+5 votes
As all gaming companies, we want to grow. The problem is most of our performing campaigns on FB are value optimization. It means we have pretty good results, but not enough players. We have just PvP games so it is a problem. Of course, we try to do AEO targeting and results are not good, not terrible, but mainly in green numbers.

CPI campaigns don't work. There is no performance at all.

Do you have any tips on how to do FB CPI campaigns profitable?

What setting do you use for CPI campaigns?

Do you have any secret golden rule?

What target event are you use (lower costs, costcap, bidcap, targetcost)?

What are the differences and what are the special reasons to use each one of these options?

In which performance metrics there is the biggest difference among these types (if any)?

Do you use another goal for CPI campaigns?

Or are you just reconciled to inefficient acquiring of users?
by (460 points)

1 Answer

0 votes

Mobile app install (MAI) campaigns using just interest targeting are very difficult to make profitable for games unless the games are very, very broadly appealing.

In my opinion, the only reason to use interest targeting for MAI campaigns is to build a large funnel into your game that you can then start harvesting custom audience LaL lists from, preferably using high-value payers. Note that if you are starting from scratch -- that is, you need to establish the initial seed lists and build a set of payers from 0 -- this is a time consuming and expensive process. I wrote about that here: The slow and costly process of scaling user acquisition

But trying to move into AEO or VEO campaigns before you have these lists won't work -- not really. You might get a small trickle of users, and they'll likely pay, but without a large enough LaL audience to target against, you'll never really be able to scale an AEO or VO campaign. So you have to do this initial step, which is harvesting lots of users cheaply using MAI campaigns and building lists from the valuable ones.

You can incrementally increase the value of these lists by building them off of increasingly valuable behavioral signals. For instance, you might:

  1. Build a list of Day 1 retainers, target that;
  2. Build a list of Day 3 retainers, target that; 
  3. Build a list of payers, target that;
  4. Build a list of high-value payers, target that, etc.

This is slow, iterative, frustrating, and expensive. But if you are starting from 0, I don't see any other way on Facebook.

by (15.2k points)
>But trying to move into AEO or VEO campaigns before you have these lists won't work -- not really. You might get a small trickle of users, and they'll likely pay, but without a large enough LaL audience to target against, you'll never really be able to scale an AEO or VO campaign.

Our experience in ZeptoLab is quite different. While lookalike audiences do help, AEO and VO work really well when broad audiences, especially VO. Generally, more of our spend goes towards broader audiences.

I guess it depends on the product, but I've heard similar things from a bunch of other gaming companies
Interesting. At what scale? My experience is that broad targeting for AEO and VO works at some low level of spend (ca. $250-500k / month), but prices start getting out of control with broad targeting beyond that.
I have to agree with Danil. VO campaign results are the best when we are targeting to just social demo. Our scale is around 600k+ monthly. But when we want to target social demo also for AEO, the overlap of users is more than 90 %. That is why we use just LaL for AEO.
On a significant scale. And yeah, at some level of spend the prices will start growing within a single region or worldwide.

For us good LAL audiences that generally have similar ROI to broad/demo targeting. Both CPI and LTV are usually higher, but the ROI is more or less the same. So for us LAL audiences are a good way to scale horizontally when we understand that we can't scale the broad audience anymore, but want to expand further in the particular region.

Another trick that we use is using a broad "games" interest with AEO purchase. While interest targeting is largely obsolete, this one seems to work well specifically for purchase. It narrows down potential reach by 20-30% depending on the region, but for AEO purchase specifically it seems to increase LTV a lot. So we typically use this interest with AEO purchase and a very broad audience with VO
Great insight. Thank you for sharing!
Danil if 'interest targeting is largely obsolete', why is Facebook still considered one of the main UA channels for games then? And are there any other channels you recommend?
What I meant to say is that interest targeting wasn't the main way to do UA on Facebook as it once was (before 2016). This doesn't mean Facebook isn't important – it's still the number one channel, it's just that the focus is now on wider audiences, lookalikes and app event optimization.

As for other channels – there's obviously Google UAC. I would also recommend paying close attention to SDK ad networks with a strong emphasis on Unity. They've been investing a lot into machine learning and app event optimization and are now on a level similar to Facebook. Ad networks are also actually easier to get started with than Google. Some other strong choices would be Iron Source, Applovin and Vungle.