McKinsey Round 1 December 2020 Case - Dispatch Radius

brainstorming brainstorming questions case structure case study
Neue Antwort am 9. Feb. 2021
3 Antworten
1,1 T. Views
Anonym A fragte am 9. Feb. 2021

Hello community, curious to know how would you attempt the following problem. Cheers.

One of the fundamental elements of e-hailing is the setting of a dispatch radius:

The example above is a rudimentary representation of how a dispatch radius works:

  • A passenger makes a booking

  • A net is then cast out to look for drivers to be allocated to the passenger

  • The size of the net (i.e., dispatch radius) is limited to a certain distance factor (e.g. net with a radius of X km)

  • If there are drivers found within the net, the "best” candidate one would be allocated to them

  • If there aren’t any drivers found within the net, the passenger will have to book again

​​How would you go about?

  1. Setting an optimal dispatch radius

  2. Identifying the best candidate

(editiert)

Übersicht der Antworten

Upvotes
  • Upvotes
  • Datum aufsteigend
  • Datum absteigend
Beste Antwort
Ian
Experte
Content Creator
bearbeitete eine Antwort am 9. Feb. 2021
#1 BCG coach | MBB | Tier 2 | Digital, Tech, Platinion | 100% personal success rate (8/8) | 95% candidate success rate

Hi there,

This is a great question.

The #1 way to help yourself with any problem is think "what is my objective"? I.e. what is my target variable? What does good look like?

I should prompt you for this, but I'll give you the answer. You want a good driver dispatched quickly. That's it.

So:

1. Radius needs to balance speed to passenger (i.e. the larger the radius, the longer the passenger has to wait) with quality of candidates.

Larger net = better candidates but worse wait. Smaller net = shorter wait but worse candidates.

2. This one's easy - customer rating and/or arrival/delivery speed

In terms of #1, you need to understand the independent distribution of driver quality in the specific location in which you sit then set constraints around the minimum level quality driver you're willing to accept based on customer research of satisfaction thresholds.

^Of course, in denser locations the radius will be smaller and in less populated areas, the radius will be larger.

(editiert)

War diese Antwort hilfreich?
Vlad
Experte
antwortete am 9. Feb. 2021
McKinsey / Accenture Alum / Got all BIG3 offers / Harvard Business School

Hi,

I worked at two ridesharing companies and fundamentally there is no single radius. Usually, you increase the radius over time.

The maximum radius is impacted by ETA to passenger and type of the area (city, rural, and even different coefficients based on the city)

As for how to select the best candidates:

  • There is a myth that driver rating or your loyalty score has an impact - that's not true. You don't get the "Best rated" driver.
  • Ideally, you would like to have a non-stop flow when the driver picks an order even before finishing his current order in the same area. So you should allocate orders to the drivers finishing soon nearby.
  • Surge pricing is important as well. If you don't have enough supply in the area, you can increase the prices to decrease demand and attract more drivers to the area.
  • You may allocate the order to a better taxi class if you don't have enough supply in the current class

Best

War diese Antwort hilfreich?
Clara
Experte
Content Creator
antwortete am 9. Feb. 2021
McKinsey | Awarded professor at Master in Management @ IE | MBA at MIT |+180 students coached | Integrated FIT Guide aut

Hello, can you paste? Image does not load for some reason.

Furthermore, can you clarify whether you did have this case on a McK round? Be careful with the non-disclosure.

Cheers,

Clara

War diese Antwort hilfreich?

Ian

Content Creator
#1 BCG coach | MBB | Tier 2 | Digital, Tech, Platinion | 100% personal success rate (8/8) | 95% candidate success rate
1.088
Meetings
76.156
Q&A Upvotes
227
Awards
5,0
150 Bewertungen
Wie wahrscheinlich ist es, dass du uns einem Freund oder Kommilitonen empfiehlst?
0
1
2
3
4
5
6
7
8
9
10
0 = Nicht wahrscheinlich
10 = Sehr wahrscheinlich