2% carried worry, fear, or acute distress
Asked on Behalf of Someone Else
43
11.3% were caregivers, partners, or parents

The TC Navigator is not functioning as a search engine. Roughly 65% of conversations carry an emotional charge, and a persistent minority run to dozens or hundreds of exchanges — patients working through surveillance intervals, tumour marker readings, and treatment decisions in real time, often after a clinical appointment has already ended. This signals a population with informational and emotional needs the healthcare system is not meeting between visits.

What patients are asking about

Share of all 379 conversations raising each topic. Conversations often raise more than one, so shares sum above 100%.

Treatment (chemo, surgery)
136 · 36%
Symptoms & Diagnosis
130 · 34%
Recurrence & Surveillance
106 · 28%
TCF Resources & Community
76 · 20%
Caregiver / Family Support
47 · 12%
Mental Health & Coping
42 · 11%
Fertility & Sexual Health
26 · 7%
Awareness & Prevention
20 · 5%
Financial Assistance
17 · 4%
Finding Specialists
16 · 4%
02

Key Findings

Most conversations carry distress

247 of 379 conversations (65%) contained worry, fear, or acute distress — and 28 used crisis-level language, including despair about treatment and, in one case, about living. The Navigator is absorbing emotional weight, not just answering questions. Escalation pathways matter as much as content accuracy.

About one in six conversations is not in English

64 conversations (16.9%) were conducted in 10 languages other than English — Portuguese leading at 34, followed by Dutch, Chinese, Spanish, and French. The non-English share more than doubled between quarters, from 9.4% to 22.4%, while US volume also grew in absolute terms (110 to 128). Demand is rising everywhere; it is simply rising faster outside the US.

The Navigator reaches people before diagnosis

80 conversations (21%) came from people who had not been diagnosed — symptom-anxious, pre-diagnosis, or both. Symptoms and diagnosis is the second most common topic overall, in 130 conversations. This is the hardest population for clinical research to observe, and the one where earlier action changes outcomes most.

Survivorship is the long tail

86 conversations (23%) came from people who had finished treatment. Recurrence and surveillance questions nearly doubled as a share of conversations between the first and second quarters, from 18% to 35%. Finishing therapy does not end the need for support — it starts a different, longer phase of it.

A minority go remarkably deep

The median conversation is 4 messages, but 30 sessions reached 20 or more and six reached 90 or more — a 338-message conversation with a newly diagnosed patient in Italy, a 336-message conversation with a US patient in active treatment, and a 315-message conversation with a Brazilian survivor tracking tumour markers among them. Median length rose from 3 to 4 messages between quarters.

One in four conversations ends with an unmet need

98 conversations (26%) ended in visible frustration, an out-of-scope request, or an unresolved question — and 124 contained at least one low-confidence retrieval from the knowledge base. This is the clearest content roadmap TCF has: the gaps are already labelled by the people who hit them.

03

Growth Trends

Monthly conversation volume
A launch spike, a trough, then a higher plateau
44
Jan*
6.3/day
33
Feb
1.2/day
41
Mar
1.3/day
52
Apr
1.7/day
89
May
2.9/day
66
Jun
2.2/day
54
Jul*
2.3/day
May peaked at 2.9/day; Jun–Jul have held flat near 2.2/day
*Partial months — Jan covers 7 days from launch, Jul covers 24
Compare per-day rates, not monthly totals — two months are partial

Read on a per-day basis, the series is a launch spike (6.3/day in the first week), a February trough (1.2/day), sustained growth to a May peak (2.9/day), and a plateau around 2.2/day since. The plateau still sits roughly double the February floor, but the second quarter is not accelerating.

Topic shift
Fewer visitors, more patients
Treatment (chemo, surgery)
Q1
26%
Q2
43%
Recurrence & Surveillance
Q1
18%
Q2
35%
Symptoms & Diagnosis
Q1
37%
Q2
32%
TCF Resources & Community
Q1
26%
Q2
16%

% of conversations raising each topic. Q1 = Jan 25 – Apr 24 (160 conversations); Q2 = Apr 25 – Jul 24 (219). Treatment and recurrence questions rose sharply while questions about TCF itself fell. Symptom questions eased slightly but remain a third of all traffic — the shift is toward people already in the system, not away from the newly worried.

International reach
Non-US share keeps climbing
31%
Q1 · Jan 25 – Apr 24
42%
Q2 · Apr 25 – Jul 24

Countries represented grew from 19 in Q1 to 28 in Q2, though roughly half that gain is explained by Q2's larger sample rather than genuinely wider reach. Brazil alone accounts for 36 conversations across the period — the second-largest national audience after the United States.

04

Geographic Distribution

238
United States
36
Brazil
9
Netherlands
8
Canada
8
United Kingdom
7
Australia
7
India
6
Denmark
6
Germany
5
France
5
Croatia
40
27 other countries
4
Not geolocated

62.8% of conversations originated in the United States. The remaining 37.2% came from 37 other countries plus four conversations that could not be geolocated — an underserved international patient population navigating TC with limited native-language resources. All four ungeolocated conversations fall in Q1, which slightly inflates the Q1 non-US share shown above (29.5% excluding them).

English 312 Portuguese 34 Dutch 8 Chinese 5 Spanish 4 French 4 Turkish 3 German 2 Italian 2 Arabic 1 Russian 1

Language of the user's own messages: 312 English, 64 across 10 other languages, and 3 conversations with too little text to classify.

05

Methodology

Data Source
Chatbase Export

Complete conversation logs from the TC Navigator assistant hosted on testicularcancer.org — built and implemented by Kenny Kane. 379 conversations, 3,988 messages, 1,839 of them from users.

Date Range
Jan 25 – Jul 24, 2026

181 days, covering every conversation from the Navigator's first recorded session through the latest export.

Analysis Method
LLM Classification

Every conversation was read and coded by a language model against 10 topic categories plus perspective, journey stage, language, distress level, and unmet need. This supersedes the keyword matching used in the Q1 report and is not directly comparable to it.

Privacy
No PII Published

Analysis was conducted on anonymized conversation data. No names, contact details, or identifiers are stored or published, and no verbatim user quotes appear in this report.

Sample Type
Self-Selected

Data reflects only users who engaged with the TC Navigator. It does not represent all people affected by testicular cancer. Geolocation is inferred from network data rather than stated residence, and January and July are partial months — per-day rates, not monthly totals, are the comparable measure.

Channel Mix
Five Entry Points

Agent page 261 · site widget 75 · Instagram 33 · Messenger 4 · internal testing 6. Instagram referrals fell sharply between quarters, from 23 conversations (14% of Q1) to 10 (5% of Q2) — worth investigating as a channel question. The 6 internal test sessions are included in all totals.

06

Research Roadmap

01

Time to Diagnosis Survey

A community-sourced dataset measuring the gap between first symptom and confirmed diagnosis. This is a known blind spot in TC research — most published studies rely on clinical records that begin at diagnosis, missing the pre-diagnosis window where 80 Navigator conversations (21%) currently sit.

02

Content Gap Analysis

Systematic review of the 98 conversations that ended with an unmet need and the 124 that triggered a low-confidence retrieval, mapped against the existing knowledge base. The goal is a prioritized content backlog derived from real unanswered questions rather than assumed ones.

03

Multilingual Resource Expansion

With the non-English share rising from 9% to 22% between quarters and Portuguese now the second most common language, a study of what native-language TC resources exist in Brazil, the Netherlands, and Latin America — and where TCF translation would close the widest gaps.

04

Survivor Quality of Life Registry

An opt-in longitudinal registry for TC survivors reporting fertility outcomes, mental health status, testosterone levels, and quality-of-life markers post-treatment — data that does not currently exist at population scale, and which the 23% post-treatment share of Navigator traffic suggests is badly wanted.

05

Caregiver Experience Study

A survey-based study targeting partners, parents, and caregivers of TC patients — 11.3% of Navigator conversations, rising to 48 (12.7%) once patients who raised family-support questions themselves are included — a population whose experience is systematically absent from existing TC literature despite representing a significant portion of TCF's audience.

TCF Research Initiative · TC Navigator findings, January 25 – July 24, 2026 · TC Navigator technology implemented by Kenny Kane; report compiled by Kenny Kane · Prepared from a complete Chatbase conversation export. Figures are exact counts unless labelled as percentages.