The importance of messaging (SMS, push notifications, WhatsApp, etc.) on user engagement

25% of mobile apps are only opened once after download

If you run a mobile app, you already know it doesn’t matter if you have the best features, or the cleanest navigation, or the most powerful search, or even the highest-quality offerings. Nothing matters if you can’t get your users to come to the app in the first place.

This is why messaging that lives outside of your app (e.g. SMS, push notifications, email, and WhatsApp) are so important.

According to invesp, sending mobile app push notifications can boost app engagement up to 88%, and 65% of users return to an app within 30 days when push is enabled. Sending  push notifications can increase app retention rates by 3-10x whereas users who opted in to push notifications are retained at nearly 2x the rate compared to users who have not opted for push notifications.

That said, the way most CRM and marketing teams approach messaging is incredibly inefficient (and, to be fair, most "modern" customer engagement tools are to blame).

Are the current segmentation "best practices" really best?

Most messaging follows what we might call the Standard Advertising Design of user engagement:

Under the SAD model, you divide users into segments, create a campaign that sends a message to each of the users in a segment, and in the end, you bring everything back together and compare campaign performance to see how each segment is doing. 

There's no cross-channel learning (e.g. SMS to Push) or cross-segment learning, and after our experiment is complete, we have to come up with a hypothesis for the next test, and none of this scales. The more segments you build, the more messages you have to write, and the more work you have to put in (with typically diminishing returns).

This is super broken. Let me illustrate:

Customer relationship management is not advertising

In advertising, you’re playing a numbers game: get your message in front of lots of eyeballs, get some low percentage of those impressions to convert, and if your message doesn’t resonate with the majority of users, well, messages are relatively cheap, so no harm done, right?

Wrong. User engagement isn’t advertising. You know these users, and they know you. You’re not forming a first impression - you’re managing a relationship (there's a reason it's called "Customer Relationship Management"). It may be unfair, but users see stale, generic, irrelevant messages as the hallmark of a stale, generic, irrelevant product, and that’s not the sort of relationship they want. Messages that miss the mark carry a cost.

The problem with the current approach to market segmentation

Segments are the problem here, as an effective messaging segment requires you to know the answers to several pretty major questions in advance: 

  1. How do you know who to put in which segment? 
  2. How do you know what the users in each segment want to hear about?
  3. How do you know when they want to hear about it?
  4. How do you know how they want to hear - what actual words to use?
  5. And the big one: once you see results, how do you figure out what to do next? Because I promise you, you can’t send the same message that worked last time. It’s just a little too obvious.

Very few people can confidently answer all of these questions ahead of time, and those who are confident in their answers shouldn’t be: you can easily Google loads of opinions about “best practices” for setting up segments, and 99% of that advice is based on assumption rather than evidence.

How do I know that?

Because setting up segmented campaigns to test possible answers to each of those questions about who, what, when ,how, and what next — that would take forever.

In reality, the people testing best practices are either just spouting an opinion with no evidence, or they’ve tested a small handful of the full range of possibilities and assumed there’s nothing more to see. 

How to use customer messaging (SMS, push notifications, WhatsApp, etc.) to build your segments for you

There’s a better way. We like to call this the Guess-Listen-Adapt Design (GLAD) model for user engagement:

Here’s how it works:

  1. Write a whole bunch of messages. Don’t just write the message that you think is going to work the largest number of users. Write messages that you think will work for a small niche. Write messages about all your major product lines. Think about all the different reasons someone would choose to invest in your product and write a message for each of those reasons. Write. All. The. Messages. and write them for all of your messaging channels.
  2. Send those messages to all kinds of users. Don’t choose which users get which messages ahead of time, because you don’t know who should get what. This is the Guess art of the design. Randomly send out different messages to different users. (If you're worried that some people might get messages that are irrelevant to them, don't worry...they already are.)
  3. Now, instead of generating results at the end, we generate them in the middle. This is the Learn part of the Design. Compare the performance of different messages to different individual users. (Did they click the message? If they did, what did they do next? Did they look at what you sent them or did they go look at something else?)
  4. Allocate users to segments based on what you’ve learned. If a user responded to a message that highlighted a certain value proposition, put them in the segment of people who respond to that value proposition. Form your segments based on the business categories you actually care about. (If you're catching on, we're actually building behavioral segments here, which is much more effective than more surface-level types of segmentation like demographic segmentation).
  5. Use those segments to choose the next messages. If you offer multiple product lines, and you have a segment of users who seem to really like one of those lines, send them another message about that - you still have all those messages you wrote at the very beginning, so it’s just a matter of picking something that fits and trying it. 

The Guess-Learn-Adapt Design model automatically crafts your segments to reflect your business priorities. You can loop through the process again and again and again, each time getting a clearer, better picture of what each user prefers, and sending them messages that fit those preferences.

So, how does this look in practice?

Unlike the major customer engagement platforms like Braze, Aampe doesn't' have a user journey builder (crazy, right?).

Instead, we spent our time building a tool that lets you quickly build a large number of messages like a puzzle with different pieces that appeal to different user motivations:

We then send these messages out following our GLAD model, and every time a user responds (or doesn't respond) to a message, we learn several things about that user:

  • Was this a good time to message that user?
  • Is the user interested in the product or offering we messaged about?
  • Do they resonate with the tone of the message?
  • Did the message address their core need that your product or service fulfills?

...and then, depending on the response, our model automatically adjusts each individual user's propensities for each of those inputs, creating a unique user journey for each user that aligns with your ultimate goal events (e.g. subscription, conversion, engagement, etc.)

So, don't spend your time building segments

You don't have to. Your customers can do it for you by letting their actions speak for themselves (and they'll be much more accurate).

If you want to learn more about how we do it, please reach out.