This post is part of a series of interviews with successful entrepreneurs. The questions are focussed on marketing, advertising and analytics aspects of  building and growing a successful online business.

This interview features Jakob Lykkegaard. Jakob is the co-founder and CEO of PlayLab, a multi-million dollar mobile app development company.

PlayLab’s headquarters are located in Hong Kong while distribution is handled from Singapore. The company’s production sites are located in Bangkok and Manila.

PlayLab has generated more than 22 million downloads and created award-winning games such as Lost Cubes and Juice Cubes (PlayLab, 2015).

In this interview, Jakob answers 7 questions on building, marketing and growing a mobile app development business.

Jakob Lykkegaard, Co-Founder & CEO of PlayLab

What advice would you give someone who wants to start developing and marketing mobile apps?

Get something small out as fast as possible and fail fast. You learn a ton from making your first few games/apps on everything from what plugins to use, how to load scenes fast, till how to market and monetise them.

There are so many tools and engines to help you these days that you can get to market really fast without having to make everything from scratch, do keep in mind that there are always tradeoffs to the different tools, so choose wisely.

If you had to restrict yourself to one operating system, would you choose Android or iOS and why?

Even as a hardcore Apple fan I would probably pick Android for 2 reasons.

The first reason is that Android starts to dominate on most new markets, so looking a few years ahead when credit card penetration and spending power has gone up, then Android would have the biggest marketshare spread over multiple telco app stores as well.

Second reason is that most android stores incl. Google Play are a lot more developer friendly, you can push out build the same day they are done, so you get data faster and can easily fix any problems.

What are the most effective marketing strategies for mobile apps, both paid and organic?

There are too many developers these days that release their games and pray to the flappy bird gods that PewDiePie will pick it up so it will go viral, in reality that is however never the case.

So most effective marketing strategy today is Paid user acquisition of real users, as you dont get anywhere near top 100 grossing without it.

That however also means that you need to make games or apps that allows people to do multiple in app purchases, as you would need to on average make more than $1 per download to keep the campaigns going.

What are the best advertising networks for driving mobile installs and why?

Facebook is still no.1 worldwide, their margins has gone up a lot over the years, but 90% of our paid traffic is still coming from there and then around 30 other networks share the last 10%, as rewarded videos are starting to pick up as well.

You are spending millions of ad dollar on Facebook ads. What targeting and optimization strategies have you found to be most effective?

We run thousands of campaigns across the different strategies as they are all good for different things, luckily then we now have a full team taking care of it so they can always test the different trends.

Most effective is however still to exactly figure out who your highest spending players are and then target more with similar interests or lookalike audiences.

It is also important that you use an attribution tool to get the whole picture, mainly so you don’t shut down campaigns that are attracting a lot of spending players, because the ads clickthrough rate is looks low on Facebook.

What analytics tools do you recommend for tracking cost per install, user loyalty and customer lifetime value?

We mainly use our own build analytics system as we gather data from so many different sources, but main attribution tool is Appsflyer and we still use Flurry for simple stuff like retention, but everything else are calculated on our backend and visualised through Tableau.

Any other final thoughts, tips or comments?

Just test, test and then test some more, then you will eventually find something that works.