Product Growth Strategy with Experimentation

Product Growth Strategy with Experimentation


When a new product gets launched or an existing product is struggling to grow, we often start employing tactics to improve usage or to bring back users. This article sets an overall framework for growth during those difficult times so that we don’t focus on unsustainable tactics of short term but rather achieve long term continuous growth benefits.

Note that any product growth strategy is futile if there is no market-product fit.

Before you start, you need to have a consistent flow of users that use the product. Then only, you can understand them better, know what works for what kind of users, what do they find exciting and where are they drop-off.

If you don’t have a good user base, then we should concentrate on customer acquisition rather than growing the existing ones.

At the core of product growth strategy, it is about understanding users and helping them complete their tasks efficiently and effectively.

The need for growth

Traditionally, businesses have been focusing on the acquisition to drive growth. When customers had fewer alternatives and businesses had limited reach, getting more customers at the top of the funnel seemed like the best solution.

However, we now live in an intensely competitive environment where customers can switch easily. With the abundance availability of customers on different channels, businesses have started focusing on optimizing the funnel by activating/engaging customers once they are bought in and building moats to reduce churn.

Also, when we calculate the compounding impact of churn on revenue, it becomes extremely clear that the businesses not only need to acquire new customers but also retain and grow the older ones.

In the classic sense, this is focusing on bottom-line growth than the top-line growth, where more income or profits are squeezed from the same revenue or sales.

Such bottom-line growth is fair for products that are the end of their lifecycle. However, as mentioned at the beginning of the article, we are talking about the new products that are struggling to grow.

Formulating Product Growth Strategy Framework

Assuming all your metrics are set up, our product growth strategy framework will involve 3 components:

Component 1: Setting up a target

In the first component, we should

Chart out a Goal – A consensual objective that we want to achieve. It could be a Northstar metric or a KPI. It is important that the metric has to be a leading indicator, nor a lagging one and should align with your vision.

Set up happy paths – Start tracking the journey you want customers to take in the form of a funnel. Needless to say, however will say that this customer path should achieve the user goal as part of our goal.

(more details on charting out the goal)

Component 2: Mining insights

Here, we will need to

Get a better understanding of our customers – Mainly analyzing their behavior around that goal and creating custom segments

Surfacing insights – Figuring out what is working for what kind of customers and what’s stopping them

Experimenting and Engaging these customers – Helping the struggling ones, experimenting with champions, motivating the undecided ones ( learn more )

Measuring the impact of our effort across user types – Finding how different types of customer are responding to our efforts.

Optional, but very important in my opinion to make sure users are delighted with our efforts. We can keep an eye on NPS or ask rating/feedback on customers whom we are experimenting with.

(more details on understanding users and surfacing insights)

Component 3: Rinse and Repeat

Now, we make the appropriate changes in the product, one by one. We keep repeating this until we have achieved the desired level of growth for the remaining type of customers.

That’s all.

These 3 components are the building blocks for your growth framework. The effort and time spend are proportional to the sequencing of these components. Ideally, spend 1-2 weeks for deciding what you want to achieve along with the other work you do, 3-6 weeks on customer insight mining with help from other teams,  and 10-20 weeks product customization as a full-time job.

Jump to the conclusion.

Below are more details about each component.

#. Charting out the Goal

As often said, “What gets measured, gets improved.

We need to set a Goal on what we are trying to improve. This objective should be SMART whose meaning differs from the HR definition – Significant, Measurable, Agreed Upon, Rewarding and Time-based.

Increasing Northstar metric is the ideal goal that one can target, if not then the Goal should be as close as it could impact it.

Some clear examples are – Increasing total messages sent in Whatsapp by X% in Y time; Reducing the time spent by X% in Y time in Gusto, the Payroll processing tool etc.

#. Understanding your customers

We might have a good understanding of our users with profiles and their usage metrics, however for strategizing growth, it is not enough. We need to understand them better, shift from vanity metrics to behavioral metrics; from demographics to psychographics; from their actions to complete journeys.

Some prefer tracking all customers for all of their activities, however this can become overwhelming and waste your time and effort.

A better approach is started by tracking only the user’s that matter and analyzing them at the specific points of their journey which can create an impact.

Later, the scope can be expanded to include more users with their complete journeys when we start seeing diminishing results for the initial set.

For example, Whatsapp would be interested in growing engagement for a set of customers who have installed the app but are rarely using it.

They would want these users to use Whatsapp more, strengthen the network effect and kill the competition in the process.

The choice is between users who have installed the app but have not activated it or the users who are seldom using it. The latter group of users can be worked on later as growing their engagement or numbers will have a lesser impact on the number of DAU (Daily Active Users) i.e. WhatsApp Northstar.

Now to better understand your users, there is a wide range of tools and techniques available – Event tracking, Funnel analysis, CDP or DMP integrations, focused group discussions, Jobs-to-be-done framework and so on.

However, the hardest part is creating user segments based on data on these tools. This is generally a skill that takes long to expertise and the more you do it, the better you become. Data science and AI is nowadays proving out a helping resource to categorize users based on their behavior.

At the end of the day, we want to create segments that reflect on how many ways the product is perceived by our users.

Our customer’s perception is defined by monitoring their behavior and usage and we will redefine this perception to achieve the defined goal.

Some simple examples of segments are:

1) Based on revenue – enterprise, high net worth individuals,

2) Based on life cycle state – newcomers, wanderers, undecided buyers, experienced power users

3) Based on usage – logged in, data exported, created task, shared report

4) Based on the journey – searched, explored products, added to cart

5) Based on Churn – low risk, medium risk and high risk

#. Surfacing Insights

Now that we know different types of customers via segmentation and analysis, it is time to find why they behave in certain ways.

We start with analyzing the best customers, studying their journeys and behavior. We try to find what excites them, why did they traverse the happy paths, what motivates them, what’s their aha moment and what do they love the most.

We get certain clues and these clues are validated to discover insights on what works and what will not work.

Some of the tools that we use to uncover these insights are – surveys/polls, Heatmaps, visitor recordings, RFM analysis, Cohort analysis, trends/research reports.

For examples –

1) We know that iPhone users use the app twice than android users. What’s the reason behind it? Maybe we should look at battery consumption or crash reports.

2) Ecommerce sells are mostly concentrated in certain cities for electronic products. Are customers thinking about service centers while placing the orders?

3) Only the higher educated urban dwellers are engaging with the financial product? Do we need to demonstrate transparency or simplify the language?

4) Users who have used Google Docs are more likely to use Google Drive. Should we make it easy in Google Drive to create docs?

5) Users who have added payment method on Uber and use the app for daily commuting to office have low churn risk. Can we experiment with the newer features with them?

Surfacing insights is the most challenging work of product growth strategy. It takes patience and you never know what you will discover. It is like science – with experience, one can get speculate insights at a high level and can validate at faster speeds, but no one knows who and what discoveries can be made.

Experimenting and Engaging the customers

Once we have insights that we think will help in achieving growth in the product, we can start making changes in the product. However, the insights are validated for the best set of users and we think that they will work for another set. This is where we start experimenting – running our hypothesis to check the impact of our solutions.

We brainstorm for ideas such as how to help the struggling customers, when to motivate the undecided ones or what to communicate to high churn risk customers. There are multiple frameworks on which these ideas can be prioritized like ICE (Impact, Confidence and Ease), RICE (Reach, Impact, Confidence and Effort), Value vs Effort etc.

Along with experiments, we would need to plan for customer engagement at different touchpoints which is a communication experiment with the customers.

To objective is to reach to the right customer, at the right time, through the right channel.

With our segmentation, we have a list of customers whom we want to reach out. For the right time, either we can take the help of data science to know when to communicate or just make educated guesses like – best time to send push notification is after 30 mins when user had not checked out their cart or nurturing in the first few days of the month when they receive their salary for a wealth management app.

Since we cannot bombard users for all the action, the right time corresponds to a most important juncture in their critical journey. This juncture can be discovered with funnel analysis for the path the most successful customers take.  For the channel, some argue that we should be little conservative and only use specific channels for which it was made. For example, using emails for promotion, SMS for alerts and push for notifications. However, the newer breed of marketers believe that customer have become channel-agnostic and since many businesses are already reaching them out in non-traditional ways, we should not limit ourselves.

That means we should be employing all channels (email, app notifications, browser push, SMS, webhooks, web popups and social) for an action we think will be beneficial to the user. 

Now the question arises on how many experiments to run and till when we keep trying to engage users?

The answer lies in the confidence and the return we start seeing. Though we can keep optimizing and run experiments forever, at some point in time, we see diminishing returns and that’s when we will know that we have reached the potential. Many marketers make the mistake of comparing the numbers across the industry, but they forget that each business is different. Amazon will be miles ahead in conversion rate in comparision to other eCommerce websites because it is built as such.

When you start noticing that they are expending more efforts and there is no substantial impact (less than 1% relative) then its time to look for other avenues for growth. 

For engagement, a few aggressive marketers believe that if the user can respond once, then we can keep them in the loop forever. And since the cost has come down drastically, it does not matter much. We can create journeys with loop until the user unsubscribes from us. 


Often few things are straightforward like incentivizing losing customers when they abandon the cart or helping out the newcomers. And since they work well, we fall into this trap of fixing individual instances by such tactics. Hoping this article provides a comprehensive framework to see the forest for the trees.

Photo by Jukan Tateisi on Unsplash

October 17, 2018

Leave a Reply

Your email address will not be published. Required fields are marked *