Freemium, brand and performance marketing: a subscription app growth guide
Marketing a subscription app is not only a matter of buying installs. You need people to discover the app, understand why it is worth paying for, and stay long enough to form a habit. That is a different problem from simply maximizing conversion on day one. The lessons below come from a conversation with Sherina Khalidi, VP of Performance Marketing at Deezer, on the Sub Club Podcast. They cover three connected decisions: what the free tier should give away, how brand and performance marketing divide the work, and how to measure progress when attribution is incomplete.
Make the free tier genuinely useful
A common early mistake is to treat the free tier as a pressure chamber. If you strip away so much that the free experience becomes frustrating, users do not upgrade — they leave. Deezer learned this when aggressive conversion pushes made free users abandon the app.
The fix was to make the free experience enjoyable enough that people stay. That changes the job of the free tier. It is not only a preview of paid features; it is a place where habit and trust can form. Some of those users may convert later. Others may never pay, but they can still contribute revenue through advertising from brand partners.
Use this checklist before tightening a free tier:
- Can a new user complete the core task without hitting a wall?
- Does the free experience build a repeat habit, or only a one-time trial?
- Are non-paying users contributing revenue in some other way, such as ads?
- Is the upgrade prompt timed to a moment of demonstrated value rather than used as a constant interruption?
- Would a frustrated user leave rather than pay?
The distinction matters: a free tier that frustrates users into leaving is a retention failure, not a conversion strategy. A paywall that converts is only useful if the surrounding free experience has not already driven people away.
Pair brand marketing with performance marketing
Performance marketing can produce quick results when you open new acquisition channels. But those gains tend to plateau. According to Khalidi, you quickly reach a ceiling beyond which you cannot scale again. The way past that ceiling is to combine performance marketing with brand marketing.
Deezer's example is a TV campaign that does not mention free trials, pricing, or even ask viewers to download the app. It simply shows what using Deezer feels like. That kind of top-of-funnel work builds a positive brand image, which can make later bottom-of-funnel offers more effective.
A practical way to think about the split:
- Brand marketing creates familiarity and preference before a user is ready to act.
- Performance marketing captures intent and converts it with offers, installs, and re-engagement.
- Neither alone is a complete growth system; the two reinforce each other.
If you are deciding where to invest, start by asking whether your acquisition channels have plateaued. If they have, more spend on the same performance tactics may not help. A brand layer can widen the top of the funnel so that performance spend has more room to work.
Measure with more than one lens
Privacy rules make it harder to trace which marketing efforts actually affect revenue. Traditional attribution models also tend to miss the effect of brand marketing. Deezer's team uses three complementary methods to get a more nuanced picture:
- Media mix modeling: estimates how different channels contribute at a aggregate level.
- Incrementality testing: checks whether a campaign caused outcomes that would not have happened otherwise.
- Brand lift studies: measures changes in awareness or perception associated with brand campaigns.
Each method answers a different question. None is a complete scorecard on its own. Use them together to decide where to shift budget, and treat any single attribution number as one input rather than the final answer.
Sometimes the data and your judgment disagree. Khalidi describes an A/B test of several new Deezer homepages in which none beat the control. The team still believed the control was not the best option, so they implemented a different version. After several weeks, they saw a clear uptick in performance compared with the previous page.
That story is a useful example, but it is not proof that gut instinct beats testing. It shows a case where the test did not capture what the team cared about, and where a later observation supported their judgment. The lesson is to know what your test is actually measuring, and to be willing to revisit a decision when the test and the product evidence diverge.
Put the three decisions together
Freemium design, brand and performance marketing, and measurement are not separate projects. A free tier that builds trust gives brand marketing something credible to talk about. Brand marketing makes performance offers more effective. Better measurement tells you which parts of that system are working.
Before your next growth review, check three things: whether the free tier is genuinely usable, whether acquisition has plateaued because brand is missing, and whether your measurement plan includes more than one method. If any answer is no, that is the place to start.
Frequently asked questions
What is the most common freemium mistake for subscription apps?
Taking away so much from the free experience that users become frustrated and leave instead of upgrading. A free tier works better when it builds habit and trust before asking for payment.
Why can't performance marketing scale on its own?
Performance channels can produce short-term gains, but they tend to hit a ceiling. Brand marketing builds familiarity at the top of the funnel, which can make later performance offers more effective.
How can I measure brand marketing when attribution is incomplete?
Use several methods together, such as media mix modeling, incrementality testing, and brand lift studies. Each answers a different question, and no single method gives a complete picture.
When should I trust my gut over an A/B test?
When you understand what the test is measuring and have reason to believe it is missing something important. In the Deezer example, none of the tested homepages beat the control, but the team chose a different version and later saw an uptick. That is a reason to revisit a decision, not a rule that instinct always wins.