Identity
“What plant is this?”
Identification uncertainty blocks care before it begins.
Independent product conceptBloom · Plant care · 3 iterations
Turn a plant into an editable care plan in under a minute.
Bloom explores one focused promise: identify a plant, adapt generic guidance to the home, and create reminders the owner can change as life and conditions change.
Role
Research · Product design
Research
Survey n=10 · Interviews n=3
Core loop
Identify → plan → care
Validation
3 iterative feedback rounds
Personal concept. Research is directional and based on a small convenience sample.
01 · Research signal
People did not need more plant content. They needed a routine that fit their lives.
A survey of ten classmates and three follow-up interviews suggested three recurring barriers. The sample is too small and narrow for market claims, but it was enough to frame a first prototype.
Context
“Will this survive my home?”
Generic advice ignores light, travel, and routine.
Consistency
“I forgot again.”
Care disappears when reminders cannot adapt.
Survey · 10 masters studentsInterviews · 3 peopleBerlin · ages 27–29Convenience sampleDirectional only
02 · Product scope
One promise for the first release.
Add a plant quickly and leave with a care plan that is understandable, editable, and easy to act on. Everything else is secondary until this loop demonstrates repeat use.
MVP loop
01
Identify
02
Confirm context
03
Edit plan
04
Act on reminder
Deferred
- — Community feed
- — Social comparison
- — Plant recommendations
- — Milestones and gamification
03 · Care-plan flow
Three decisions turn generic advice into an owner-controlled plan.
The screens below are concept artifacts. The important story is the handoff of control at each step.
01
Rank identification instead of pretending certainty.
Signal
A single incorrect result asks the user to trust the model blindly.
Response
Show ranked matches, confidence, visual cues, and manual search fallback.
Value
The owner chooses what enters the collection.
02
Ask only what changes the plan.
Signal
Long setup forms create work before the app has delivered value.
Response
Pre-fill known plant needs, then ask about light, pot, and the owner's routine.
Value
Less setup with visible reasons for every question.
03
Generate a plan users can override.
Signal
A fixed schedule ignores travel, weather, plant condition, and personal routine.
Response
Create an editable schedule with complete, snooze, skip, and reschedule actions.
Value
Guidance stays useful without becoming authority.
04 · Iteration
The best changes came from watching where three people lost confidence.
Three rounds cannot validate a broad product, but they can expose navigation, trust, and setup friction. I used each round to narrow the loop rather than keep adding features.
V1 → V2
“I sometimes get lost.”
Simplified navigation and made the add-plant path explicit.
Findability
V1 → V2
“The identified plant is wrong sometimes.”
Replaced one answer with ranked matches and manual fallback.
Calibrated trust
V2 → V3
“Some details should be auto-filled.”
Pre-filled known care needs and made the rest optional or editable.
Progressive disclosure
05 · Flexible care states
A care plan needs recovery paths, not guilt.
Plant care is variable. The interaction model supports uncertainty and missed tasks without silently changing the plan or pretending to diagnose plant health.
Low-confidence identification
- 1Explain uncertainty
- 2Guide a photo retake
- 3Offer manual search
Missed reminder
- 1Complete later
- 2Snooze or skip
- 3Reschedule explicitly
Plant looks unwell
- 1Ask for observed condition
- 2Offer cautious guidance
- 3Escalate to expert help
06 · Reflection
The project became stronger when it stopped trying to be a plant social network.
Research generated many appealing ideas—community, milestones, recommendations, and photo timelines. Product judgement meant separating delight from the promise that needed proof first.
What changed
The scope narrowed from a broad lifestyle app to identification, editable care planning, and recurring action.
What remains unproven
A three-person feedback loop cannot establish retention, horticultural accuracy, or recommendation quality.
Next evidence
Run a two-week diary study with new owners and measure setup completion, reminder action, overrides, and repeat care.
Explore the interaction model
Identification, setup, care actions, and deferred V3 explorations.