So, now you are promoting subscription-based mobile projects. Why this particular niche and not, for example, games or e-commerce?

During my work for AdQuantum, most of the clients I worked with had this monetization model. Gradually, I was engaging more and more in building predictions for subscription apps, and then in user activation too.

A little over a year ago, I was invited to lead the marketing for a brand-new mobile non-gaming publisher. For me, this was a very driving challenge, as the model for this kind of business is not fully clear. Before Union Apps, we had only heard of about 5 similar companies in the market across the globe. After starting at Union Apps, I no longer faced the question ‘why this particular niche?’ :)

From the UA point of view, how much does traffic buying for subscription apps differ from traffic buying for non-subscription ones? What are the features, nuances, and tricks?

I will list the main nuances in UA for subscription products, point by point. P.S. This works for over 80% of subscription-based apps.

• The buying works only for the iOS apps (which implies no satisfactory attribution). Android apps don’t pay off at all;
• Event optimization works only for paywalls. Other types of optimization (install/proxy events) don’t perform;
• Performance UGC creatives take a huge part of the total spend;
• If UA figures add up on the US, then they will add up on other countries;
• ASA is super often overheated for the majority of app categories. You can increase the conversion to install, for instance, with Custom Product Pages;
• If you face a deadlock of spend per application, try out web pages with payment there and transfer users to the app after they make a payment.

What do you normally do when your marketing economics looks good and you want to drive even more traffic? How do you scale wisely and how do you determine that an app is ready for scaling?

When working with subscription-based applications, it is crucial to remember that money is spread over time. Many users take a weekly/monthly subscription, and some of them are not converted into paying users on the 1st day of installation (it’s typical for freemium apps). When you scale your app actively, you may eventually face a cash gap. Judging by predictions, you drive traffic with a positive ROI. However, in reality, you can no longer afford to run campaigns or pay salaries to your team.

To prevent this from happening, you have to build a model for the arrival of money. Internally, we call it Payment Profile. This is a template fitting any app that shows how much money will come in a certain month. With this model, you can build a cash flow model and figure out how much money you need to have for operating so as not to get into a cash gap. Here’s the link to it. Hopefully, it will be useful to someone.

The second important thing to understand is that you will have positive results of 15% or 30% from the Apple commission. If it’s 15%, then calculate your economics at 30% of the commission. It may turn out that you are not able to drive traffic positively, as Apple eats up all your margins by increasing the commission. If this is the case, I would advise not to scale the traffic, but keep improving your product so that you can make UA profitable even with a commission of 30%.

Thirdly, use factoring if your model allows it. Basically, you get fast financing, though you have to pay 20% or more per annum for it.

At Union Apps, we do a scaling when UA delivers 30% predictive ROI on Apple’s 30% commission. Such an ROI margin allows us to start scaling and be sure that after $1M in revenue per calendar year (when 30% commission comes) we will not have to stop our campaigns, but will keep on scaling.

If we talk about mobile ad platforms, what traffic sources are most in demand now, and which, in your opinion, are underestimated?

Historically, Facebook and Google have performed well. For many apps, these two sources account for 80% of all paid traffic.

ASA comes after them. However, as I mentioned earlier, we see that the auction in ASA is extremely overheated. Therefore, it’s difficult to get a lot of traffic from there.

The next traffic source for me is three other social networks — Twitter, TikTok, and Snapchat. I would especially recommend Twitter, as it is now showing an increase in efficiency according to the AppsFlyer Performance Index.

Apple’s policy change hit media source performance hard, so the split is changing dramatically now. The main source for us is still Facebook, followed by Twitter. Then come ASA, Snapchat, and TikTok (we are talking about the split specifically for iOS apps, where we buy iOS 14.5+ audience traffic).

The latest AppsFlyer report shows that our split is in line with the market: Meta is starting to lose ground while Twitter is moving forward.

Why do you think Facebook is no longer the #1 traffic source for non-gaming apps? Is there a chance for it to return to glory?

Facebook has always been an SRN source that uses IDFA for attribution. The lack of IDFA weakened it, so now you have to spend more money to get a paying user. At the same time, Meta remains one of the largest companies whose social networks count over 2 billion people worldwide. Moreover, they have one of the strongest auctions, as Facebook Ads algorithms perfectly understand what ads to show and to whom. It’s just that now this advertising is not as effective as it was a few years ago.

There is only one solution for them in that situation which is to become a 1st party data provider and start making attribution with no additional parties. This will allow them to enrich the data within their social networks. Snapchat released a similar solution about a year ago. I assume this is the only possible way out of this situation.

Let’s then talk about another important subject, mobile analytics. For subscription products, this is a particularly «painful» topic, as the privacy policy and SKAN make it difficult to correctly attribute traffic and evaluate the performance of channels on iOS. And for apps with a trial period, this is a real pain in the neck. How do you cope with this?

The mobile market offers some attribution analytics solutions. These companies allow a cohort look at the RR of subscriptions, refunds, and all payment statuses of subscriptions. We personally partner with Adapty. In addition to subscription analytics, Adapty provides A/B testing services allowing the growth of user activation. There is also other good software on the market, such as RevenueCat, ChartMogul, Apphud, and Qonversion.

Since we don’t have good attribution on iOS, we evaluate mobile traffic in general. With Adapty, we can look at the install funnels —> paywall show (paywall impression) —> trial —> trial_converted —> 2nd payment... Actually, this is how we evaluate traffic by GEO and subscriptions and make decisions.

We transfer payment events from Adapty to AppsFlyer and optimize SKAN campaigns for these events. We evaluate the campaigns by trials, direct subscriptions, and cancellations of trials. In particular, 80% of users cancel trials in the first hour after installing the application, and for us, this (cancellation of trials) is a proxy metric that allows us to estimate the conversion to a paying user from a trial.

How do you and your team understand that an app will become a hit, even if everything isn’t going so smoothly for it at the moment? Do you have some kind of checklist for defining a potential hit? What are you looking at before the launch and during the tests?

Before a development team gets to publishing with Union Apps, we come a long way (about 3-4 months). This is necessary to understand whether we are on the same page and can give each other what both of us need.

The first metric that you can use to measure the «hitness» of a project is LTV. However, not every app has a high LTV, so next, we look at two other things: subscriptions RR and the percentage of paying users.

Subscription renewal RR indicates how successfully an app has found product-market fit, or in other words, if the app solves its users’ problem more efficiently than the competing apps. For me, this metric digitizes the quality of the product itself.

The percentage of paying users indicates how good a developer is at audience monetization. We at Union Apps can help a developer grow this metric. We test multiple hypotheses on user activation. Therefore, if an app has a low conversion to trial rate and at the same time the RR of a 6-month subscription is over 20%, we can sum up that this app might become a hit.

For each subscription duration period, we have an internal RR benchmark, so with a high probability, we can predict whether we can grow an app together with a partner.

Speaking of the initial tests, at that stage we look at CPI and the install funnel —> trial/direct subscription —> trial_converted/payers total.

Besides the fact that an app can have a high RR and a reasonable percentage of user activation, it can also have expensive user acquisition. This mostly happens with niche products that have too few audiences for UA or it can also happen when the app’s unit economics does not add up when you upscale.

It is important for us that the product we work with may be scaled hundreds of thousands of dollars in paid UA. That’s why the CPI and funnel from install to paying user are critical in initial tests. At the same time, we understand that a high CPI during tests may be caused by bad creatives. In this case, we create a new pack of creatives or conduct a second test.

Is there a way to quickly and painlessly increase the number of subscription-based product activations? Or does it always require an individual approach?

We internally collect an A/B tests database that we’ve always conducted for our partners. Over half of the tests work for almost any app, however, we still always have to adjust the test for each product separately.

In my experience for the majority of apps, you can conduct three onboarding/paywall tests which will significantly increase the conversion to trial/subscription. But at least one test is definitely worth trying right away, as, for most applications, a registration form comes before a paywall which reduces users’ reach before choosing a subscription. If they move it after the paywall, then the reach will increase and 50% of apps will increase their conversions to trial/payment. Try it out too :)

Artem, please tell us what the most extraordinary or complex product tests that you and your team have done so far are?

In my understanding, the most difficult tests are paywall price ones. This is what we call tests that affect paywall subscriptions. These kinds of tests involve changing the subscription price or subscription duration and adding a 3/7/14 day trial period, etc. These tests can not only change an app’s economics but also greatly affect media buying results. Almost always, we see that Facebook starts to provide another audience with a different CPM when you change the subscription from monthly to weekly or revise the prices for current subscriptions.

Such tests have to be conducted in a cohort. In one version of the app, you drive traffic to one paywall, and then you launch the next version with another paywall. Afterwards, you compare the data of those two versions.

How can you explain the phenomenon that in a world where the vast majority of users prefer to get something for free, subscription apps have become this popular? Is this an omen of some major changes? What should we expect in the next couple of years?

I guess there are two aspects of why the subscription monetization model works well.

First, there are major players (Netflix, Apple, Amazon, Tinder) who drive this model to users in different areas of life from movies to e-commerce. Large companies form a model of user behavior implying that you have to pay for content and additional features, making you stand out from other users.

And secondly, the sharing economy is more suitable for the younger generation. A subscription is a more profitable model for them, as it allows them to have something temporarily and quickly part with it. In this sense, the content in a subscription-based app is something they understand: today you pay and enjoy using it, but tomorrow you can switch to something else.

The thing that very few users can estimate is how much money they spend on subscriptions. According to the Subscription Trade Association, more than 60% of millennials do not know how much money they spend on subscriptions. I suppose in the next 2-3 years, Apple will make it possible to control subscriptions on devices. In this way, the number of people who pay for apps and don’t use them will decrease. This will lead to fewer low-quality apps on the market, as their economics will deteriorate badly.

Finally, our traditional question: what knowledge and skills should a mobile marketer have to always keep up with trends, especially if they work with subscription-based products? What resources can you recommend for the most curious ones?

To successfully scale a subscription product, you need to be able to buy traffic (and work with SKAdNetwork) and to be able to calculate subscription forecasts and ROI. Also, a very important skill is to know how to work with ad creatives. To be successful, you should be able to set a task for a designer on a great creative that you came up with yourself or got inspired from a competitor.

As for the resources, I personally read a lot to keep up with current trends, so here’s my list of the most useful sources of information for mobile marketers:

https://grow.co — a newsletter with the most significant mobile marketing news. Once a week, the news is collected and sent to your inbox. Convenient.

https://mobiledevmemo.com — high-quality articles and analysis of mobile market news.

https://www.consumeracquisition.com/blog — one of the coolest agency blogs that I’ve ever seen (in terms of content). These guys run insane amounts of spend for their clients. Their articles are cases of their expertise with hundreds of millions of spend. I pay special attention to articles about ad creatives since the analysis of data on creatives is done on a $100M+ spend.

https://www.thesistesting.com/insights — another agency that shares its clients’ specific hypotheses and averages their results. Here, I like to see concrete numbers and clear conclusions. The tests themselves are quite ordinary, but I believe every marketer sooner or later thought about the tests these guys talk about.