Unsolicited teardowns of products I use. Written as a paying customer first, PM second — each one names a real friction I hit, then what I'd ship to fix it.
The friction: "Unlimited OPD" in practice required pre-approval from an empanelled doctor before visiting your own — unworkable for ortho, skin, eye, and paediatric cases that need physical examination. Reimbursement uploads for outside visits needed support-team escalation. And app-only onboarding was a wall for senior-citizen members.
What I'd ship: Borrow the banking funds-transfer model. "Add Beneficiary" → members nominate their own doctor to the panel (with a cool-off, then full cashless OPD); "Quick Transfer" → visit any doctor now, cashless capped per consultation. Plus a 3-visit open-upload quota before mandating an empanelled consult, and assisted telephonic onboarding for members who can't do app-only.
The interesting call: the fix isn't loosening controls — it's giving the control a self-service path. Members do the empanelment work for you, and the network grows where members actually live.
The friction: as an owner, the voice assistant failed exactly where hands-free matters most — song search forced touch-screen use while driving, no voice control or readout for cabin temperature, poor recognition of Indian contact names, and no handling of spoken number patterns ("nine double eight").
What I'd ship: a five-part PRD — voice-native music search (streaming-app integration), temperature readout + set-point by voice, contact recognition tuned with contextual understanding of the user's own contact list, numeric-pronunciation handling, and a proactive-assistance layer that learns preferences over time.
The interesting call: in-car voice isn't a feature checklist, it's a safety product — every interaction that falls back to touch is a design failure, so "recognition accuracy on local names and number idioms" is the real North Star, not intent coverage.
The friction: the ISP's website is its top acquisition channel, but the path from landing to enquiry submission leaked users — unclear content, deep navigation, long forms.
What I'd ship: journey-map the landing→enquiry flow against analytics (drop-off points, heatmap friction), then attack the three biggest leaks: content clarity (what happens after I enquire?), navigation depth to the enquiry form, and form length.
The interesting call: for an ISP, the website is not a brochure — it's the top of a feasibility-check funnel, and every field before "check availability at my address" is a tax on conversion.