Download the model here to follow along with this lesson (click --> opens Dropbox

window --> click "Download" --> click "Direct Download")

In this section, we’ve been building out our product quantitative model.

In this lesson, we’ll establish the base model for Uber.

The key question we are trying to answer is, what is the simplest quantitative
representation of my qualitative model?

In other words, we're trying to build the scaffolding we can build more and more detail
around.
We're not trying to represent every single dynamic and variable in our model just yet.
Instead, we’re creating the base scaffolding we can build around. Here are the seven steps
to building our base model:

1. Setup

2. Outputs

3. Acquisition

4. Retention

5. Monetization

6. Costs

7. Connecting

Let’s dive in.

Step #1: Setup

Our first step is setting up the main components of our model.

 om
.c
What you see here is I’m in the summary sheet.
u ys
Here is my quantitative model for Uber. We’ll go through and set up the main components.

Download the model here to follow along with this lesson (click --> opens Dropbox
pb

window --> click "Download" --> click "Direct Download")

ou
gr
sy
Ea
In this summary sheet, we’ll pull all the major outputs from all the other input tabs into this
sheet. We can think about the summary view the main components anyone on the team
would want to know. You can also see here at the top, I started to put my time scale here
as monthly. We recommend building the model for at least two years. But you can drag it
out even farther if you'd like.

You can also see here, down at the bottom I've set up some of my main input tabs. So we
have one for acquisition, where we break down the different parts of acquisition.

We have one for retention, where we break down the different parts of retention.
 om
.c
ys
One for monetization, where we determine how much revenue we're generating.
u
pb
ou
gr
sy
Ea

And one for costs, which is the area for the major costs important to scaling in our loops.
 om
.c
u ys
pb

Once you have these major components set up, we have the basic set up in scaffolding for
our quantitative model.
ou

Step #2: Outputs

gr

Now we want to identify the outputs we’ll use the model to solve for.
sy
Ea
 om
.c
Here in my Uber model, I'm sitting in the summary sheet.
u ys
pb
ou
gr
sy
Ea

This is where we’ll identify the main outputs we want the model to solve for. These are the
major growth components anybody in the business would want to understand, especially
the executive team. One thing we can think about are the major outputs in each area. Here
I have acquisition, retention, monetization, and cost. We can also think the different value
receivers in the growth model I want to account for.

Uber is a marketplace, so we need to think about both sides, both drivers and riders. You
can see here under acquisition, I want to solve for the outputs of new driver registrations,
as well as new rider registrations. In retention, I want to understand the monthly active
drivers, as well as the monthly active riders. Now when we think about a marketplace like
Uber, we need to understand how these two sides of the marketplace, or these two
different types of value receivers, come together and interact to create liquidation. What
you can see here is I'm tracking the number of rides taken on both types of products, Uber
X and Uber Pool.

But I'm also creating a space for marketplace dynamics to understand how these two value
receivers are interacting. I've created a place for understanding which side of the

om
marketplace is constrained. I've also created two line items for the amount of utilization for
both drivers and riders. This means out of the total potential amount of monthly active
drivers or riders, what percentage are finding liquidation and being able to use the
product?

.c
Under monetization, I've identified a couple of different outputs. The first is GMV, gross
ys
merchandise value, which is the number of rides taken times the total price charged to
riders. Note this isn’t the amount of revenue Uber is getting. Uber only takes a certain
u
percentage of each ride, so for total revenue, we've created another line item here.
pb

Additionally, under costs, we're tracking total costs. So once again, we're identifying the
major outputs. We’re also taking into consideration how the different types of value
ou

receivers interact together and what we need to track around each interaction. Finally, we
are tracking the different types of products changing my model, which are Uber X and
Uber Pool.
gr

Since we've identified the main outputs we want the model to solve for, we can start
sy

working on some of its inputs.

Ea

Step #3: Acquisition

Let's start by breaking down the main components of acquisition.

 om
.c
I'm here in the acquisition tab of our Uber growth model.
u ys
pb
ou
gr
sy
Ea

What I've done first is laid out areas for both the driver and the rider side of the business.
We need to understand the number of new rider registrations as well as the number of new
drivers. Those are the two things we're trying to solve for within the tab of our growth
model.

The first thing we can do is think about each on a very basic level. On the driver side,
where are the majority of our drivers coming from? What are the major loops and sources
of new registrations we need to think about and solve for?

First, I created an area just for linear acquisition channels. The main things we want to think
about are the primary loops driving drivers. The first would be paid acquisition. Uber uses a
lot of paid acquisition with TV ads and billboard ads to drive additional drivers to the
platform.

We also have a cross side network effect. A certain percentage of active riders convert into
becoming drivers for Uber X. In addition, we've identified a financial buyer loop, or an
incentivized referral system with drivers referring a new driver to get a financial incentive.

We need to think about the same thing on the riders side of the equation as well. What are
the major sources of new riders, as well as major loops? The two areas I've identified are a
couple different linear acquisition channels, as well as the use of another financial buyer-
loop of a rider referring another rider to get rider credit. Both of those will add up to new
rider registrations.

The next part is thinking about how to break each one of these down. Let's start on the
driver side with linear acquisition.
We think we're going to get a number of new drivers from direct. As we grow the Uber
brand, a certain percentage of them are signed up to the platform. I've created a space for
a number of new drivers from organic search, which would also be coming through the
linear acquisition model.

The next thing I can think about is the paid acquisition loop. We need to think about the
inputs we're solving for. The output is the number of new drivers from paid acquisition.

om
.c
u ys
pb
ou
gr
sy

We need to think backward in terms of the inputs. The first input I need to know is the
conversion rate to sign up. Out of the impressions I'm getting from my ads, how many of
Ea

them are converting to signing up to be new drivers? So we need to know the number of
impressions per ad, which you see with the conversion variable here. This indicates we
need to know the number of ads we're running, which is going to be a function of the
amount of budget we have to spend on paid acquisition, as well as the amount of money
per ad. So we've created those two variables here.

All of this adds up to our new drivers of paid acquisition. We find the number of ad
placements by dividing the spend by the amount of money per ad. We get the number of
impressions by multiplying the number of ad placements times the average number of
impressions. We get the number of new drivers of paid acquisition from taking the number
of impressions and multiplying it by the conversion rate.

I've also created an area to keep track of my approximate costs of customer acquisition for
paid acquisition. This is a combination of the budget we spent, plus any types of bonuses
or referrals we're giving drivers when they sign up.

The next area we can think about is this cross side network effect.

A certain percentage of riders convert into drivers. The output I'm trying to solve for in this
loop is the number of new drivers coming from the riders side of the business. In order to
understand this, I need to know the percentage conversion rate from active rider to driver. I
get this by multiplying the number of active riders times the conversion rate to drivers.

Now we can think about the financial viral loop on the driver’s side. What I'm trying to solve
for here is the number of new drivers from this financial buyer loop.
 om
.c
ys
Working backward to the inputs, the first thing I need to understand is the conversion rate
u
of referrals to new drivers. This indicates we need to understand the total number of
pb

referrals being sent out. To get the total number of referrals I need to know the number of
referrals per referer. This works backward to understanding the total number of drivers
referring other drivers. To get this number, I need to know the percentage of active drivers
ou

who end up referring at least one person.

You can see this adds up to the number of new drivers of the financial buyer loop. I'm
gr

taking the percentage of active drivers who refer, multiplying it by this conversion rate to
get the total number of refers. Then multiplying that figure by the referrals per referer to
sy

get the total number of referrals sent. Then multiplying that figure by the conversion rate of
referrals to new drivers to get my end output of new drivers of the financial buyer loop.
Ea

We can add up these different sources to get the total number of new driver registrations.

Now let’s switch to the riders side of the marketplace and think about the major sources of
new riders.
 om
.c
ys
The first thing I've created is a basic space for any type of linear acquisition, such as those
u
coming as the Uber brand grows, or those searching for a ride-sharing in the app store or
pb

organic search.

The other area and major loop driving riders is the rider to rider financial buyers loop.
ou

Riders are referring other riders to receive a financial incentive or credit to get free rides.
gr
sy
Ea
 om
.c
ys
This output I need is the number of new rider registrations from the financial buyer loop. I
u
need the conversion rate of referrals to new riders, which indicates I need the total number
pb

of referrals sent out. To get the total number of referrals, I need to know the referrals per
referrer. And to do this, I need to know the total number of referrers. And to get this, I need
to know the percentage of active riders who end up referring.
ou

I can start by calculating the number of new riders of the financial buyer loop. I can add up
the linear as well as the financial buyer loop to get the number of new rider registrations.
gr

As you can see here at the bottom, I'm just summarizing the two major outputs I need, the
number of new driver registration, and the number of new rider registrations, to pull into
sy

my summary tab.
Ea

Step #4: Retention

Now let's look at retention for Uber.

This is the retention tab for our Uber growth model.

The three things I want to figure out are the number of active riders, the number of active
drivers, and the total rides taken, which is the liquidity component between those two
sides of the marketplace. I've set up a side for the drivers and a side for the riders, as well
as an area to track them along the way. The liquidity and market dynamics are happening
in between those two sides of the marketplace. Thinking about retention for a marketplace
business is a little bit more difficult than a non-marketplace business.

We have to think about a few different states. First, we need to understand the total
potential number of drivers and riders in each side of the marketplace. Depending on those
numbers, the drivers create a certain amount of ride capacity. Then riders fulfill the supply
with a certain amount of demand based on the number of rides they're requesting.

The number of actual and active drivers and active riders depends on how those two
things interact. We could have a lot more capacity on the driver's side than we do demand
on the riders side. This means out of our total potential pool of drivers, only a certain
portion of them are going to be active. We can also think about the reverse. We could have
a limited supply of capacity per rides and a lot more demand of rides requested from the
riders side, which means we are not going to be able to fulfill all of those requested rides.
As a result, a certain percentage of those rides will be inactive.

So let's reveal how to solve this. We’ll start with the number of drivers.

We have starting registered drivers, who are people who have registered for the app.
Starting potential drivers are people who have opened up the app and want to use Uber.
Then there’s the total amount of starting active drivers who are able to accept rides based
on demand. Those flow through to a number of key actions, the main ones being the
number of rides per driver per month they can handle.
We have two different products, Uber X and Uber Pool, who each have different ride
capacities. Out of our pool of potential drivers, we’re trying to figure out the total ride
capacity. To do this we take our starting potential drivers and multiply it by the number of
rides per driver, per month, on each product. Then multiply it by the percentage of people
who are accepting Uber Pool, or are Uber X only drivers. Then we add those two up to get
the total ride capacity.

This flows through to our acquisition tab and all of our acquisition loops, which depends on
the number of people flowing through our financial referral loop and others, giving us new
registered drivers down here. Before we get to ending active drivers, the total number of
drivers who are active on the platform, we have to figure out the demand from the riders
side.

We have the same thing on the riders side, with starting registered riders. We also have
starting potential riders, who people opening up the app and trying to request a ride. Then
we have the total number of starting active riders who are people actually completing trips.
What we need to figure out is the total number of rides requested.

The average rides requested per potential rider is 15. So we take the starting potential
riders and multiply it by 15 to get the total amount of rides requested. Now, I can figure out
what my ending active drivers and riders are.

To do this we have drivers creating a total amount of capacity for rides at the top, which is
220 million rides. Drivers are creating a certain amount of rides requested, which is 300
million rides. To get the total rides taken, we take the minimum of those two. You can see
we're getting more rides requested than we can fulfill because we don't have enough
capacity. As a result, the total rides taken is 220 million.

With the total number of rides taken, we can work our way back to ending active riders and
ending active drivers. To do this we take the total rides taken and divide it by the number
of average rides requested per potential rider. So 220 million divided by this 15. This gives
us our ending active riders number of 14,666,667 ending active riders.
As a result, we have more potential riders than we did ending active riders because the
capacity was constrained on the driver's side. So even though we had more people
wanting to use Uber, we ended up with a lower amount because we weren't able to fulfill
the demand.
We do something similar with ending active drivers. But since we're constrained on this
side, this ending active drivers equals our starting potential drivers as we are fulfilling 100%
of our supply for drivers.

Now I can also create an understanding of my marketplace dynamics. The first thing to do
is to figure out the utilization on both the driver and the rider side. In other words, what are
the percentage of total potential drivers or riders who end up being active? We simply take
the ending active numbers and divide it by this starting total potential.

We can also figure out the number of unfulfilled rides on either the driver or rider side and
the specific amount of unfulfilled monthly active drivers or monthly active riders. We can
flip the constraint from supply to demand side, back and forth. So depending on how our
acquisition dynamics and retention dynamics are playing out, the constraint could flip from
supply to demand. This will affect how this marketplace grows over time.

Step #5: Monetization

Now we can look at how acquisition and retention connect with revenue generation.
 om
.c
This the monetization tab for our Uber growth model.
u ys
pb
ou
gr
sy
Ea

The first question we ask is, what do we want to solve for in this specific tab? Since we're a
marketplace, we need to figure out what the GMV. This means the liquidity, or the total
number of rides taken times the average price for a ride. Once again, GMV in a marketplace
is not the total amount of revenue.
But we also want to solve for total revenue. In order to do this, we need to take into
consideration the dynamics of our products behaving differently on the monetization side.

We've created two different sections for our two Uber products because the monetization
for Uber X is going to be different than Uber Pool. Let's look at how we figured out the total
amount of GMV and revenue we earning with each of those products.

In order to understand the total amount of revenue from Uber X, we need to understand
two different components. First, we need to understand the GMV. Then multiply it by the
take rate, or the percentage Uber is taking from it.

om
.c
u ys
pb
ou
gr
sy

For GMV, we need to know the ASP of a specific ride on Uber X, which you can see is
Ea

about $10. To get GMV, we are multiplying the ASP by the total number of Uber X rides
taken, which we figured out in the retention tab. Then we take the GMV number and
multiply it by Uber's take rate, which is 20%, to get our total amount of revenue from Uber
X.

We do the same exact thing for Uber Pool. So we take the ASP, which is going to be much
smaller, six dollars, and we multiply it by the total number of trips taken for Uber Pool to
get the GMV. Then we multiply this by the take rate. At the end, we add up the total GMV
from each product, as well as the revenue, to get our major monetization outputs.

Step #6
Now let’s break down our costs.

The first thing we want to think about is the primary output we're solving for.
We are trying to solve for the total cost, in terms of the costs relevant to scaling our major
loops. We have three different acquisition loops involving money. The first is the paid
acquisition loop on the driver's side. The second is a financial viral loop on the driver side.
And the third is the financial viral loop on the rider side.

For total acquisitions costs for drivers, we pull in the budget we're spending on ads from
the acquisition tab. Then we need the amount of money we're spending on the signup
bonus for our people from this channel. So we take the sign up bonus and multiply it by the
number of people we are acquiring from this channel. We add up these two costs to get
the total costs for paid acquisition.

For financial viral loop costs, we need to know the bonus we're giving to referrers as well
as the referral. On each side, we are giving each person $150. So we take these two and
add them up, then multiply it by the total number of people we acquire from the financial

om
buyer loop.

.c
u ys
pb
ou
gr
sy
Ea

We do the same thing on the rider side. We have the signup bonus for the referrer and the
signup bonus for the referral. We take these two and add them up, then multiply it by the
total number of people signing up from this financial viral loop. At the end, we just add up
the costs from the paid acquisition loop, the costs from the financial viral loop on the driver
side, and the cost of the financial viral loop on the rider side to get our total costs for the
month.
Step #7

Now we want to pull all this together and connect it to complete our model.

There are a few things we did in the summary sheet to layer on the final details for our
base model.
The first thing we did is pull all the major outputs from our input tabs into the summary
sheet. There should be no calculations in the summary sheet. We are just pulling from the
major outputs we solved for in the other tabs.

The second thing we've done is dragged every tab out to our full-time period, which we’ve
set at a few years. The last thing we want to do with our model is add summary graphs at
the bottom.
 om
.c
ys
This means taking our outputs for each tab and creating a simple visual to understand how
u
they change as we adjust the model. For example, in the retention tab, we have graphs for
pb

active riders, active drivers, and total rides taken.

ou
gr
sy
Ea
Once we've pulled everything into the summary tab, dragged all of the different inputs out
over its entire time period, and added the graphs into each different input tab, we've
completed our base model.

om
.c
u ys
pb

Those are the seven steps to build our base model for Uber. We set up the main
ou

components, identified the outputs, broke down acquisition and retention, and looked at
how acquisition and retention connected with monetization. Then we looked at the cost of
scaling our loops and connected all our outputs together in the summary tab. We'll use this
gr

base model in order to layer on additional details in future lessons.

sy

Download the model here to follow along with this lesson (click --> opens Dropbox
window --> click "Download" --> click "Direct Download")
Ea