Download campaigns – Utilizing your traffic
In this post I will focus on download campaigns (PPI) or in the professional term – bundle software, which is referred mainly to as the Download Valley. Download campaigns come in various shapes and skins and the common denominator between them are FLV Players, Free Codecs and several small downloadable games. The campaigns working on mass audience without any particular targeting except for the media type and the payout are great considering their deceptive funnel.
The download distribution scene is frequently in the headlines due to its notorious reputation and association with malicious software, adware, hijacking and others. But, in reality, there is a lot more to bundled offers than its reputation. The fact is that almost any free software websites are using some kind of bundle (additional software) monetization from Download.com to any OpenSource software out there.
Before rushing into optimizations and data collection you will have to understand in depth, on the tech side, what you are actually promoting. Understanding this will ensure you better utilization of your traffic and overall knowledge about our industry which is just as important.
To start, I will give you a quick overview about the Installer structure and the various software a singular distributor offers in one install process. This stage is crucial for the optimization process and the traffic shifting between campaigns.
Download campaigns have more than one event, in fact there are many and these are the key events you should better be familiar with:
- Started: Once a user opened the installer
- Complete: User finished the additional component installation process
- Pixel fired: User generated the expected revenue from the installation process
Any install process includes 1 Search Offer and several other non-search, side offers. If you download and install one of the campaigns you are promoting (I suggest to use any virtual machine for that) you will see between 3 to 5 screen offers and 5 to 7 check box offers. Anything additional you install might have an unknown number of offers under the term “silent install” which the user doesn’t know about and there is no compliance on them. Right after the installation you will get an open source software which commonly known as a carrier.
A random installation package:
1 PC Optimizer
1 White spice
Other components as DNS tracker, contextual, video, inText etc.
When a user downloads software that is wrapped with a bundled installer, the installer core’s first process will be to check the local machine, for known registry keys (registry keys – Wikipedia). The installer won’t offer or install a component if its registry key exist on the user’s machine or any other software that might cause a conflict in the monetization process. After the installer core has finished the registry key’s scan, it will show (or not show in many cases) the users any additional software available in its inventory. The common user will probably just press “Next” till the Install Button appears, and once that done he installed several software.
As a marketer that promotes download offers, you aren’t getting paid for an install–you are getting paid for an install AND IF the user generated minimum revenue to the advertiser. That means that the pixel the advertiser is firing back to you will be triggered only if the user completed the installation and met certain conditions.
Most advertisers won’t set a revenue condition but will focus on the search component as a condition to fire a pixel. This means that if the user completed the installation process and by that installed 4 additional component but skipped the search component, on purpose or because he already had it, the pixel won’t be triggered and you won’t get paid.
The solution – Saving and crossing user’s data
Now that you know about the installation process and flow along with the untold advertiser’s conditions, I’ll reveal the real numbers behind the download offers. The download “take rate” in download offers is 30%-40%. Only 30% of the people who will engage with your landing page will actually download the .exe file. The average success rate, from start to completion, is around 75% and the “take rate” of search components (if were avail to the user) is around 70% and below.
For every 5 completed installs, the average advertiser will fire only 2 pixels. This is partly due to the reasons I outlined above—but there are others. This means that even if you managed to bring users to your landing page, they downloaded the product, started and finished the installation process you will get paid for only 2 installs. One of the most important reasons is, a user that installed a download offer will most likely install a few more in the future, even in the same day.
This situation isn’t going to change but you will have to adopt a new way to maximize every impressions and install process. The solution is to cross data and play with different download offers to make sure you will get paid for each one of the installs.
First step: advertisers’ variety
Since the majority of advertisers lean on search conditions, the first step will be to find as many offers as possible with different search components. Protip–se a virtual machine to do so. You can start with the list below:
Second step: creating a database to cross installs history
You can use any sort of database you are familiar with, from SQL to files DB, everything goes. In the database you will need to save data to the following columns:
Unique ID – User’s unique identifier and his cookie value!
Country – User’s origin
Offer – The offer name or ID you promoted to this specific user
Download – If the user requested to download the file or not / Clicked on the download button in the landing page.
Pixel – Successful conversion on the user or not
Third step: going to production
For each time a new user visits your download’s landing page, you will need to set a cookie with a unique identifier as a value and write his visit and action to your database. Once they pressed the download button, request from your server to deliver one of the campaigns you chose earlier. If the user is a new user, simply deliver the highest paying offer (you can create per-scenarios with prioritizations). Write to your DB that if the user clicks on the download button and what exactly what offer the server delivered. Within 10 minutes (the average time) if the install process is completed and the user passed all the conditions, the advertiser should fire your pixel. If your pixel got triggered you may write in your DB if not, you will consider it as a fail. Either way the next time you will meet this same user you will have some data on them.
If you met the same user on the same day and they requested another file to download by clicking on your download button, you will deliver them another offer from your inventory with a different search component. It is not important if you already got a conversion today from this user or not, since you are offering a completely different search component from what recently was installed on his machine. If you didn’t get the conversion earlier, this is probably because they didn’t complete the installation or they already had the search component that offered in the first delivery.
You should repeat and play with the download offers you are delivering to your users until the 3rd time. If you offered 3 different search component to the same users and you got didn’t see a single conversion, the user’s potential is too low and you should redirect him to any other vertical instead or exclude them if possible.
Using this method I was able to increase my complete to pixel rate from 40% (2 for 5) in to 85% and maximize my impressions revenue potential. There is a lot of technology and data in this world, you just need to understand the map of things in each one of the offers you are promoting. There is no simple flow in any one of the campaigns, not in download campaign nor in lead generation. In any of them we can use data and technology to our advantage with one goal and one goal only, maximizing our impression potential.
Till next time, Michael