CaddieOpen in CaddieDownload for Mac

Mobile App Engagement and Retention

Shared by Caddie Library · Updated
App Masters14:38Open on YouTube
Chapters show their main points

The bottom line

Retention and engagement drive app growth, and machine learning on detailed behavior data can personalize and re-engage each user. Without them, churn shrinks your real addressable market far below what acquisition promises.

Key takeaways

  • Retention, not acquisition, sets how big your real market is. 4:15
    • The model assumes a 100 million addressable market, a 10% conversion rate and about five shares per user.
    • At 80% retention, that market settles near 43 million users, and it keeps shrinking if you stop spending on acquisition.
  • Growth has four stages, and most teams only work on the first. 6:56
    • The stages are acquisition, onboarding and activation, engagement, and retention with loyalty and referrals.
    • About 90% of the talk at conferences is on acquisition. Casual game makers keep launching new games because they lose users in the old ones.
  • Tracking user actions and attributes shows what to build next. 8:40
    • Instrumenting too many events without detailed attributes hides how users really behave.
    • Tying behavior to a user ID, plus web and mobile traits and CRM data, lets you personalize down to the individual.
  • Machine learning can predict churn and trigger campaigns automatically. 11:55
    • The model can flag a user who is likely to leave in about three weeks, then fire a coupon or a game offer.
    • Combining Android and iOS data finds the segments, and a one-line capture means no engineers are needed to instrument.
  • Within three or four years, every user will see a different app. 13:48
    • Screens, content and offers will differ per person, and each segment will level up at its own pace.

You're not really eliminating the data scientists, but what you're doing is you're applying them to understand the business rules much more than the technical rules.

Scott Ritchie, Pyze

Chapter summaries

0:00 Every user will get a different app

Personalization by segment, driven by machine learning, is where apps are heading, and small tricks can still move acquisition.

  • Each person will see different content in the same app. 0:52
    • Everyone in the room has the CNN app, but each person would open it and get different content.
    • Offers differ by segment, so some segments level up faster than others.
    • Games change for each player too, not only news feeds.
    • Slower segments get more discounts to push them up a level.
  • A small growth hack can push an app to number one. 1:55
    • Pyze's co-founder built an app that reached number one in its vertical.
    • The data came from SDKs worldwide: manufacturer, usage, and which screens people view.
    • Behavioral data like this can be bought and used to drive acquisition.

2:58 Retention decides how much of the market you keep

Acquisition fills the market, but low retention drains it again once the spending stops.

  • Growth by sharing climbs toward the market ceiling, then falls back. 3:05
    • It starts with 100,000 organic or acquired users, and every one of them is asked to share.
    • The curve passes 50,000, then 80,000, and approaches the carrying capacity.
    • Past the peak, with no new acquisition spend, the market shrinks to about 17 million.
    • Roughly 36 months after the peak, the app is close to empty.
  • Only casinos escape the churn, because real money keeps players. 6:13
    • Casino apps have huge retention because it is gambling.
    • In casual gaming, users leave the core game for other apps.

7:19 Analytics can carry engagement and retention

Analytics with machine learning can cover engagement and retention without leaning on product development and creativity.

  • Acquisition costs money for every user, so keeping users needs its own system. 7:29
    • You pay for each new user, and analytics is what drives the rest.
    • Machine learning takes on engagement and retention work that creative teams would otherwise carry.
    • The point is that product teams don't have to guess what to build to hold users.
    • Ritchie says this is his belief, not a proven rule.
  • A checklist runs from instrumentation to automated campaigns. 7:54
    • Instrumentation, usage analytics, attribution, segmentation, personalization, engagement triggers, retargeting triggers.
    • Segmentation asks which groups of users behave alike.
    • Personalization covers screens, content and future gameplay.

8:40 Track user actions and their attributes

Detailed action data shows what users already do, so features can follow real behavior.

  • Measure what users do, not which features exist. 8:43
    • Look at clicks and decision points inside screens and content.
    • Build new features that match how the current base behaves, not something brand new.
    • Start from the big picture: where the game should go and which metric a feature should move.
    • Analyze usage rates for that feature and apply the findings from that subset.
  • Attributes on events steer product direction in real time. 9:42
    • A gaming client defined very detailed attributes for how users engage with each event.
    • New events can then be attributed to a segment or cohort of users.
    • Traits should follow a login from web to mobile and back.

12:25 One tool instead of a scattered stack

Today's analytics tools are split across teams, and a single system could replace the patchwork.

  • Large organizations spread analytics, segmentation, recommendations and messaging across separate tools. 12:30
    • Product managers, data scientists and marketing analysts each use different tools.
    • Coupons.com is the example of a big organization built this way.
    • Recommendation engines sit in one place and messaging campaigns somewhere else.
  • Better engagement means more purchases, such as in-app coins. 14:06
    • Ritchie ties engagement directly to in-app purchase revenue.
    • He also ties it to the value of an exit, when selling the company.

In the growth library

Scott Ritchie, Head of Business Development at Pyze. Acquisition gets most of the attention, but retention decides how much of your addressable market you actually keep, so build analytics, segmentation and automated engagement around real user behavior. The talk sizes the effect with retention models and gives a checklist for personalization and churn prevention.

More from App Masters · The whole library, and ask it anything

Transcript301 lines
0:00

coming up is a presentation from our app masters connect event in Santa Monica where Scott Richey head of business development at PI's will share why you're addressable market is a lot smaller than you think the customer value growth checklist and finally what actually moves the needle for your app business

0:38

before I get in the so what I want to talk about is after tension and engagement but before we get into that what and I'll post regular bed law applies what we see is the future is if you've got CNN out and I haven't seen a nap and everybody has that CNN app this room you're going to open that half and you're getting a different contest as an individual you're going to get to prove communication as an individual if you play a game that game is going to change right so the way one segment levels up based on their engagement look at versus the other segments that levels up it's going to be different he offers me to be different to get them to level some will level up faster develop shorter you're going to give a lot of discounts and we are trying to use machine learning to do all this yeah so you're not really eliminating the data scientists but what you're doing is you're applying them to

1:39

understand the business rules much more than the technical rules of data sets languages modeling right so that's the ultimate goal spice my co-founder also set up a high seat he created the sinkhole that will fit in okay we're talking a little bit of a growth hacking here this was the number one downloaded momently so what he did was he fighting this number you know what datasets need to be dry all behavioral aspects of who used this app right what data says we captured from SDK around the world manufacturer behavior usage what screens are on what scenes there playing and you can buy all that behavioral data and you sort of use this app to drive it but in terms of what made this number one anybody hit yes exactly bad drove me to number one for its vertical so little things can drive it to user acquisition to get to number one but we're here to talk about user retention and engagement you have a

3:05

hundred thousand organic users or acquired users right you typically have five so Facebook's number when we were working with Facebook they're like forty two people with spot you have to get five people to share right if you did do five you get everything in your power to get it to five five years which you're an addressable market of one hundred million users right so that was a target that these folks had in terms of how you share have you have a typical conversion rate of 10% right these are typical numbers on when you look at an addressable market what should you be after right but today I think you know going to conferences and speaking with Steve and thousands of app developers from companies like ah get it down to like Indies so many people are featured like every conference do not even center focus on acquisition the money for acquisition acquisition technologies but let's say you did keep through your high

4:15

musician you you look at addressable market about hundred million users based on this going on and over time if you started my wand this is how you would grow to that of Escobar right based on that 10% you know you're looking at 50,000 and then on top of that 80,000 you get to your hundred carrying capacity as I recall it so if you have a 90 percent retention rate on that hundred million addressable market this is what you can expect and sort of look 19 you're going to hit your capacity or your decimal market a 58 knowing based on those retention dollars right which is a 90 percent off those numbers that you saw earlier and 36 much you're gonna be below 10 then when again if you didn't throw any more money that user acquisition or didn't grow again right so 90 percent retention here's eighty percent retention you're looking at about 43 million users as the addressable market so you're still

5:23

throwing money out acquisition right but what is how many actually downloaded and use what's anybody want to take a guess some of these numbers it's 30% so what that means you really tension anyway how many actually stop you can get back to the person so that's what you're really tension how do your customer markets after the peak without going more money at your acquisition pause and everything else we're trying to do through growth market so essentially now you look at your dress will market you look at 17 million adjustable market you know looking at 98 minute you started doing user acquisition whatever growth whatever it is basically as an app developer your banker by 36 months right after doing whatever acquisition techniques and sharing techniques so that's why so many these gaming apps making another version making that version except for maybe the casino industry they have huge retention right

6:37

this is money it's gambling there's a reason why it works the bike sister but in casual gaming that core game meaning a lot of the other usability apps put it through the apps you can end up losing users so they create more and more apps to go through this file right so if you look at the growth cycle these are the major four right you have acquisition that's what ninety percent talk about an acquisition you got your onboarding recorder activation first purchases whatever it is for shares your PB either on your engagement which is sustained usage of conversions personalizing experiences this is something I'll dress loyalty referrals attrition risks reengagement so that's root growth so much business book there's a Mac side and on morning without on the side so you pay for usury leisure user so what drives on this analytics right so analytics and we believe I believe machine learning can cover the engagement retention pieces

7:44

without suffering from product development and creativity so if you look at that checklist what is it right so you have instrumentation of your architecture and I'll go through one of these case examples because there's I don't have enough time to go through all this usage analytics and you have attribution essentially analytics segmentation very important what groups what codes will huorchi have what group here then users retailers users you have present personalization of accidents content in the future gameplay engagement triggers retargeting triggers and campaigns if we just go through one of those you may be for the use cases if we're looking at doing the architecture for this it's important to just figure out sort of your metrics first which is track user actions right and not features there's been a couple of presentations today that's that we focused on features but you lost our old

8:54

users right if you have really detailed behavior analytics on how your users are behavior within your content within your app within your screens you can build future features that match that kind of behavior of your existing user base rather than something branding so look for clicks decision points the big picture I think they pointed out earlier the big picture is where do you want to go with this game is there a feature to is a feature 3 what's the metric they're trying to drive to usage rates and look at that and actually analyze that and take the data from that subset to apply it to what you need to do with that by mary game events versus attributes too many people look at you know when they instrument the analytics and so forth they look at the events and they try to instrument too many events without looking that each attribute I think this was pointed out earlier today so we're working with a gaming company for each

9:59

of them to have took the attributes backs define very very detailed in terms of how their users are engaged with those events and allow the attribution data to sort of dry product development and product direction with your current base of users and have that on the real-time basis so then you can add another event attributed to a segment of cohort users that you're engaged with traits are very important so if you have log in to be able to trace traits from web to mobile to mobile web so if you can get all that bundled in as well that's very important to engage your users and if you have CRM data that's also beneficial right so you can get down to personalize down the user I'm curious though if you have your users that are so far what's the benefit of adding these different so it depends on where those users but ultimately you want to be able to capture exactly what they're doing and know who got exactly

11:16

that person is in the app that's doing it as opposed to just usage behavior but if you can vary the usage behavior to an ID then you have an exact idea about how that's done right and particularly if you have machine learning because then you have that ie you have that particular person regardless of where this behavior changes in the events the Machine looking to keep updating that would be so then it would identify what he's doing yeah and you can start triggering automated campaigns to that person through machines right machine will say this person is going to ensure you're going to lose this guy when you bought 4x twice and you're going to lose that person in about a month for three weeks right letting all the campaign and against this guy happen whatever it is keep them involved in the a coupon well the game keep you involved in here what events that they're creating within your ecosystem within the apps and so that's

12:25

one checklist there's a lot of other checklist as I mentioned and we'll go through this so how does that ecosystem look today right so these are sort of the tools that are being used for analytics by you know product managers data scientists then you have marketing analysis bi this is what a big organisation looks like like a coupons.com for example you then you have a marketing close to the product manager your segmentation on these particular users and then you have stuff that has recommendation engines for a personalization but you have a messaging campaign them somewhere else right that's what the ecosystem looks like what we're trying to attempt which we do I take what we have the name where we're going to go but the ability to caption it automatically one-liner throw there's no engineers needed to instrument what you need to capture we have machine learning that those segments of users to

13:24

tell you when you're going to lose your users then you can combine these users across devices so if you have an Android user iOS user if you have five minutes you can combine the data across it to identify those particular segments where you can run machine will anything like auto campaigns off right and in the future everybody is looking to get to you is to drive engagement which is the personalization which is every individual is going to get different different screens different apps different contact within each particular so that's kind of where we see that things will evolve in the next three or four years and obviously engagement means a lot to you this is right it makes a lot of its we're at many purposes purchases for in-app coins and for your exits that you're looking at somebody to present your company that's pretty much all I got so why can we meet you here [Applause]