Lime is a Data Company
Using Claude to analyse 3 years of my daily lime bike commute in London
Lime bikes have been my primary mode of transport in London for past 3 years. I would take a lime when it rains, when it’s faster to take the tube, when the destination is more than an hour away, always.
Showing off my 6k+ kilometres stats became a common ice breaker and party trick. I was pretty sure I was a 1% power user, but couldn’t prove it.
That’s when this idea crossed my mind to combine the two best inventions of humankind, GDPR and Claude, and properly analyse my geospatial trail.
I sent lime a request to provide all of my data they store.
Ultra Emerald Commuter
Once I received the archive and fed it to claude, I started from the basics.
Summarise what files are in the archive
- Trip history — every ride with timestamps, GPS coordinates, distance and cost
- App event log — every in-app interaction with timestamp, IP address, device info
- Payment history — subscription charges, trip costs, arrears, refunds
- User profile — name, email, payment card details, account settings
- CRM profile — rider tier, loyalty scores, user segmentation
- Identity verification — selfie photos (multiple poses) and government ID scan
- Customer information — template document with privacy policy links and DPO contact
Bingo!
What marketing data is stored about me in the crm?
| Field | Value |
|---|---|
| Rider Status | Diamond |
| TOP_1_PERCENT_CATEGORY | TRUE |
| USER_SEGMENT | Ultra Emerald |
| RFM | Engaged-High Frequency-Very High Value-Diamond |
| Growth Segment | Engaged |
| COMMUTER_SEGMENT | commuter |
| TOURIST_SEGMENT | local |
| MAJORITY_RIDE_TIME | Weekday Day |
| MINUTE_QUARTILE | 4 (top quartile) |
| AGG | more_than_10_trips |
| MOST_USED_VEHICLE_TYPE | Old LimeBike |
Now I am officially a diamond top 1 percent ultra emerald high rfm local weekday commuter! 🗿
I then proceeded to build some dashboards with my stats.
I remember a day in November ‘25, when I discovered dog 💩 in the basket of a lime bike that i was about to rent. That was the moment I downloaded forest bike app and decided to give it a go. You can clearly see it on dashboards above.
But how much did I spend over 3 years?
Third Space
I then played around with mapping out my rides for a bit.
(nb: for discretion, i filtered out all data points after summer ‘25 when i moved houses and changed jobs - so my current house / office is not on the map)
This turned out to be the best illustration of a concept of third space I’ve ever seen (ironically that’s exactly the name of my gym).
I could also see a bunch of other smaller points of interest on this map - a couple of my favorite brunch spots, favorite park walk, my dentist’s office, my badminton and salsa class locations.
This got me thinking: based on these rides alone, how much info about my life could be deduced?
My favorite brunch spot
As exciting as analysing raw data could be, the real gem is the data that could be derived from it.
I started by asking:
Based on the rides, where do I live? where do I work?
Homes:
- Home 1 (Aug 2023 — Aug 2025): Camden Town / Kentish Town (51.551, -0.115)
- Home 2 (Sep 2025 — present): [redacted]. Moved in September 2025, clean overnight switch
Jobs:
- Work 1 (Aug 2023 — early 2025): King’s Cross / St Pancras (51.538, -0.125)
- Work 2 (Oct 2025 — present): Fitzrovia. Gap in commute signal from ~March to September 2025 between the two jobs.
I was blown away! Not only did it correctly identify where I live and work, but also when I moved, when I left my previous job and when I started a new one, with the gap in between!
identify other spots of interest and figure out what are they
-
[redacted location 1] — Your most visited non-home/work location by far. Every day of the week, peaks 5-8pm, consistent over the entire timeline. Almost certainly a gym. The pattern is textbook: go after work most days, occasionally at lunch or mornings.
-
[redacted location 2] — 43% weekend, peaks Saturday morning and 1pm across all days. This is a brunch/cafe spot. The Saturday 10am-1pm pattern is very strong.
-
[redacted location 3] — never weekend, overwhelmingly Tuesdays, exclusively 12-3pm. A standing Tuesday lunchtime appointment — therapist, barber, physio, or similar.
-
[redacted location 4] — Almost exclusively Thursdays, almost exclusively 5pm. 22 arrivals, only 6 departures (you arrive by Lime but leave differently). A regular Thursday appointment or class that starts around 5-6pm.
It perfectly identified my gym, my favorite Saturday brunch spot, my dentist and my salsa class locations — all without any extra context!
DIY
Having read this, you probably want to do this with your data. Here’s how:
- Request your Lime data using the app, or by emailing dpo (at) li.me
- (recommended) Enable private / zero retention mode in your agent (codex, claude, …)
- Feed the dataset to the agent. Explore it together, talk to it, ask questions, ask to build dashboards for you.
The magic of GDPR + Claude
Now, the most interesting bit - you can do it with any app that has your data - as long as you’re in EU / UK.
Your whoop, deliveroo, uber, revolut and hinge. Your messages, social media and emails.
You can request everything, combine, and analyse with an agent - it would likely tell you things you didn’t even know about yourself.
This article reflects my personal experience. I am not affiliated with, sponsored by, or endorsing Lime, Forest, or any other company.
← Back to home