2016-12-08 | Article written by David Ismailov

How to Evaluate the Quality of Traffic in Mobile Marketing?

What is nom-incent and incemt traffic, and how are they different? Wha payment model is more profitable: CPC or CPI?

When it comes to the promotion of mobile applications, it’s important to properly understand these issues and adjust their monitoring to bring the highest quality traffic.

So, mobile traffic comes in 3 kinds: organic, advertising (non-incent), and incent (where users pay for the install):

  1. Organic Traffic - installing applications from Google Play/App Store by searching them or from recommendations of the app stores.
  2. Non-incent Traffic (advertising) - installs occur by the classic advertising model: users notice an interesting ad about an app and proceed by clicking on it.
  3. Incent traffic - users are rewarded for installing an app. Frequently, the application itself doesn’t interest them. This traffic is used to get the app to the TOP.

You can buy mobile traffic by a number of payment models, such as CPC, CPI, CPM.

How do CPI networks work?

For example, an advertiser appears on the market and orders 1000 installs/day at the price of $2 per one in the CPI network. The CPI chain accepts the order and gives it to its partners (with the deduction of the commission of 30%, hence it’s now $1.4).

In turn, many of the partners buy incent traffic for $0.4 from their partners, and then pay $0.2 remuneration to their implementers.

So what happens?

The advertiser receives 1000 installs at $2/install, but the majority of members that have been involved have no further interest in the use of the app. (At the same time, the first-level partner received $1,000, the CPI-Network received $600, the second-level partner and all the implementers received $200).

If you are using the CPI-model, be prepared for some difficulties:

  1. 1. Mobile analytic systems overload.
    When buying large volumes of traffic, the analytic systems may miscount the number of installs, which leads to a 20% - 30% loss in the tracker. The traffic supplier is usually blamed for this, although it might not be his fault at all. Trackers are not always accurate when determining the quality of traffic (incent or non-incent).

  2. 2. Fraud.
    Fraud - adulteration of incent traffic as non-incent, click-fraud of advertising by robots, which does not lead to further installation, and other similar actions.
    The best way to deal with it is to correctly configure the analytics and to analyze traffic on a regular basis. If someone supplies poor traffic, you can see it in the metrics and turn it off.

  3. 3. Continuous work with the app’s internal analytics.
    To understand how much quality traffic a particular source give you, you need to keep track of the internal analytics of the use of the app by users: when the use the app - in a week, after 30 days, implementation of required targeted action, making payments, etc. Only on the basis of these parameters and the comparison of these parameters between sources you will understand the real quality of each source.

Let’s finally talk about the differences between incent and non-incent traffic!

Using appropriate analytical services, such as Adjust or Mixpanel, advertisers need to track behavioral indicators of the users:

  1. 1. LT - time the users spends in the app.
  2. 2. Churn Rate - user outflow rate.
  3. 3. ARPU - Average Revenue per User (for applications with in-app purchases).
  4. 4. Does the user use the app or not.
  5. 5. Are the target action(s) performed (registration, purchases, number of searches, etc.)

On the basis of these parameters with sufficient probability, you can tell if the traffic is high-quality and alive.

CPC or the CPM-model?

If you are not confident that you can control the attracted installs sufficiently by CPI models, it’s better to use the old-fashioned schemes to purchase traffic - CPC or CPM. When fine-tuning the analytics you can attract more quality traffic and optimize the cost of installs due to the more specified targeting.