1

Rational Growth
An introduction to growing user signups via
data and analytical thinking


By Sandi MacPherson
Via interviews with Andrew Chen

Kindly sponsored by:


2

Rational Growth
We live in world where it's easy to write code, but still hard to get the code into the
hands of customers and users.

Luckily, the same skills that make technology products possible the analytical thinking
that drives the engineering skills for product development can be applied to
"engineering" the growth of your users as well. The foundation of this thinking is to
build a spreadsheet that models out how people discover and sign up to your product.
By refining this model with real life data, you can to simulate different scenarios and to
prioritize product changes. Finally, by deploying this code to real life and observing its
effects, you can then further refine your model to make better changes in the future.

This eBook will focus on thinking about your signup flow, which is often the highest
point of leverage in your product. Products often lose 80-90% of their users within the
first couple screens and the first couple minutes of their experience. Improving this is
key to having a successful product. But the ideas explained here can be applied
anywhere- whether you're trying to encourage more inviting, sharing, or more.

This is an introduction to the topic there is much more literature out there, but I
hope this is a good starting point to learn to think analytically about growth.

Written by Sandi MacPherson (@sandimac)
Based on interviews with Andrew Chen (@andrewchen)

3
From Visual to Spreadsheet
"All models are flawed, but some are useful" George Box (Statistician), 1979

Before you can start this eBook, you need a product. And you need some users
who are signing up to your product, even a small trickle will do. If you haven't done this
yet, no optimization process will help you! So start there first. Even better is a product
that has deeply engaging usage and a vibrant community around it, even if the userbase
is small. An optimization process will turn good products with small audiences into
good products with big audiences. But rarely will it turn a bad product into a good one.
Once you have some data on signups and growth, the first step in thinking
analytically about your signup process is to create your first model. The goal is to
reduce the complexity of your product down to a simpler, abstract spreadsheet-based
representation. This lets us examine each variable in more depth, and see how they
interact with each other. This offers clues for what you can optimize.
Usually this kind of model works by breaking down a signup process into a row-
by-row representation of what percentage of users finish each step. This lets you
identify bottlenecks, and then brainstorm ideas for how to fill in these gaps.

Roughly, here's your plan of attack:
1. Map out the major steps within your signup process.
2. Enter baseline data into a growth model.
3. Come up with hypotheses for how to increase the %s.
4. Examine their impact by altering the growth model to see the potential effects.
5. Prioritize the potential hypotheses based your calculations.
6. Execute the best idea.
7. Repeat from step 2.




4
Your Signup Flow

One of the foundational elements in crafting an accurate growth model is to think
about your product as a step-by-step funnel from the standpoint of the user, rather than
a content hierarchy. Rather than the homepage to start, think about what a new user
experiences first often some kind of profile page or detail page.
What does your signup flow look like? What are the different actions that are
required for a new user to signup to your website? What are your completion rates at
each step? * Here's an example signup flow:






Based on this flow, and the %s you could reduce this to a spreadsheet model with
something like below, with a few distinct steps:



*Each page in any signup design has at least one action for the user to complete (e.g. enter
details, confirm account, etc.). The click-through rate (CTR) is the percentage of users that
complete that action and continue to the next page.
Page
Action
CTR
Landing
Signups
30%
Profile setup
Ask for name, bio, etc.
60%
Tutorial
View tutorial on homepage
80%
Total users signed up
14.4%
Landing
Profile
setup
Tutorial
60%
30%
Confirmed
new user
80%

5
Build Your Spreadsheet Model
Creating a spreadsheet model is a great tool to simulate potential changes to
your signup flow, allowing you to get an understanding of how those changes will affect
the number of users you're able to sign up and for what cost.
Modeling your signup flows like this isn't intended to give you the "answer" to
what you should do. Instead, it's a forecasting tool, and a different way of thinking, that
gives you some quantitative output in addition to the qualitative decision-making that
you're already used to.
Here are two example signup designs to demonstrate this thinking both
products are made up, but include some rough mockups so that you can understand
the details:

1. DailyDiary will show a website that has a long signup process, versus a
website with a similar signup design with fewer pages and will estimate
changes in both Customer Acquisition Cost (CAC) and converted users.
2. TeamShare will compare a signup design that requires payment versus one
that offers a free trial.







6
Baseline example
Let's say you have a new diary website, called DailyDiary:







1. On the landing page, the
user is prompted for their
full name, a password, and
email.

Often, landing pages are
profile pages or detail
pages, not the homepage.
2. Next the user is
brought to a page asking
for more details, their
notification schedule,
terms to accept, and a
CAPTCHA.

3. The user is then
asked to select what
style of template
they'd like to use.

4. Next the user is
brought to their
homepage and a
tooltip suggests that
they write a new blog
post.

7




After creating this visual flow, the next step is to quantify the drop-off at each step.
To measure this, you can use a funnel-tracking feature from a product like Mixpanel or
Google Analytics, or you can track this manually in your database. For the sake of this
conversation, let's say you come up with the following:

5. The 'new blog
post' page is simple
with few actions, and
makes it very easy
for the new user to
create their first
post.

6. A follow-up confirmation
email is sent to the user
after setting up their
account.

8
Page
Action
CTR
Landing
Signup
25%
Details
More data and captcha
50%
Template
Choose template
75%
Dashboard
Follow tooltip
60%
Post
Create post
50%
% of new users who post
5.6%


When multiplied together, each step's completion rates gives a total conversion
rate of 5.6% who ultimately write a post, which you get from multiplying together the
variables.

Looking at the model above, you can tell exactly where people are dropping off.
This is the first step in trying to understand how to go from 5.6% to something better.
Question is, how would you make this better?



9
How to Brainstorm About Growth

Creativity combined with rapid iteration are the keys to making progress on user
growth. Remember that you can get to 10X growth by a combination of 2Xing a few
different metrics, hitting one out of the park, or getting 10% increases across the
board. They all multiply together to be 10X. If you can brainstorm a lot of ideas, going
for quantity over quality, you'll have a lot of ideas to evaluate for impact versus cost.


A few tips on getting the most out of your ideas:
Think about each step of your flow and come up with as many ways as possible to
make that step better, with a higher % conversion.
Try to shorten the flow. What could you take out and ask the user to do later?
What's really required? Try to make it possible to sign in with just the mouse, and
don't require anything on the keyboard.
Try to rearrange the flow so that the highest value proposition for the user comes
firstnobody likes to start by giving their email address
Explore classic landing page techniquestry very long pages or very short pages.
Simplify each page, removing navigation and distractions so that there's just a
single call to action for what the user can do. Don't encourage the user to explore
the siteget them to sign up first and foremost.

There are tons and tons of ideas out there, and many I get out of inspiration from
looking at other products. Try out a lot of other well-designed products and see what
you can experiment with too.



10
Baseline example

For the sake of this discussion, let's take the classic method of trying to shorten
the signup process and making it so that the user can do as little as possible to get
started. To see how the model plays out, let's say you remove 2 pages from the flow to
get the user to create a blog post.
It might look something like this:



1. This landing page
asks the user for their
details, and includes a
CAPTCHA and terms of
service. The page is a
bit cluttered, so the
CTR of this page might
drop versus the
previous simpler
landing page.

2. The new user is
brought immediately
to their new
dashboard. A tooltip
shows them how to
create their first blog
post.

11


In a typical situation, you should make mockups of a simplified signup flow like
this as well as complete your spreadsheet model to try to figure out what would happen.
When you do this, it's hard to figure out what are realistic expectations for how your
numbers would change, but with experience, you'll get better and better at simulating
changes.
In this proposed set of changes, you've combined the first two pages of the
original signup design into one page, and have removed the 'Choose Template' page.
You could guess as to what this might result in by removing and combining a few
variables in your signup flow, which might look something like this:
Page
Action
CTR
Landing
Enter details and CAPTCHA
20%
Dashboard
Follow tooltip
80%
Post
Create post
50%
% of new users who post
8.00%
3. The 'new blog post'
page is the same as the
previous version, and
aims to get the new
user to create their first
post. Also similar to the
previous example (but
not shown here), a
confirmation email is
sent to the new user to
confirm their account.

12

This spreadsheet is now the basis of your forecast of what might happen when
these changes are made. The idea here would be to get from your original conversion
rate of about 5.6% to something more like 8%, which is a relative increase of about
+40%. Is this worth it? Are there other product changes that might make a bigger
impact? This is the kind of thought process you can go through to figure out how to
prioritize your growth projects.
Of course, reality is what counts. The next step would be to start implementing
these changes on our website, and track CTRs to see how our estimated data compares.
As you get more data from your own iterations and within the context of your product
and audience, you'll get more comfortable guesstimating in your models.
You might wonder, what's the point of this if you can't get exact numbers right
away? That's the point of modeling this out. No one will have your particular audience,
product, and signup flow, so each set of numbers is entirely situational. The most
important thing is that this can help you think systematically about your decisions.
You can create conservative versus aggressive models and fill it in with any variables
you do know, and use that to figure out if it's worth it.
You'll never get a black and white answer for this, but it's better than randomly
adding product features based on the emotions and needs of the team. While that's
great for refining core product, when it comes to systematically growing a user base, a
more scientific approach is warranted.

13
Convert Now versus Free Trial

Here's another scenario: How would CAC and conversion rate compare between
one website that requires new users to signup (and pay) immediately, versus one that
offers a free trial? You're probably thinking that the free trial conversion rate would be
higher, but do you know by how much?
You can use the same approach as before to estimate the before and after CAC
and conversion rates. Let's start with your example, TeamShare, for team collaboration:





1. The customer arrives at
the landing page and is
prompted to signup and
purchase a plan.

2. A plans and pricing
page is displayed, and
the new user selects a
plan to continue.
Bundles are offered to
encourage larger up-
front purchases.


14




Before changing the signup design to one that offers a free trial, you'll use a
signup model to simulate the changes and estimate the resulting changes. Having a
model of your signup metrics (based on actual or estimated data) before making the
changes versus making changes without first consulting one helps you forecast the
potential impact and prioritization of this change.

Page
Action
CTR
Landing
Get more info
50%
Plan/pricing
Choose plan
20%
Signup
User/payment details
3%
Total paying users signed up
0.30%

3. The customer is then
prompted for their contact
and payment information.
An optional promotional
code allows for sales reps to
encourage customers to
signup to receive deals.

4. The dashboard is
presented and the
customer can begin to
use the full version of
the website.


15

You've estimated a current total conversion rate of paying users at 0.3% of those
that visit the landing page, which equates to a CAC of $200.00 (based on a CPC of
$6.00).

Let's adjust your model to simulate a signup design with a free trial. You're going
to estimate increases in CTR on both the Plan/Pricing and the Signup pages versus the
original required payment signup design CTRs.

Page
Action
CTR
Landing
Get more info
50%
Plan/Pricing
Choose plan
30%
Signup and usage
Lots of user activity
60%
Freemium
Payment details
15%
Total free trial users signed up
1.35%


Assuming that a large % of users become engaged, you might ultimately find that
a large % of users who sign up ultimately end up converting to a premium offering.

Your model of the free trial signup design gives you a much better conversion rate
of 1.35%, resulting in a CAC of $44 drastically lower than the original $200.00.
However, keep in mind that this kind of model only suits specific kinds of products that
have high-retention and where users have a path to a freemium wall.

However, this data is integral to your decision making process when considering
whether or not to offer a free trial, and the costs associated with doing so.
For completeness, let's see what this new signup design might look like:


1. The customer arrives at the
landing page and is prompted to
sign up for a free trial or to
review the details of the paid
plans.


16







2. A plans and pricing page is
displayed and a free trial of a
plan is selected. The new user
is expected to begin a free
trial, so that option is the
most prevalent.
3. The new user is then
prompted for their personal
and work details.

4. The new user is brought
to their dashboard. A notice
on the top of the page with a
countdown clock shows how
much time remains in their
trial.

17






It's important to note that after you create this model, and these mockups, the
next thing you have to do is to evaluate the ROI of the experiment versus the other ideas
you have. Then if it makes sense, implement the experiment (most likely in some kind
of A/B test), check the resulting numbers against your forecast, then keep going.
To reiterate, this is all basic stuff, but is really a different way of thinking about
and assessing the impact of your features. It's a more rational and more scientific way
to measure the impact of signup and conversion rates, so that you can spend your time
working on projects that have high impact with low cost. Combine this with a high
iteration speed where your team can push out many of these experiments every week,
and you're well on your way to 10Xing your growth rate.
Oftentimes, you'll find that only 2-3 A/B tests out of every 10 will perform better
than your baseline case. Given that, you'd have to run at least a half dozen per week to
see a positive result on a regular basis. If you only run 1 or 2 A/B tests, it's easy to get
discouraged early.
5. A confirmation email is sent to
remind the user about the length
of their trial and how to contact a
representative if they're
interested in a paid plan. Trigger
emails will also be sent closer to
the end of the trial date to
encourage them to purchase a
plan an important step in
converting trial users into paying
customers.


18
Other random notes on things to try
Here are some other popular techniques that you can try to improve your
completion rates and lower your customer acquisition costs:

1. While a shorter signup flow can be a big help in raising conversion numbers, it's
not a cure-all solution. Some
pages can become too cramped and are new users
may find the discouraging. As you saw in the blogging example, 2 pages at 30%
each will yield a 9% conversion rate at the end so if you can combine those 2
pages into one with a conversion rate greater than 9% do it!
2. Any 'high value' pages should be upfront as more people are present at the
beginning of your signup flow than the end (e.g. an 'invite friends' page). If you
have multiple goals in your signup flow prioritize, and move the important ones
nearest to the beginning. The same concept applies for high conversion rate
pages. If you can get more people through the beginning steps, there'll be a
higher chance that more of your signup goals will be completed.
3. Try different registration types (e.g. sign up with Facebook or Twitter). The
success of these will depend on your audience, and could yield positive or
negative results.
4. 'Lazy registration' is a technique used by some websites (e.g.
http://www.geni.com, http://www.doodle.com) where the user experiences the
product during the signup process. Signing up for the site is an expected part of
using the website itself. This type of signup is very easy for the new user to
complete, but can only work with certain types of websites.
5. With ecommerce sites, the costs associated with acquiring a customer are not
necessarily your only concern, and you are more concerned about your overall
ROI. This means you must consider up-sell and cross-sell techniques to push the
customer to purchase more items.
6. Bundling is a common tactic with SaaS products. You can increase the amount
that your customers spend at one time by offering them a discount when they

19
spend more at one time. If you set the price high enough, you'll potentially be
able to cover your CAC with one purchase, versus trying to bring the customer
back to your website multiple times to pay for separate chunks.
7. Many SaaS products have begun offering a free trial, which allows you to obtain
the user's contact information to follow up with them if they don't initially make
the purchase.
8. For any type of website, the payment step is often high churn. This is a good
page to experiment with to try to optimize for the highest possible CTR.
9. It is always important to optimize your ads. There's no point in paying for clicks
that will never convert into users. Great ads are key even a perfectly designed
signup flow can't convert people who are uninterested.
10. There are many online resources about optimizing landing pages. Read up on
best practices (see our Resources section later for suggestions), and A/B test
them until you come up with a winner.

Remember that overall, none of these specific techniques are guaranteed to
improve your signup flow. However, with a growth spreadsheet, you can now start to
model changes to your signup flow and simulate what might happen, helping you to
make data-driven decisions.



20
Conclusions and next steps
This eBook primarily focused on the key step of signing users up, but of course
that's merely one step out of many. The more important lesson is the process of
mapping your product's UX into a spreadsheet, simulating and prioritizing product
changes, and then running the experiment to validate your hypotheses.
You can take this rational approach to many other pivotal moments in the lifetime
of your product as well, including:

1. Activating users
2. Increasing usage and activity
3. Retaining and re-engaging users
4. Getting users to invite their friends
5. Converting users from free to paid
6. and much, much more

Each one of these moments is just as important as a signup, and each one can be
broken down into a step-by-step funnel and optimized. The same techniques as you can
use for optimizing signups can also be used in these other activities just build your
model, put in baseline data, try out a bunch of experiments, then measure again to see
if you've improved things.
On a final note, it's important not to go overboard on optimization. New products
are about providing deeply engaging experiences, and optimizations can help you bring
that to more people and make sure a larger percentage of new users have positive
experiences. Optimization won't help you fix something that's broken, or get you out of
a market that's too small. Consider this just another tool in your tool belt, to be used
once a product is working well and just needs more growth.