Lele Canfora is head of user growth at Lovoo and he spoke at App Promotion Summit Berlin 2014 about the topic of app retargeting. Drawing on his experience of keeping Lovoo’s social discovery platform as one of the most successful around, he explained the best ways of bringing a user back to an app they might have previously abandoned.
In his talk, he spoke about:
What exactly app retargeting is and why it matters.
How Lovoo approached the problem from a technical perspective.
Advice from their case study about how to make the most of it.
You can read the full transcript of Lele’s talk, watch the video or listen to the audio below.
Developing a Retargeting Ad Campaign to Drive App EngagementVideo:
Developing a Retargeting Ad Campaign to Drive App Engagement Audio:
Developing a Retargeting Ad Campaign to Drive App Engagement Transcript:
I’ve been here last year at the App Promotion Summit, it was a very good experience. I was talking about Lovoo Apps. When we started last year we were already pretty big in Germany, in Austria and Switzerland, and we were exactly starting expanding in Europe and France, in Italy and then UK and then Brazil as well. It was a very good experience at the App Promotion Summit and we were starting in October last year, two-year acquisition. I really want to talk about that last year because it excited me to talk about something that we were studying. We started to get some results but we really just had the beginning of it. This year is exactly the same topic. I will run you through exactly our experiences in re-targeting for our user base.
This is our app. I will talk about mobile re-targeting, how to drive app engagement and reduced their churn rate. I’ll go quickly through Lovoo as an App. Lovoo is a social discovery network, is a social network where you can definitely meet new people. It’s an iPhone and Android app and is as a website as well. It’s currently the strongest, largest social discovery network in the Austria, Switzerland and Germany region and we’re very, very strong also in Brazil, second at the moment, Spain, France, Germany, France, UK and Italy. We have a pretty young year of user base, 90% definitely on the app and 10% on the website. Good split between women and men, has again very young user base, very active.
I’ll quickly explain about social discovery networks because some people know more or lesser the concept, but what’s most important to remind is that these apps and these websites are highly engaging, they’re usually highly monetized. You will always have inner purchases. They’re used more than once a day. They’re very engaging [inaudible 00:02:11] other times and then they’re use and pair it or with other Apps. As you can see this is something not everybody noticed. Social discovery networks are one of the hidden champions of the App Store. This is the Germany iPhone and Android top grossing, and as you can see we both are on those platforms in the top ten. As I discussed last year, I announced that we had reached 6 million registered users. It was very exciting and, as again, now I can definitely approve we are 219 million users.
We’ve been growing and growing, growing and acquiring user all over the place, mostly definitely organics and then help by a lot of very well-thought data-backed marketing. We’ve done this, we’ve grown extremely well. I’m very proud of it. The team did a very, very good job. Once you reach this amount of users, once you reach a very large presence in a country, it happens something like this, we have a large active user base, and then in the course of a year, of course, some of these user churned. Some of these user became inactive. This is a very, very classic pie. We have active users, some inactive users for a longer period of time, users that churned in the previous month and then the new user that you’re acquiring. This is after, of course, month to month to month and the most you grow, you keep on acquiring users from getting organic users. Probably the inactive pie will get bigger. It’s just a numeric thing.
What do we do at this point? We thought there is probably something that we need to do. This is a nice image because it conveys exactly like a metaphor that it’s used a lot during conferences and it conveys exactly what I mean. You’ve got a bucket and this bucket come in lot of water. There is some also in the bucket. You lose some of this water. I am a marketing person. I’m doing marketing analytics now, but of course my biggest advice is always go back to product and fix the patches in the hole. The most important thing is that you retain the user themselves. You stop the churn before or during marketing, but us, again, we’re doing marketing so we cannot just sit there and wait for developers to create some very good new features that will retain users, which, by the way, we will have a very good release next month.
What do we think? We start analyzing patterns of usage inside the app. We understood that, as usually happens with social discovery networks, our app has a very intensive usage for a period of time and then user become inactive. Some of these come back for another period then they become inactive again, and of the small amount will then ever, let’s say, a third life cycle. What we definitely wanted to reduce, what we definitely wanted to eliminate, the users, of course, never came back after a certain period and make these inactivity periods as small as possible, or actually eliminate them. So we thought, “Let’s do mobile re-marketing.” I’ve heard a speech from Simon before me and from Eric as well that was definitely related to the topic. One was about deep linking, nice introduction, and one was about programmatic buying which has absolutely to do with re-marketing.
We thought, “Okay, what are our options?” We have an internal system, of course, of retaining user with pusher notifications. We’re doing that brilliantly. We send notification at the right moment. We have social networks and when you have a real interaction we send you a notification. This works really well because it’s a real thing that you want to check out. Let’s say the users have turned them off. Let’s say the users have become inactive. What are our other options in the market? We’ve had several different channels that we can actually go there and tell our previous users, “Come back to the app. There’s something interesting to see.” Social networks, Twitter, Facebook, definitely, Ad networks, Ad words, RTB, so Real Time Bidding and then we have the option to send emails to those users.
Our goals in our re-marketing strategy were to stimulate the recent inactives that became inactive let’s say seven days to 14 days, reduce the overall churn rates of our app, reduce the size, as I told of the inactivity period, increase the amount of interaction, and then, of course, in the end create more revenues from the returned users. We, at Lovoo, have a lot of information about our users. We’re a social networks, so they usually tell us their location, their sex, some of course their preferences they create in app purchases. What did we decide? We decided, “Okay, let’s see, we have geo, we have gender, of course we know which platform they’re active on, we know if they’ve done purchases or not, and we know exactly when did they become inactive.”
Let’s say we created several audiences, 85 I believe, and we say, “Okay, we have all these audiences, we know exactly what these audiences are worth it. So let’s go there and reactivate them.” And then of course came a lot of issues. Re-marketing is not easy, I will tell you. There is a lot to be done, lots need to be done in the market, and there is a lot of work behind it. When you start analyzing your audience you will see will see you have several different IDs. You have mobile IDs that actually can be used in Ad networks and social networks, and real time bidding, you have STK, or automated generated audiences that will be flashed in from your tracking partner, from your analytics partner through to the app networks. We have social networks ID in case you have, let’s say, Facebook connect, Twitter connect, Google+ connect and then you have email addresses that you can fill in.
And these vast amount of IDs is to be mixed and matched in the right channel, and some of these channels realize that both, some of these channel will let you use all of them together, some of these channel not, and you will have to understand that. Of course maybe an automatic created audience is the fastest, is the most clean, but sometimes you’re not sure what’s going on in there. It’s a black box, so you prefer to extract your data, send it to the user and exactly re-target these ones. And then we thought, “Okay, definitely going to do social networks, we’re going to do real time bidding and we’re going to do Ad networks.” What happens in this case? We have very, very large number of users in the millions. We extract these IDs from these users, 90% are find-able. Let’s say some of these users opted out of mobile IDS. Some of these users have, let’s say, emails that are not valid any more, some of these users have telephones that are dead.
So you have inactive users, some of them you find IDs, then you send it to the networks or to the real time bidding and they match it. This is a black box. You don’t know what’s going on in there. You send it and you hope that they get matched. Then you have to, of course, start bidding on these users and you have to hope that user sees that. The users have to click on the ad, and then you opt that the user does not de-install the app, so it goes directly through a deep link, and it’s a much shorter funnel or the user actually de-install the app, so you have to actually convince them to install the app again. This is an additional step and is much more complicated. What are the limitations? It’s beautiful if you have a lot of audiences. You can make it as small as possible and you can really say, “Okay, I want to spend so much on this cohort and I want to spend so much on this other cohort,” but there is the huge limitation that your audience must be large enough for the networks.
This is true, of course, also in user acquisition when you sell like some parameters and it’s extremely true on re-targeting. Anything below 300,000 is too small. Then there’s the problem of user inactivity. Let’s say you have a lot of users accumulated in years of activity, probably users that are one, two, three months old will not convert. The users that are really worth it are the ones that stay de-lapsed seven days, 14 days, maximum one month. There are some issues in the market. Tracking is complicated. There are a lot of partners that have different metrics for actually what is that reactivation. Each channel that you’re going to use has a different way of actually calculating this deactivation. Sometimes these networks are not that good. They’re not that re-marketing ready, and they will just not optimize in this value.
My advice is that you abstract a little bit some times on the automatic optimization of this network, and you just, let’s say, do it your own. So you know exactly the user that click on a certain ad then you’ll optimize your own on this click. It’s, of course, not the best way, we’d all loved to have automatic optimization, but that’s sometimes at the moment not possible. Again, what is the most effective audience? Men, women, purchase, no purchase, it’s definitely inactivity. After I’ve been talking to a lot of DSPs for real time bidding and that’s what all of them told me. The most important thing is that you go after, as fast as possible, to the seven-days old, seven days inactive users, you spend as much as possible on them because they’re easiest to reactivate. Then the 14 days are still okay. They might have not de-installed the app, and they usually don’t. One month, they start to get old. They might have really lost the interest in it. Three months and more are probably lost.
What is the best way to work on a bidding system? It’s definitely to know how much money you can spend on your users. Don’t just collect a bunch of IDs, just send it and start reactivating. Just know exactly how much you can spend on these ones. Know exactly, let’s say, in the limit of the minimum accessible amount of users, I want to spend this amount on these users because they were X way, or they spend a lot of money. I will spend a little bit less on the casuals, so let’s say a 10, 20 cents will be enough. They are our star users that have been very active for a month and then now they are inactive for seven days, lets really go after them. Geo is very important as well as in user acquisition, and in our case, gender, because we are a social network. It’s one of the most important variables.
Good link, two speeches before me. When you have time constraint, let’s say I want to do re-marketing but there’s a lot of technicalities into it. Should I go with dynamic creative or should I go with a lot of deep links? Which one is the one that works best? Again, from my experience, as we’re talking to a lot of partners, that it’s not really for a lot of categories, a proof that dynamic creative works. It’s enough probably that if you know your target, if you know what they’ve done and then you create a set of creative that is a catch-all in this specific segment. What’s much, much more important is that you create the right deep link. First of all, because you really don’t want to waste time on the reactivation to get the user to your target, and second because the user that did not de-install the app and you can send it to a deep link is much more valued that the user that de-installed the app.
Always prefer custom deep link, several deep link, very elaborated sophisticated deep links. Now to the key metrics that we use. As again, we started recently doing re-marketing and this is just to tell you what our issues were, where we encountered during the journey and what my advises are. For us, what’s very important was, first of all, successful come-back. We want the user to come back to the app. Again if he still has the app, successful deep link with an app open, if he de-installed the app we want the user to install the app again. Then again we want to look at what is the post-reactivation retention rate of these users. Was he active before, is he active more or less than before? Because don’t forget that you spend money on bringing this user back. And then again, which is the average revenue? In our case, it’s a 30 days average revenue after the reactivation against the before activation. Is it worthy to spend this additional money to bring him back?
Now, for results, it’s not so easy. Re-marketing is definitely still complicated and there are a lot of things that need to be considered. In our case, we discovered that we have a very low uninstall rates in the 7-14 days, and this means it was easy to reactivate those users. We successfully managed to drive part of the churn rates overall, massively targeting the deactivated user. And regarding new revenues, we will see you in the next few months if it’s interesting, if it’s valuable.
The post Developing a Retargeting Ad Campaign to Drive App Engagement appeared first on Business of Apps.