Product Growth Strategies Using CRM

Planting the seeds of sustainable growth, not just temporary changes

This article covers a portion of a presentation given at the recent Uplift event.

Anyone working with app or web products has probably felt the intense hype around CRM over the past few years. Typically, it's used in the form of push notifications or in-app (in-web) messages to boost in-product engagement for acquired users. Alarmy also used to leverage it to pitch a specific feature to users who hadn't visited it yet, or to raise awareness for newly launched features. However, what made us a bit different from other organizations was that we utilized CRM as a core part of our product growth strategy. Let's briefly dive into how this differs from the conventional way CRM is used, and how leveraging CRM contrasts with standard growth strategies.

#0–1. What is growth?

Unlike a standard development process where you plan a product by pre-determining what to ship (What, How), a growth approach starts with what problem to solve (Why) and then develops a hypothesis (How). Because of this, the goal isn't just deployment itself, but rather solving the actual problem, followed by a validation process post-deployment to ensure the issue was truly resolved. Sometimes a winner is chosen through before-and-after comparisons, but to make a more statistically valid judgment, we run A/B Tests (hereafter referred to as "experiments"). Nowadays, there's a plethora of SaaS solutions designed to make running these experiments a breeze.

Whether it's messaging via CRM or UI/UX changes within the product, we tweak variables across control and experimental groups to find the variant that yields the highest target conversion rate. What's crucial here is verifying whether the change is statistically significant. A statistical significance of 95% means there is a 95% probability that this experimental group will continue outperforming the control group moving forward.

For this growth approach to work properly, two things must happen in tandem: 1) proper variable control, and 2) continuous accumulation of lessons. If variables aren't properly controlled, it becomes impossible to pinpoint exactly why (which variable made) B better than A, preventing you from setting up effective follow-up experiments. Furthermore, even with perfectly controlled variables, if you don't actually run those follow-up experiments, the lessons you acquired previously won't ever see the light of day.

#0–2. The background behind CRM for growth

Sadly, our resources are limited, and we already have more than enough massive hypotheses on our plates. Running experiments to verify every granular variable within even a single major hypothesis is too much for our resources to handle.

This is where CRM swoops in to save the day.

It allows you to relentlessly test the hypotheses you want to validate without engineering dependencies (meaning, zero resource constraints).

That being said, CRM has its own unique pros and cons compared to general product growth, so you need to leverage it by maximizing its strengths while remaining mindful of its weaknesses.

#1. Growth through rapid iterations

Without a doubt, its biggest strength is the absolute lack of engineering resource constraints. Thanks to this, you can inductively accumulate lessons at lightning speed.

If we send message L at M o'clock to a user who hasn't opened the Alarmy app for N days, they will start using their alarms again.

There are three variables in the hypothesis above: (N, M, L).
Don't hesitate—just control the other variables and start hunting for the winning values.

First, we run a Language Market Fit experiment where we control the N and M values and only tweak the L value.

We fired off different messages at 11 PM to users who hadn't visited the app for 7 days, and successfully unearthed a statistically significant winner.

Right on its heels, we followed up by applying the exact same winning message across the board, but this time sending it at different times to those 7-day inactive users. Once again, we managed to dig up a statistically significant winner.

Finally, keeping both the winning message and winning time slot consistent, we ran an experiment using the number of days since churn as our variable. Sure enough, we found a specific day mark that yielded the highest conversion rate for alarm usage compared to the control group.

In the end, without a single shred of engineering support, we ran 10 campaigns and uncovered 5 winners. Thanks to those winning combinations, we were able to bring back an additional 25,000 churned users every single month. By taking full advantage of the zero-development constraint, we flawlessly executed 1) proper variable control and 2) continuous accumulation of lessons, ultimately driving highly meaningful growth.

#2. Growth through messaging

Whether it's a push notification, an in-app message, or an email, CRM is uniquely characterized by altering the user experience through "messaging." Many growth veterans place immense weight on variables like the size, color, structure, and interaction of key UI components, yet tend to downplay the importance of on-screen copy. But reality paints a different picture. The power of a message can significantly impact even highly difficult metrics like user payment conversion. Especially with CRM, where messaging makes up almost all of its leverage, you really need to pour your heart and soul into crafting that copy.

Therefore, before running an experiment sending a message to a specific segment at a specific time, you must conduct a Language Market Fit (LMF) experiment first. Conversion rates can swing wildly depending on the message, naturally shifting the statistical significance along with it.

If your product is a global service, it's most efficient to run the LMF experiment first in the language you're most confident in, and then only translate the chosen winner for subsequent experiments. You can quickly verify whether that winner performs meaningfully in other languages. If it falls flat in a specific language, simply re-translate it for that market and keep the experiment rolling.

#3. Growth while disrupting the user journey

Fundamentally, compared to just leaving the user alone, CRM is an act that disrupts the user journey. Truth be told, the easiest thing to try to boost conversions would be blindly firing off messages to every single user. The reason we all know this but don't do it is because we also know that if we did, there wouldn't be a single user left in the product. This is where proper targeting (segmentation) takes center stage. By narrowing the messaging target down to users who will feel relatively less disrupted in their journey, you can minimize the negatives while maximizing the positives.

The experiment above kept all other variables and experimental group settings identical, changing only the segment. For Ring Alarm O (users who have used an Alarmy alarm at least once), the control group won, while for Ring Alarm X (users who have never used an Alarmy alarm), the messaging of experimental group 4 won. Considering the user base for Ring Alarm O is much larger, if we had run this experiment on all users without proper segmentation, it could easily have ended in a fruitless control group victory. This is a prime example showing that how you slice your segments can dictate whether you uncover a meaningful area for improvement. Conversely, it also implies that depending on the segment, sending a message can actually harm your conversion rate.

When utilizing CRM for growth, you must pay close attention to your segments. Even if your conversion rate goes up in a statistically significant way, you need to check if app retention took a hit. Segments not only impact your conversion rate delta and statistical significance, but they also impact app retention. With the right segmentation strategy, you can secure a meaningful lift in your target metric without dragging down retention.

In short, CRM can absolutely be a powerful tool from a growth strategy perspective. There might be some operational headaches since it requires separate management (and maintenance) outside the main product pipeline, but with just a bit of extra care, you can build up a highly impactful growth engine right within your product.

We still have an overflowing backlog of growth hypotheses to explore. For the rest of Q3, I want to keep creating fun growth case studies using hypotheses that are even more impactful while fully maximizing the strengths of CRM.