TapGravity: performance-based mobile ads – what you need to know

Kirill Sofronov TapGravity
When CPC or CPM ads do not work for advertisers, CPA ads where they pay only when a user installs the app or performs some other action seem a risk-free alternative.
In this episode of AppInTop Mobile App Marketing Podcast we speak with Kirill Sofronov, mobile marketing professional and the founder of the mobile-first marketing agency, TapGravity to understand the costs and risks associated with buying performance-based ads.  Listen to this podcast episode or subscribe on iTunes.
There’s a lot of demand for performance-based ads in the mobile industry.  The most common one is an ad where an advertiser pays per install.  Can you walk us through what kind of performance ads are available to mobile advertisers?
Currently we have ads which are available on a cost per install or a CPI basis, as well as cost per engagement and cost per sale.  Advertisers are becoming quite conservative on what they want and moving from buying ads on CPC and CPM basis to CPI.
There are also mobile content offers whereby advertisers are asking users to submit various forms. A single opt-in ad is where you have to fill in a certain form or an email.  Whenever someone submits it, it is counted as a conversion and advertisers pay the publisher for it.
Another variation of such an ad format is a double opt-in, where users have to confirm the email address to count it as a conversion.
What is really hot right now, is pay per call. It’s a form of local advertising where advertisers are getting billed whenever a call results. Let’s say if you’re an insurance company, you have an offer and when a prospective client calls and stays on the call for one minute or more, there is a conversion and you get paid.
There is even a hybrid model where an advertiser pays per install the first time, and when the user, for instance, reaches Level 5 in the game, they pay a second time.
What is one thing that you’re seeing in the mobile advertising area that no one is touching anymore?
I wouldn’t qualify is as not touching, but incentivized traffic is not considered the best option.  One of my colleagues from back in my days at HitFox, Thomas Sommer, wrote a really nice article about the three waves of mobile marketing.
The first wave was in 2009 to 2012 where incentivized installs were really kicking in and everybody wanted to get into the top ranks in the App Store to get organic installs. This is one of the things that is getting bagged because the quality of those installs is not really good and the market, for instance, in the US, is so competitive right now that only a few players can actually afford to pay this price for incentivized install in order to hit the ranking.
At the end of the day, if you don’t reach the volume of the top 20 or even 10, you don’t break even on the traffic that you’ve bought.
When we’re talking about a cost of paying users for games, for travel, for dating, what industry pays most per quality user?
Dating is really specific because it’s a bit easier to monetize.  In our experience it really depends on the country or the device.  In general, dating pays a bit more. Otherwise, games and utility apps are on the lower tiers.
You can always find extreme price levels.  There are games which bring only 30 cents per user, although app publishers already know their users so they can adjust the acquisition price accordingly.
On the other hand, gaming advertisers are really sophisticated. They know what their paying users are like so they usually cap the traffic volume and it’s not that easy to promote.
Currently, the biggest upswing I see is in the utility apps. One company which is turning out to be really, really successful is publically-traded Cheetah Mobile, because they created a new kind of market for utilities promoting antiviruses.
Banking and automotive, they’re not so much into app installs so far. Then there are sweepstakes offers where you can win a car, but they lack mobile promotion strategies, and I don’t see so much of an upswing in those industries.
If you have really, really relevant traffic for, let’s say, finance, you can monetize it well.  When the traffic is relevant, you can have really good quality because users are actually paying, and because financial institutions pay per sale or a commission on a purchase of insurance.
What is the market capacity for quality users for games, travel and entertainment when you’re paying per install and when you pay for other actions, say, a purchase?
There are two billion smartphones and tablets in the world, so the opportunity is really enormous.  In general, the market is staying big and it’s still evolving.
On the other hand, you always have to consider that there are more elements in that business. There are advertisers who are giving budgets for a certain type of offer, and then there are affiliate networks who are distributing these offers. Then there are affiliates who actually have to make their bread and butter every day and they don’t like cost-per-sale offers because it’s much harder to buy traffic and make conversions of click to install or a purchase on that traffic. It has to be relevant to be successful.
So in general, capacity is huge, and it’s increasing every day, but for certain kinds of offers, it’s really hard.  Organic CPI installs are still the default thing in the market right now.
For mobile commerce, an advertiser pays per order transaction. How does that work technically?
They’re starting to do it because sometimes when advertisers are not very successful in getting installs on a CPI basis, they are looking for models that work for them so that they don’t have to overpay.
Technically, there is the possibility of sending a device ID, and a click attribution.  Analytics and tracking platforms can do ID-based attribution which lasts around seven days. One of the companies I work with, Adjust.com, does just that and I highly recommend it to my clients because they can track most of the attributions and the sales as well.
Are there risks involved with that?
In general, the risks are more on the publisher side when they go for performance marketing.  For advertisers it makes much more sense, but for publishers, sometimes it’s really hard to make a profit on certain offers.  So, they have to buy a lot of data.  For instance, if you have an offer which pays $50, you have to spend a lot of money to gain a lot of data to see what converts.
If it is an ad where advertiser pays per order, what happens when the customer returns the order?
If the purchase is a commission-type of a deal, an advertiser might offer a commission in the range between 1 and 20% for a closed order.
When it comes to returns, it’s usually specified in the insertion orders so it really depends on the advertiser.  In general, they might take the risks by themselves and they make a ‘no refund’ policy on commissions.  On the other hand, it might happen that they’ll say, “Okay, if there is a refund, we’re going to charge you back on whatever has been refunded.”
When a game publisher buys guaranteed installs from the marketing agency, what do pricing and conditions look like? Are there risks of paying a higher price per install?
Usually, advertisers already have established a certain base for install. So if an unsophisticated advertiser just starts with user acquisition, it makes sense to test out a certain price range in a certain country. An agency like TapGravity will give them feedback as to whether it works or not.
At the end of the day it comes down to communication between two parties on what works and what does not. Many companies are offering technology products which aim to optimize the installs on an LTV basis, it slowly becomes a standard in the industry so that people do not overpay for their installs.
Do major CPI networks elsewhere have their own traffic?
Facebook actually started offering CPI installs as well as Google.  Twitter still doesn’t do it, but I think it might come along with their solution as well because it’s becoming a trend in the industry. Advertisers are becoming more conservative.  They don’t want to spend money for clicks. In Russia, Target@Mail.ru started to do it quite a long time ago as well.
Most CPI networks use external publishers to drive traffic, although many have turned into self-serve platforms, a kind of hybrid ad network.  CPI networks do not typically have their own traffic. They aggregate publishers in order to drive installs for the advertiser. CPI networks came along because none of the bigger players offered a solution.  That’s why most of the bigger players are beginning to offer a CPI product.
Unfortunately, the prices, let’s say, on Facebook and Google are still quite high so sometimes it doesn’t make sense for advertisers to pay such a big price. Everything depends on optimization and what kind of price advertisers are willing to pay for their app install.
Since advertisers are getting more picky and making sure that they’re getting quality people, how can the CPI network manage the risk and still make a profit?
A CPI network doesn’t have many risks. They are only taking risks with working capital when the advertiser cannot pay them but they don’t have any risks on traffic delivered.  They may have a risk of fraud when a publisher actually sends poor-quality traffic or mixes it with incentivized installs.  On the other hand, they don’t do CPC or CPM models so they usually don’t have any risks.
Recently, many affiliate networks started to create their own media buying teams internally which are buying traffic as well though. So they have a lot of insights about which offers are working for their affiliates. They have an advantage because they can actually promote their apps with 100% margin or without any margin at all.
The cheapest users on the market come from incentivized installs.  When a CPI network optimizes its profit, the most profitable strategy is to buy incentivized installs and sell them as non-incentivized at a premium price.  What’s the risk of buying incentivized installs from any of those CPI networks?
Generally, it’s actually classified as a breach of insertion order where advertisers usually specify what kind of installs they want, otherwise, a CPI network won’t be paid.
For the network, there is risk as it is hard to control publishers when they are mixing traffic.  It’s about the communication between the advertiser and the CPI network.
So CPI networks are trying to make it quite transparent where the traffic is coming from, although there are also cases when you have to scrap a certain amount of a budget because some publishers were not delivering the quality that they were asked for.  Otherwise, many CPI networks are working on certain fraud algorithms so that these cases don’t happen.
How do you protect your clients from buying dead users, aside from the communication?
You use proven top networks that you are buying the traffic from. You have certain ad traffic sources which you already tested and you know they worked and you know that the ad network won’t send you bot clicks or dead users.
In general, it’s about knowing the industry and having feedback from the advertiser in time and combination with the fraud mechanisms of affiliate networks which prevents you from getting dead users.
If the CPI network sold you dead users, how can you get the money back?
Usually, advertisers specify in the insertion order that for all leads that come from a fraud source, the CPI network isn’t going to get paid and affiliates aren’t going to get paid.
So it’s a general practice that at the end of the month when an advertiser sends a report on the quality of the traffic, the affiliate network scraps the leads or contacts the publishers, or they’re going to switch the users upfront.
In TapGravity, we do it on a weekly basis usually, so we have reports coming from the advertiser and we’d see what kind of users they’re actually getting, what are quality traffic sources and what are not.  When we are actually buying traffic, we only pick certain traffic sources which always comply with the rules of the insertion order.
A lot of advertisers actually got burned because they started with a really high CPI price.  They launched a campaign with a 50K budget and spent the whole budget, let’s say, in a couple of days, and then they realized something like 50% of it is actually dead traffic.
It happens frequently, I would say, with many networks.  Because right now the industry is super competitive. There are so many players which are not proven.  Nobody knows about them.
It’s really hard to establish yourself, let’s say, as a quality network.  So that’s why a lot of advertisers work exclusively with, let’s say, two to three networks because they’ve already worked with them for a long time and they know that they bring quality leads.
Let’s talk about burst campaigns.  Sometimes you need incentivized installs to run burst campaigns to get you to the top of the ranks within a very short period of time.  What volume of incent traffic can you get per day?
It’s really highly dependent on the country and traffic availability for this specific timeframe.  If you aggregate sources, you can get dozens of thousands of installs per day.  It only depends upon what kind of actual sources you want to use.
So we only use the sources that are proven and which we have worked with for a long time, and know that they can deliver.
What often happens, when you start a media buying or incentivized burst campaign during the weekend, some partners don’t deliver. This happens very often unfortunately so you always have to be prepared to engage another source or come up with another solution.
So whenever we have a client who needs a burst campaign say during the weekend we first and foremost research what volume can be delivered during that time.
Of course it might go bad. Most of the installs are actually coming during the weekend because it’s the best time to have burst campaigns.  So managers who are my contacts in those networks might be out during the weekends and you usually have to be active all the time in order to monitor the campaign and get as many installs as possible.
For example, on one of the burst campaigns Wooga worked with 23 ad sources. Give us some examples of ad networks advertisers should work with.
There are really high profile sources like Fyber, Tapjoy or Supersonic Ads.  I highly recommend those sources for any company who would like to get incentivized installs or who want to have a burst campaign.  I’ve worked with them for quite a long time, and, I think they are one of the best in the market.
On the other hand, there are other sources which are not that high profile but can work as well.  But at the end of the day, everything comes down to experience.  If you have worked with these sources, and it worked quite well for you, you should continue working with them but also see if you can test others for more volume.
Let’s say, a mobile marketing agency wants to implement a burst campaign, do you have any advice for them?  How many hours does it take to prepare?  What tasks are the most time consuming?
Yes, it’s actually a bit time consuming.  We talk to the client about date, volume, price, countries where the campaigns are going to run, as well as what goes with the campaign, the relevant KPIs that we would like to achieve.
It takes several days to prepare a campaign because sometimes some sources are already running other campaigns and won’t have the availability in a certain country. So I would definitely recommend preparing the burst campaign at least several days or even two weeks in advance to increase your chances of success.
On a global scale, what should mobile developers keep in mind when selecting countries for a global launch?
Most people want to launch in the US because it’s the highest quality market.  There are a lot of users, but at the end of the day, it is the most competitive market. Recently, a lot of advertisers turned to emerging markets like Brazil or Russia. There you can find a lot of quality users who are able to pay so you can win big there, too.
The pricing as well will be much higher in the US than, let’s say, in Thailand. The US is actually really not recommended currently because there is a lot going on there.  There are a lot of apps launching.
You need a big budget as well.  Only a few players in the market can actually afford to launch burst campaigns.  For a developer who only just started and developed his first game, user acquisition in the US doesn’t make sense, at least on a burst basis.
I’d really recommend starting with the organic installs and start small to see what the quality is. Keep communicating with the agency or a network to see how can you increase the volume, and the quality.
What’s one thing a mobile developer should most avoid when it comes to a global launch?
Definitely avoid shady networks. The industry is booming and there are a lot of players trying to operate in the market.
Right now there is a lot of re-brokering.  So for instance, Supercell stopped working with Appia because unfortunately they were re-brokering some ads. That’s what’s happening in the market because there are so many players.  They don’t have contact with direct advertisers so they take the offer and they re-broker it within other networks and there is substantial chaos.
There was an example of that when I was still working in HitFox.  We had a burst campaign and we engaged five networks, and the campaign went through so many sources that it actually came back to me and somebody offered me the same campaign that we were the ones actually promoting, and I was like, “Hey, what the hell is going on?  You cannot re-broker.”
So in general I’d avoid networks and agencies without any references, otherwise, you won’t know where their traffic is coming from.  If you don’t care about the quality and you only need any traffic, of course, you can engage with any kind of player but in general, just keep in mind who you’re working with.
What kind of budget would you be talking about to get, say, to the top 10 ranks of the US gaming category?  
I think you have to have at least 50,000-plus installs per day in order to get to the top ten ranks.  That means you have to run for a couple of days with this level of installs in order to be counted into the ranks of the top 10.  It would seriously cost you something like dozens of thousands of dollars, if not hundreds of thousands of dollars, to achieve that kind of a success.
That’s why I mentioned that only big players use the burst campaigns in order to get the traffic because not many developers can afford it.
There are many products, which allow developers to acquire users organically at a lower price.  Burst campaigns are only good for a certain amount of time.  From statistics I read a long time ago, for each incentivized app install you get probably around half of an organic user.
In general, the virality factor is not that big, and it makes sense only if you have a big budget, but for smaller developers which is a majority of eco-system, it doesn’t make much sense.
Why should an advertiser go to a marketing agency?  Who is your typical client, and what should somebody be expecting to pay?
TapGravity’s main focus involves advertisers for whom we help reach additional volume in their current global marketing activities.
There are two business models.  One is an agency model where we help direct advertisers to achieve their goals and revenues.  The primary model is though working with a variety of ad networks and drive installs to their offers by buying traffic on the CPC and CPM model, and risk our own money to get our own return on investment on arbitrage.
In general, the fees for an agency model are around 10 to 15% on the CPI or on the budget.  In terms of media buying, we just usually go with the offers from the variety of affiliate networks which advertise directly and start buying on the ad networks or we just do both models combined.
Why wouldn’t an advertiser go to a marketing agency?
If somebody wants to try media buying, they usually just go through an affiliate network. If you are starting, it doesn’t make sense to actually go to the agency but instead go to a network directly, but sometimes, if you really don’t have the experience, I think it makes sense to consult first.  We can give you all the information you need to start with as well as proven partners you can work with.
This is an abridged and edited interview from the AppInTop mobile app marketing podcast produced by AppInTop, an automated mobile app marketing platform. Listen to this podcast episode or subscribe to the podcast via iTunes.