Facebook Crushes the Competition With Mobile App Install Ads

19 Dec 2012

January 08, 2013 update: Interesting Results from Facebook's New Mobile Ad Features

This post is about my experience running mobile ads on iOS with the goal of acquiring new users for an iPhone app. My findings and recommendations are based solely on this use case. For information on setting up ads on different ad networks check out my related post Getting Started with Install Ads for iPhone Apps.

Why buy users?

When we launched our first app, Miso, over two years ago we were very proud of growing our users organically. But in early November when we launched a new iPhone app, Quips, we decided to try something different.

We wanted acquire new users as quickly as possible. The theory goes like this: the faster we get a strong base of users, the faster we can test features with significant results, the faster we can validate what works, and the faster we can ship a successful product. Also, because Quips is a social app we believe that the more users we have, the better the experience; and the better the experience, the more likely users will become retained and refer friends.

So to test our theory and acquire new users for Quips, I set out to answer these questions:

• What do different mobile ad networks offer for app developers?
• How does the cost-per-install (CPI) model work?
• What ads and placements work best for our app, Quips?

After some quick research, I created a list of the following ad networks to try out:

• AdMob (Google)
• iAd
• Facebook
• Jumptap
• MdotM
• mMedia (Millennial Media)
• TapJoy (incentivized and non-incentivized)

After attempting to set up accounts and deploy test campaigns across all of these vendors we ended up focusing most of our efforts on Facebook, MdotM, and TapJoy. In a separate post I detailed my experience in getting started with all of these networks.

My Top 4 Takeaways

1. Get Facebook Mobile Install Ads While They’re Hot!

Facebook mobile install ads delivered us the best results across the board.

• $0.45 Average cost-per-click (CPC)
• 26% Conversion rate of installs (actions) to clicks
• $1.73 Average cost per user

However, don’t expect this great deal to last forever. When I first ran these campaigns I had never even seen a mobile install ad in my Newsfeed despite my best efforts to find one. Now it seems like I see one every time I open Facebook on iPhone.

Over the last several weeks I’ve seen suggested CPC bids increasing to over $1.00 in some cases, though typically the range was $0.20 to $0.50 on our campaigns. Bids change depending on the target audience, so it’s not surprising to see higher demand for socially active college grads with iPhones.

I’ve also seen the number of impressions served in our campaign slow down recently, despite continuing to raise our bid to stay within the suggested range. Moreover, my gut tells me that these ads are performing well because they are still new to users, and I expect to see click-through-rates (CTRs) fall over time.

So seriously, get them while they’re hot. Start today. And if you have the Facebook SDK already integrated you won’t have to release any new code in your app. In fact, you can probably get a campaign up and running in less than 60 minutes from reading this post.

Pro Tip: Check out Facebook’s official tutorial, but ignore step #5 and use the Power Editor to access advanced targeting options. Here are instructions for using the Power Editor

2. Outside of Facebook, Stick to Buying Ads on a CPI Basis

One of the reasons that Facebook ads worked well for us is our ads have an extremely high click to install conversion rate, around 25%. As a result, Facebook was the only network where it made sense to pay CPC. I did try setting a CPI on Facebook for $3.00 (done through the optimized CPM option, by setting a max bid on the action) but it ended up serving so few impressions that I could only acquire a couple, sometimes even zero users in a single day.

Aside from Facebook, I would recommend using CPI in order to control cost, since the conversion rates on clicks outside Facebook are much lower and vary widely. I ran CPI campaigns on MdotM and Tapjoy (non-incentivized) and saw much lower conversion rates (0.5% to 5%) – nowhere near Facebook’s 25%. One ad expert told me that industry standards are typically 2% to 5%. So consider this hypothetical CPC campaign:

• You pay $0.20 CPC
• $100 dollars yields 500 clicks
• At a 5% conversion you end up with 25 users
• Each user cost you $4.00

Let’s say you only wanted to spend $2.00 a user, you could lower your CPC to $0.10 (though you’re starting to risk that your ads might not run). But you still can’t guarantee your conversion rate will stay at 5%. If it ends up being 2%, you’re back to paying $4.00 a user. I ran several CPC test campaigns on AdMob and Jumptap and didn’t crack a 1% conversion on either one of them. These tests were capped at a $100 day, and the net result was that that we ended up spending between $10 to $20 per user.

Pro-tip: Think about aligning incentives. When you pay CPC, the network just has to deliver clicks to get paid – the burden and risk of conversion falls on you. When you pay CPI, you are aligning your incentives making it easier to lean on your ad vendors for support in optimizing the entire funnel.

3. The Rate of User Acquisition Depends on Your CPI

As I mentioned in the introduction, one of the perceived benefits of buying users is speed of acquisition. So while CPI has its benefits, be aware that the rate at which you acquire installs is a function of the CPI you set.

Your daily budget (which you can set with any of the ad networks) will control the upper bound of installs, but the more common problem is not getting enough installs. If your bid is below average bids, your ads won’t get served. Raising your CPI is one obvious solution, and you may need to constantly adjust this number if you are trying to hit a daily goal.

Your best bet is to chat with ad manager to get detailed reports about your campaigns and have them help you hit your goals. If your conversion rates are low that means your effective CPC and CPM are low, which may cause the network to slow down your campaign. Remember your incentives are aligned when paying CPI, so have them help you optimize across all these areas.

This chart demonstrates how installs fluctuated for one of our campaigns over a 10-day period when CPI was fixed at $2.00:

Pro Tip: Don’t forget to look at your click-through-rate (CTR) and install conversion rate. CTR can indicate effectiveness of your creative and/or ad placements, and the conversion rate may point to the effectiveness of your app stores page, all of which you can optimize

4. Incentivized Installs Do Not Equal Signups

This should obvious right? If you are thinking about buying ads, your app probably isn’t compelling enough to win over users from incentivized installs, so don’t bother. Logically we can assume these users don’t give a rip about your app, they only want points, coins, ponies, or whatever it is they get in exchange for downloading.

The counterargument is that if your app is awesome enough you could get a user hooked, if only they install it. The problem I saw in our campaigns is that most of these users don’t even bother giving the app a try. According to our data, less than 30% of the users who launched the app signed up to use it – and even among those who signed up, most of them never opened the app after their first session.

Below is a chart I keep in a google doc to track signups to downloads. In the last week when downloads peaked, we saw our percentage of signups drop from 70% to less than 30% of total downloads. This was the week we ran the TapJoy incentivized install campaign.

Pro Tip: If you run an incentivized campaign, remember that while the number of downloads and installs may look good, you should follow your conversion funnel to the end to see if these individuals actually become users.

Conclusion

I was very happy with the results from Facebook mobile ads and we are going to keep running them as long as we see great results. Given my experience, I would recommend Facebook mobile ads to any developers who are already using the Facebook SDK and looking to inject some life into their user growth.

If you found this post helpful, do me a favor and check out Quips! If you want to get notified of my next post – Getting Started with Mobile Install Ad Networks – follow me @katiesmillie on Twitter or check back in about a week.

UPDATE: As I was finalizing this post, I saw that Facebook released new features for their Mobile App Install Ads yesterday. The option of a larger custom image will probably lead to another streak of great CTRs as users experience these ads for the first time. Also developers can now run different ad creative simultaneously. More good reasons to jump on these now!