
Zoom in, zoom out
Building individual customer relationships at scale
‘Winwood Reade is good upon the subject,’ said Holmes. ‘He remarks that, while, the individual man is an insoluble puzzle, in the aggregate he becomes a mathematical certainty. You can, for example, never foretell what any one man will do but you can say with precision what an average number will be up to. Individuals vary, but percentages remain constant.’
Arthur Conan Doyle, The Sign of Four
Public transport networks are built on this assumption about aggregate behaviour. Until recently, it didn’t really matter who exactly got on the train or the bus, just as long as enough people did.
Arguably, that approach was creaking and the need for greater personalisation in transport was already evident, but COVID has laid it bare.
Transport can’t rely on the aggregate anymore. The industry needs to build stronger, more relevant relationships with its customers. It needs to get to know them.
This short animation gives an overview of our technology, GKT and its possibilities.
This piece then goes on to show examples of recognising when customers change what they do, so operators can respond.
GKT analyses passenger behaviour at scale. But what does that look like and what does it offer?
Passenger A
Passenger B

This plot shows the number of days per week that Passenger A travelled during the first 36 weeks of 2020.
Before lockdown, they showed some signs of flexible working. Usually travelling 5 days a week but sometimes 1, sometimes 3.
When lockdown happened, they didn’t travel at all. And during the summer months, they travelled reliably 5 days a week.

In this plot, you can see the time in the morning that Passenger B travelled.
Before lockdown it was generally around 07:45.
After lockdown, this person continued travelling but opted to do so about an hour earlier.
Understanding behaviour change
The red line shows when the behaviour was recognised as having changed. The system then tracks back to find when the change happened – that’s the green line.
The variability in the behaviour before a change influences how quickly the behaviour is recognised as different.
So, in plot A, it took a few weeks after travel had initially stopped for this to be recognised as a new pattern rather than part of an existing variable pattern.
Whereas the abrupt return to travel after not having travelled for weeks is recognised quickly.
What does this offer?
It creates the opportunity for operators to segment customer communications based on how customers behave.
If Passenger A had regularly travelled 5 days a week pre-COVID but begun to exhibit more variability after, the operator may have concluded that they were at risk of relinquishing their season ticket.
This transition would represent a revenue risk and the operator could begin targeting them with a more appropriate product to retain the relationship.
Passenger B’s new, earlier travel time means they are no longer travelling at peak time.
The operator may want to encourage that behaviour and could, for example, work with a retail partner to offer a free coffee if redeemed before 07:00.
This offer wouldn’t have been relevant to Passenger A.
A richer picture
Frequency of travel and time of travel are just two data points. Every time a passenger responds to a message creates another data point.
In the examples above, does Passenger A buy another type of ticket? Does Passenger B take up the offer of early morning coffees? What do other similar passengers do?
And the system could include other data points. The obvious ones are demographics but can also include, for example, usage of connecting modes for a mobility as a service virtual circle. The more data points, the richer the picture.
People may be insoluble puzzles (and who doesn’t want to retain a little bit of mystery), but this technology means operators can zoom in to be relevant to their customers, while also zooming out enough to do so at scale.
Do you want to build a better network through stronger customer relationships?