The concept is pretty neat really.
Take all the mobile phone mast data between various geographic points (which all mobile operators sell along with anonymised demographic data of mobile phone owner/SIM subscriber).
[snip]
Let’s move from that concept to the real world.
About a decade ago the mobile operators and a few consultancies started pushing the idea of origin-destination matrices from cell exit data, even better multi-mode trip stages would automatically be linked andtrip purpose and other segmentation could magically be inferred. Expensive surveys would be history.
We tried to validate a sample we’d been given. Assigned to the network not even the road flows validated acceptably against count data and the bus data seemed more like random noise.
I can guess why, all the assumptions and sub-models involved in turning cell data into transport data add unknown error.
So please point me at a solid piece of research showing that this stuff validates. I retired over 5 years ago so I’ve no idea what happened since.
Salesmen always tell you this kind of thing works, but often you are buying an expensive random number generator.
You can even see how often that subscriber is travelling etc, where else they go etc - but you can't link back to a specific individual in terms of personal level information.
In the demo, probably O2 were showing, they were proud of linking all the stages of a car wandering around south London with frequent trips to Gatwick and back. So probably a private hire driver. Parked in the same place each night apparently. Pretty close to personally identifiable information in my view.
Then Detica started telling us how good they are at linking apparently unlinked data. Nowadays Palantir are probably doing it even better.
Is everyone happy their movements are being sold to marketing firms and others?
Back on the main topic, I’d really like to see some hard data on these revenue increases from selling low value shortish distance advances. Are there genuinely enough new trips to offset the reduced average fare paid? If so, how much comes from new versus existing users? Does it persist after marketing ceases? What if the non-advance fare was reduced to the new yield, would you get the same revenue with increased user convenience?
www.gov.uk
