Mobile Paid Growth
How we increased monthly growth 4X and decreased customer acquisition cost by 3X: Long Tail Mobile Paid Acquisition in 4 steps
In this article we are focusing on our paid acquisition initiatives at my company. Historically, paid acquisition has represented a small portion of our user growth.
Most companies spend a significant amount of their user acquisition budget on a few channels to ensure they spend enough time and effort optimizing each channel prior to scaling. This has long been the conventional approach.
We took the opposite approach and produced tremendous results in a short time frame. In this article I explain why we used a long tail-portfolio strategy and will also provide examples about how a healthy mix of multiple networks enabled us to both decrease the Customer Acquisition Cost (CAC) by almost 3X and increase the average monthly growth by 4X within 6 months. This approach worked for us in our performance marketing efforts.
Among all the optimization efforts, we invested in multiple channels with low bids; instead of investing in few of them with higher bids. Imagine you are an apple farmer and you pick all the apples from one tree to fill your basket. If instead you find apples — customers — from 100 different trees, then you can pick only the lowest hanging fruit from each tree — or network, in this analogy — and you still fill your basket without much effort.
Step 0: Know your LTV
Before spending any money on customer acquisition, you need to know how much value you can gain from each customer and their long term impact. We call this Customer Lifetime Value (LTV). Here is how we broke it down:
LTV = marginal value of each customer * average life time
There are also more sophisticated LTV formulas here depending on your need. Companies define After calculating LTV, you can define your customer acquisition cost (CAC) target. CAC is basically how much are you willing to pay to acquire a user. “Acquiring a user” can be sign up, install, subscription or any other metric depending on company KPIs. (Kissmetrics explains CAC this way)
As Andrew Chen mentioned in his article, a healthy ratio would be 3:1 (LTV:CAC).
Step 1: Start with the user
The most successful advertising is said to be a reflection of who we are, what we do, how we think and what we feel.
Ideally we’d be able to customize each ad message for each user that we target. However, in consumer products, it’s difficult to scale that way. Instead, the best we can do is to create personas and try to put customers in each persona bucket. Some examples are: “Tech-savvies,” “Family Shoppers,” “Holiday Shoppers,” “Outdoor Enthusiasts,” etc.
While preparing these buckets, you need to ask yourself two questions:
1. Who are the customers I currently have?
2. Who is the ideal customer I want to attract?
Try to develop these buckets with as much detail as possible. (User-centric design techniques and Stanford’s d.School workshops would be helpful if this is new to you.)
Once you have the target personas, the next question is where to find them and how to convince them to use your product.
Step 2: List and classify every possible advertising and growth channel
Given today’s massive mobile advertising industry ($5.3 billion first half of 2014 with 76% growth rate), new ad networks and different growth tools are being launched every day. At Slice we used Feedly to subscribe to growth and marketing related blogs so we’re in the know about all new ad networks and tools. We then created a list of tools and ad networks that we work with or may work with in the future.
We classify growth channels into the following buckets:
· Social (Facebook, Instagram, Twitter, Pinterest, Snapchat… )
· Google (SEM, display, Gmail ads)
· Incentivized (Tapjoy, Fyber, Adaction, …)
· Discovery (Drippler, …)
· Video ads (Youtube, Vungle, Unlockable, …)
· Influencers (SocialX, Fango…)
· Programmatic (Rocket Fuel, AppNexus, … )
· Affiliates (Rakuten)
· Retargeting (Criteo)
· Data tools (PersonaGraph, …)
After you have the list, you might ask yourself “How and where do I start?” Self-service platforms such as: Facebook, Google, Twitter, NativeX, Aarki etc. are perfect starting points for two reasons.
First, you don’t need to start with a large budget . Second, they will help you figure out a rough upper bound on your CAC . (i.e If you are paying $3 to acquire a user on Facebook, you can use $3 as a benchmark and you know roughly what you should pay in other ad channels) Also this would help you figure out your appropriate test budget.
Some ad networks will not be able to hit your CAC upper bound and some of them will. Don’t worry! Just keep integrating new ones; you can use Ad mediators or aggregators. You will find enough channels that can match your CAC benchmark and yet provide good traffic . From our experience, just 1–2 weeks of data was enough to know the performance of new ad networks. Since there inherently is a trial-error method when choosing ad networks, it’s always good to start with as low budget as possible.
Also, know that each social network channel provides different value. For instance, we have found a sustainable and high-quality user base from “discovery” networks , while incentivized networks give us better rankings in app stores. Video ads are a great way to explain our value proposition, while social channels (e.g Twitter and Facebook) provide a demographic and behavioral data about user base.
Step3: Long tail/Portfolio Strategy: Have a healthy mix of many networks
When it comes to user acquisition, conventional wisdom suggests focusing on a couple of channels and optimizing them through time. However, this
approach has a drawback when we consider bidding mechanisms of ad networks. (For those of you who are not familiar with bidding mechanisms, the link above has Facebook’s bidding algorithm, whose logic is similar to other ad networks.) In a bidding algorithm, the only way to get higher traffic (assuming an optimized ad copy) is to bid higher. On the other hand, you can only bid as high as your LTV.
What if you want high-volume customer acquisition with low CAC?
We solved this problem with our long tail-portfolio strategy. Basically, instead of working with a few channels (i.e Facebook, Google, Flurry, Radiumone) we worked with 20+ different user acquisition channels to acquire the largest volume of users with the least possible cost.
We bid low on each channel so that we could maintain a low CAC. This brought us low traffic in each channel, but since we had so many active channels, we were still able to get a large pool of users (i.e., Instead of getting 1000 users from Facebook, get 100 from Facebook, 100 from Ad channel#1, 100 from Ad channel#2 ….. and 100 from network #25. Remember the apple tree analogy.)
Besides the low CAC, this strategy gave us another advantage in targeting and optimizing ad copy. Since we were working with different networks, we got many different data points. For instance, Facebook can provide demographical data, Twitter can provide different information for keywords and interest targeting, and another ad networks provide insights on geographical, creative, and ad copy data.
By only picking a few data points in each channel, we were able to optimize each individual user campaign.
Each channel has a different dynamic and brings a different value to your customer acquisition strategy. For instance, we run incentivized networks to improve our download volume (app store ranking), which places Slice in the top 20 rankings before weekends and, consequently, provides us with organic installs. Additionally, Facebook increased our viral coefficient and video ads allowed us to explain our value proposition and maintain our “stickiness”. Each channel will drive different value; try different formats to find the best performing one(s).
Step4: Pre-install and Post-install Optimization
The best way to optimize a campaign is to focus on only one metric at a time. For us it is the number of signups. Ruthless focus on the single most important KPI made our life much easier.
After choosing which KPI to focus on, we separated optimization efforts into two categories: pre-install optimization and post-install optimization.
Pre-install optimization focuses on evaluating your target (demographic, behavior, interest, etc.), messaging, ad creatives and so on. Because we had simultaneous campaigns with different channels, we had robust data for our targeting, creative and ad copy. We constantly collected data from each of them, created A/B tests and optimized each network accordingly.
During our conversations, we realized many companies do pre-install optimization but not post-install optimization, though we believe it’s equally important.
Post-install optimization has been available for a short period of time due to the advanced tracking technology of App Acquisition Analytic Tools (e.g., Adjust, Kochava, Amplitude, Branch.io etc.). We used a couple of tools to measure our key post-install events (KPIs). I highly recommend Branch.io, a very strong tool for deep linking, where you can tailor fit your conversion funnel for your users.
As mentioned, the number of sign-ups was the most important metric for Slice. We were able to see our sign-up/install ratio by ad network and even by publisher level (the ad networks’ sources of traffic). Once we had this data by publisher level, it was much easier to pick the right ad networks and the right sub publishers within the ad networks. Most of the time ad networks wouldn’t share sub publisher info but you can either ask for publisher IDs (so that you can’t see the actual name) or you can simply sign an NDA depending on your relationship.
On the other hand, in general, ad networks work with a fixed cost-per-install value (CPI). When both parties are transparent, the best way to buy ads is to imply dynamic pricing. In other words, changing the ad price automatically depending on post-install KPI to increase traffic when you have higher than normal post-install conversions.
After one cycle of selecting and optimizing ad networks, you should update your list of ad networks (increase the budget or fire them), update your user study based on what you learn from all the networks, look for new ad networks and restart the cycle.
To sum up, I tried to explain the benefit of introducing a healthy mix of multiple networks instead of focusing on only a few of them while spending on performance marketing.
Feel free to reach out if you have questions or like to talk more about it.
Happy user acquisition!
