Slide 1

Slide 1 text

Fleet positioning optimization at felyx How we match demand and supply

Slide 2

Slide 2 text

Agenda § Introduction § Why do we need to optimize our fleet positioning within cities? § How do we optimize our fleet positioning within cities? § Dynamic pricing § Active rebalancing § Future plans § Questions

Slide 3

Slide 3 text

3 Introduction

Slide 4

Slide 4 text

Who are we? 4 Introduction Daan Stroosnier Head of Data & Analytics Mick Eijkens Data Analyst Annanina Koster Data Scientist

Slide 5

Slide 5 text

felyx launched in 2017 and offers shared electric mopeds through an intuitive app, now operating more than 7,500 e-mopeds in sixteen cities 5 Introduction Register, locate, and operate e-mopeds through our app Keyless,app-based (reliable and seamless mobile integration) with built-in 4G &GPS Phoneholder ,to enable easy navigation Over-the air updates Smart anti- theft systems andalarm (siren) High- end braking system Allweather tires Easy interchangeable dual battery solution from Panasonic, with100+km range Stand-alone on center stand (locked mechanically)designed to make relocation difficult silent and electric 2-peopleseat, and legcover in winter 2 Helmets/ luggage boxes unlocked by the app > 50 million km driven Focus on product quality, having the safest and highest industry standard moped on the streets. > 600 thousand customers

Slide 6

Slide 6 text

Our mission is connecting people through smooth, green and shared rides. 6 Introduction With felyx, the idea is to travel quickly, comfortably and sustainably through the city - benefitting from all the advantages without having the disadvantages associated with owning your own e-moped. Offering a new door-to-door mobility service • Adding flexibility and choice to modern travellers • Providing free-floating mobility as a service

Slide 7

Slide 7 text

Felyx has amassed a rich amount of data and is continuously enriching this data by increasing its volume, variety and velocity… 7 Introduction Geo info Dama- ges Batteries App usage Cus- tomer info Promo- tions CS data Ops data Weather Mobility market

Slide 8

Slide 8 text

… which allows for a huge potential to generate business value through increasing operational efficiencies and unlocking customer value 8 Introduction Automated reactivation campaign triggers to increase CLV RDM : Overall pattern well captured Predict market demand & supply dynamics to support CS, Marketing and Ops on operational and tactical decision making Recommendation engines for battery swap and maintenance operation to improve efficiency Service area optimization to monetize ride behaviour

Slide 9

Slide 9 text

We have several use cases to optimize our fleet positioning, but will focus on inner-city fleet positioning during this presentation 9 Introduction Intra-city rebalancing Dynamic pricing Active rebalancing High demand Service area optimization

Slide 10

Slide 10 text

10 Why do we need to optimize our fleet positioning within cities?

Slide 11

Slide 11 text

By nature of our free-floating model, the supply and demand within a city is typically imbalanced 11 Why do we need to optimize our fleet positioning within cities? 11 § Customers can park mopeds anywhere in our service area. Situation Complication § Surpluses and deficits of mopeds § High idle times of some mopeds § Loss of revenue § Vandalism § Complaints from neighbours/municipalities § Low availability of mopeds § Revenue loss § Lower customer satisfaction

Slide 12

Slide 12 text

12 Future plans

Slide 13

Slide 13 text

We have accomplished that dynamic pricing and active rebalancing is running in production and have identified synergy in improvements to increase value creation 13 Future plans Step 3 Text goes here goes here goes here goes here goes here goes here goes here goes here goes here. Current § Scaling to more cities § Improving financial and operational tracking § Improving algorithm accuracy and granularity Past § Create rebalancing data product § Pilot in 6 cities Step 3 Text goes here goes here goes here goes here goes here goes here goes here goes here goes here. Current • Defining a better experimentation set-up • Making the engine more dynamic again • Making a model-based ranking of worst mopeds in town Past § v0.1 tested for feasibility (1 city) § Developed + piloted engine v1.0 (2 cities) § Improved scalability v1.0 and developed v2.0 that eliminates cannibalization § Large scale A/B test with difficulties Future § Integrate moped ranking with active rebalancing § Run experiments to gain more knowledge on e.g. (continuous) effectiveness, price sensitivity § Start using flexible price levels § Automate recommendations for active rebalancing § Fully integrate all demand/supply matching initiatives Dynamic pricing Active re- balancing

Slide 14

Slide 14 text

Thanks for your attention! Questions?