I asked Victor Dekalov from Scoompa to give a short interview. To those of you who don’t know Scoompa, it’s a mobile-apps start up founded by Google veterans. They have a diverse portfolio of popular Android apps that have been downloaded over 200M times.
Q: Tell us a bit about yourself and the company you work for
I’ve been managing monetization and analytics with web & mobile publishers for the past 5 years. I was lucky enough to experience the evolvement of these landscapes first hand. I started by monetizing with premium direct campaigns and evolved with the industry through the rise of programmatic ad channels, to full stack programmatic mediation and content discovery in mobile.
Today I’m leading revenue & analytics at Scoompa – our apps are used by 15-20M monthly active users
Google and Facebook are enough
I asked Victor to describe his mix of ad-types, networks and mediation providers
Up until now our approach was not to over complicate things.
We have simple flows inside the apps that trigger interstitials on natural breaks in the app usage (finished creating doc, etc). We also have banner ads on most of our screens, which in the past few months are being gradually replaced by native ad slots that fit in nicely in the apps’ UI.
We use our own in-house auto-pilot mediation system that optimizes the waterfalls, on a daily basis, separately for each placement and each country.
The vast majority of our ad traffic is split between FAN (Facebook audience network) and Google (Admob, adx) + few more we try from time to time. We consider different ad-networks from time to time but usually the opportunity cost is too high and we stick mostly with FAN and Admob.
Don’t over optimize focus on the big picture
Since ad-operations is a relatively new practice for a lot of mobile app developers. I thought it would be good to learn a bit from the expert. I first asked Victor to describe his day to day and was surprised to learn it’s not so much about optimizing the eCPM (or RPM as he calls it)
Q: What is included in your day to day as the one heading Ad-Operations
Since most of our ad-ops are “managed” by our auto-pilot mediation platform, it frees me up to focus more on planning ads related product changes and analyzing their effect. This is key: as ad-ops people we tend to focus too much on optimizing our small piece of the puzzle like increasing RPMs (revenue per mile) at any cost. We forget that this is only the very “end of the funnel”.
You can get more impact by creating better ad interactions that perform better (lead to better conversion but also better UX and retention), working on cross-promotion and general growth, analyzing the usage of features inside the product to identify the best monetization opportunities.
Q: Can you give us an example and tips how to do it?
One thing we noticed is that users mindfully CLICK on native ads. They do it intentionally. On the other hand banners, whose CTRs are so low that they might look like “margin of error” more than a KPI. A lot of my role is to find these native opportunities rather than just filling the screen with banners. We found that the best implementation of native ads for us is in a list of discoverable content items. This yielded a steady 2% CTR and 4-5$ RPM in US. Even when having several ads per screen.
Another thing we found is that interstitials, If implemented correctly and with the right partners, can bring higher RPMs than rewarded-videos.
So back to my 2 cents – over optimization leaves you focused on the tiny details. In the long run, being blind to the big picture will always loose you more money than what you can earn by lifting RPM by another 0.5%.
LTV is hard to calculate – power through the excel
A lot has been written about KPIs and the importance of LTV. I asked Victor if he shares the same views and he shared some interesting perspective about the challenges with LTV.
Q: What are the KPIs you are actively tracking and some benchmarks
ARPU/LTV – this is important since it’s the ultimate monetization goal. For a utility app you should shoot for 2-4 cents in top GEOs and 1 cent elsewhere. At least that’s what we are seeing.
Tip – Don’t give up on ARPU and LTV analysis. In most cases, it’s not easy. You need to work with several data points, they don’t always “speak the same language”, you need to spend hours pivoting and vLookuping in excel…… It’s worth it! You’ll find great insights that’ll contradict your instincts, and in the end you might double down on a BI tool that will help you streamline those analysis.
Impressions per session – we don’t try to optimize this one up but rather balance it – too much ad inventory devalues each impression. Too many ads reduce retention. Those 2 things are hard to come back from, and rather than obsessively optimizing for the upper limit I suggest to use common sense and practice restrain (of course, after you’ve analyzed your range). For us we balance for 1 interstitial per session and 4-6 banners.
RPM/eCPM – did I do something to change performance of the ad format? Is there an industry/seasonal trend? The variance is pretty high between ad formats and GEOs. For USA: interstitials should average 6$, natives should be 2-3$ (for very large native ads can be same as interstitial)
Note that RPMs can change industry wide by 50% from one period to another. It can happen in a matter of days, and can influence just one ad-source and not the other.
Finally, I recommend finding your top 5 GEOs and optimizing them. Even when you have a very diverse WW user base, often 5 countries can account for over 50% of ad revenues. Think about it when you’re spending all your energy on localizing for Africa.
Calculate ROI on new SDKs and factor implementation effort as well
With new tools and ad-network SDKs coming to the eco-system all the time, I had to get Victor’s views on how to find the right partners.
Here are some of his thoughts about tools
I mostly use excel and SQL (workbench) to combine ads data with analytics data from Flurry, Google Analytics, Fabric, etc. Don’t spring for expensive and time consuming BI tools, if you can’t estimate how much ROI you’ll have from such a project.
From previous experience, and if you have the time resource, I extremely recommend creating pipelines to streamline all this data into a unified database (redshift, bigQuery, etc) and using a data querying and visualization tool to give you easy day-to-day access to all your main KPIs. You don’t have to go for the big and expensive solutions (like Tableau), there are great and lean (and MUCH cheaper) solutions out there that will do for 90% of use cases. I recommend redash.io
And about adding more ad-networks
We always want to add more networks to our stack and increase average yield per impression, but we always ask ourselves these questions : can this action improve our LTV by more than X %? Will it be more profitable than investing the same development time in new apps and features? So far the answer is usually “NO”.
Cross promoting is key
Q: What do you think about user acquisition for ad supported apps?
Buying mobile users, for a utility app, hoping to be ROI positive by monetizing with ads, is a dead end. Unless you truly believe that you have a home-run app that just needs some advertising to get discovered and become viral. If you plan on generating an average LTV that’ll be higher than the cost of user acquisition (actual CPI + overhead cost), know that for 99% of use cases it’s just plain impossible
A great publishing business is one that has diverse properties. It can utilize cross-promotion to expose masses of users to new releases, instead of relying on paid advertising or chance.
Q: Any other tips you want to share?
Our landscape is full of people eager to share their experience and results. You don’t always have to reinvent the wheel or try-fail every single idea. On the other hand – be sceptic. Not every blog post is relevant to your business and not every piece of data is objectively true 😉