The most consistent operational friction across all three sources. Travel of 30–60+ minutes is common. Several users couldn't book at all due to no availability in their area after paying. The compounding problem: clinic locations are not visible before payment, meaning users commit £249 without knowing if the service is geographically accessible to them. Specifically flagged for: rural areas, Scotland, East Anglia, Plymouth, Bath, Cambridge. A postcode checker before purchase would eliminate a significant share of these complaints and is consistently suggested by users independently.

Appears as the reason for 3-star and 4-star reviews across Trustpilot — users who liked the blood tests but felt the programme didn't deliver on the personalisation promise. Tasks feel generic, don't evolve week to week, and often suggest habits users already practise. The WhatsApp group specifically noted that when task cards do show which biomarkers they target, it "increased motivation" — direct evidence that the fix works when applied. This is not a content problem; it's a connection problem. The logic layer linking results to recommendations is absent or invisible.

Joy is universally praised but consistently flagged in 3–4 star reviews as the reason for not giving 5 stars. Two specific failure modes: conversations that repeat rather than evolve over time, and occasional inaccuracies (wrong information about which tests are included, suggestions that conflict with blood result guidance). For power users in the WhatsApp group, the depth of engagement is high enough that these gaps become visible quickly. Joy's quality is the ceiling on NPS. Raising it from 43 to 50+ runs directly through Joy's memory, accuracy, and conversation evolution.

A specific and pointed signal from the WhatsApp group: a user ran a half marathon and returned to find their plan still showing race preparation tasks. "I feel like I haven't got any support or goals this week." The plan has no mechanism to recognise that a user's life has changed or that a task is no longer relevant. Joy can't fix it directly — it required manual team intervention. This is both a product limitation and a trust signal: users expect the product to listen. When it clearly hasn't, the sense of personalisation collapses entirely.

Multiple WhatsApp users noticed their biomarker ratings (oestradiol, cholesterol) changed without a new test. The clinical team had made a silent age-range adjustment, which is clinically correct — but the communication was reactive, not proactive. The fix happened in the WhatsApp group, not in the app. This creates a credibility problem: if a marker can change rating without explanation, users don't know what to trust. Additionally, when Do Health and NHS ranges disagree (MCV, MCH, ferritin), users have no framework for resolving the conflict. The product needs a protocol for both: proactive notification of any rating change, and clear guidance on how Do Health ranges relate to NHS ranges.

Users have these measurements taken at their clinic appointment but they never appear in the app. Multiple WhatsApp users explicitly asking for them to be integrated into their biological age results. This is a quick win — data is already collected, it just isn't surfaced. Users expect it to be there and are surprised when it isn't. Adding it gives users another dimension of progress to track between blood tests, which is exactly the feedback loop the product needs more of.

The WhatsApp group is proving the demand — users are actively helping each other navigate clinic locations, sharing results, discussing what's working, and providing accountability. This is a product feature waiting to be built, and users are building a DIY version of it outside the product. One Trustpilot user explicitly describes feeling lonely doing the programme with no one to share it with. The Do Health principles framework specifically identifies Connection as a core pillar. The product doesn't yet deliver on it.