Mobile app personalization
The benefits of personalized marketing
Jess Overton, Director of Demand for ironSource Aura, was welcomed by AppsFlyer to speak at MAMA Tech Virtual. Here’s the breakdown of his talk and what we learned about engagement and personalization.
How to personalize your apps
It’s always been top of mind for marketers to create targeted user acquisition strategies - but in the context of today’s increasing CPIs and CPAs and declining organic discovery, it’s more critical than ever for marketers to drill down and get as granular as possible. In this piece, we’ll discuss best practices and actionable tips for building a targeted marketing strategy, which we’ve developed over the years optimizing campaigns for some of the biggest app brands in the world, and the customization possibilities available to you across media channels. Let’s dive in.
1. Define your segments and microsegments
The first step in building out a targeted marketing strategy is defining your segments - in other words, finding out who exactly you’re customizing for - university students, young professionals, and retirees are just a few examples.
In our experience, we’ve found that defining eight high-value segments is the ultimate sweet spot. That’s because fewer than eight negatively impacts how personal you can get with the messaging, while more than eight ends up drowning you in unnecessary data.
Likely, you already know a few of your highest value segments. But still, there are always surprises to be found in the data. It’s important not to make assumptions, and to analyze the data on a regular basis. For example, by diving into the data, one of our food delivery app partners in Europe discovered that in addition to their bread and butter segments (university students and young professionals too busy to cook), working mothers were in fact one of the highest LTV users, despite having smaller order sizes compared to other segments.
With these segments in hand, the next stage is mapping them against the user journey, in order to create microsegments. From brand discovery and initial consideration, through to app download and eventually conversion, you need to know at which points in the user journey you’re gaining users and losing users, where marketing is effective, and where product is effective. The key is identifying the highest value microsegments to focus on. That food delivery app, for example, combined the segment ‘working mothers’ with the user journey ‘app download’, giving them a microsegment of ‘working mothers who have already installed the app’ to target in a nurturing or retention campaign.
2. Planning the targeting strategy
Now, how do you get in front of these microsegments? There are not many - if any - user acquisition platforms in which you can input ‘working mothers’ and immediately be able to engage them. Only by distilling the user profiles you’ve just built into granular targeting parameters will you be able to locate them. So what parameters do you need to input to market towards ‘working mothers’?
Depending on the media channel, the parameters will differ. On OEM and carrier channels, you could use device model, geographic region, age, gender, apps already installed, and more. For example, to locate retired men, you might use specific zip codes, look at high-end $1000 devices, and cap the age at 55+.
Your targeting strategy boils down to one main aim: you don’t want to be paying for irrelevant traffic outside of your microsegments, and getting smart and identifying a good targeting strategy is imperative for customization.
3. Planning the creative strategy
Now that you know you’re speaking to the users you want, the next step is to start speaking their language. To resonate with your prospective users, you need to be able to offer value, which boils down to messaging that is relevant and timely. Creating those initial messages is only the first step - and A/B testing each creatives against your matched micro-segment is the best way to make sure you’re going with the strongest copy and visuals.
Let’s take a look at a different app - a news app. Last year, during the midterm elections, one of our news app partners was looking to promote their election analyses. Together, we identified their micro-segments and parameters and ran A/B tests according to state, building creatives that included the state name and imagery of leading candidates in that state. We then ran those creatives against a control, which was relatively bland and not customized in any way.
After some time, we saw that the uplift in the CTRs for the customized creatives were about 30%, and the improvements held throughout the entire funnel. Essentially, users who viewed the customized creatives were not only more likely to click and install, but also more likely to stick around and become loyal, retained users. In general, we’ve found that it takes 4 or 5 iterations on creative and copy to get the best results - anything beyond that brings minimal lift.
4. Planning the pricing strategy
The last element of a customized marketing strategy comes down to budget and pricing. Often, marketers fall into the trap of paying the same amount for all users, no matter their segment. However, paying different prices for users from different segments is key to reducing cost and ensuring that you’re not overpaying for irrelevant segments, and underpaying, or failing to compete, for your most valued ones.
To determine how much to pay for each segment, it’s best to start with the same bid for all of them. After 7, 14, or 30 days, analyze the data to see where the LTVs are lowest and reallocate spend from those segments to your high value ones. This will allow you to generate quality at scale, ensuring you’re paying the right price for the right users.
Let’s take a look at how this works in practice. Take a look at the chart below, from an eCommerce app partner in Southeast Asia. We started with a single bid and ran with it for 30 days under a ROAS goal of 200%. After those 30 days, we discovered that the results varied pretty widely among countries - while we were underpaying in India, we were overpaying in Malaysia and Indonesia, as shown in the chart on the left. But after tweaking our bidding strategy and changing each bid per country, each micro-segment in each geo met the 200% ROAS goal by day 30, as you can see in the chart on the right - perfectly hitting the original KPI and achieving peak efficiency.
By paying the right price for the right users, the eCommerce app ensured they were competing in the microsegments that were most valuable to them, and letting go where it mattered less.
As customization becomes the norm, we’re likely to see hyper-customization become more popular - essentially using multiple datasets across platforms to build a 360 degree user profile, perhaps even combining virtual and brick and mortar in one. Perhaps, for example, we could find ourselves shopping in-person at Zara, and unable to find our size, find it waiting for us in our cart online when we get home. Of course, building this sort of omnichannel marketing strategy will require cross team and cross platform collaboration, but with the way the future is going, it might not be too far off.