When we talk about mobile ad fraud prevention, we usually think of it in terms of financial theft.
Someone is taking our money, and only pretending to give us something for it.
This is likely to cause panic among Chief Revenue Officers and Chief Marketing Officers, who are prone to react by asking, "How much are we losing?"
Undermines Data Integrity
But looking at ad fraud in terms of dollars and cents minimizes the bigger problem of dirty data.
User acquisition fraud means the datasets marketers operate on — the data that they’re using to make decisions about where to spend their budget — can no longer be trusted.
Fraud reallocates the marketing budget to the wrong places — to the tune of billions of dollars per year.
While many mobile marketing teams have been addressing fraud prevention, it is time to recognize that most of us have been coming at it from the wrong direction.
When working with impressive names like Rovio, Zynga, Miniclip and so on, we’ve learned a lot about fraud prevention for user acquisition, and the needs of marketers when faced with this huge problem.
In a recent study, we found that three of the five most common sources of error in marketing datasets are caused by fraudulent activity, whether from a sub-publisher spoofing conversions for cost per install (CPI) payouts or from individual users that try to access content without paying for it.
In short, user acquisition fraud makes bad, underperforming or fraud-ridden channels look good, and by comparison, highly performing channels look bad.
Misattribution of Users
Current fraud schemes that affect advertisers either poach organic users, causing them to be attributed to a fraudulent publisher, or fake conversions altogether.
Organic users are normally more engaged than the average user acquired over even a high-performing advertising medium, like social.
So when the high organic engagement is poached, it boosts the average engagement of the channel to which it’s attributed. The apparent return on investment (ROI) on cheaper channels that are exposed to fraud thus increases, and the opportunity cost of investing in the right channels appears to increase.
This causes marketers — who trust their dataset with their life — to re-allocate more budget to channels that are affected by fraud.
Here, Not There
When we speak of the ad budget “lost” to fraud, we have to remember that it’s not truly “lost”.
It’s just money spent on the wrong channel. While this sounds like one in the same, actually when we activate our fraud filters, our clients don’t take the money they’re now not spending on the fraudulent channel and put it in their piggy bank.
They take that budget, and they spend the money in a new channel.
Ensuring Data Cleanliness
To effectively help your marketing team make better decisions, you need to look at things more holistically than just squashing the individual fraudster.
That’s our realization, and here’s how it should work:
- 1.Your fraud filters should detect anonymous IPs, and prohibit attempts at spoofing installs through emulating devices
- 2.You should refocus your distribution modeling to prevent poaching of organic installs
The result of both actions is that your install and post install metrics will be realistically allocated to organics, paid sources and untrusted devices. This way, your user acquisition manager can make confident informed decisions on their install and post install metrics that are now correctly aggregated.
As the advertising industry finds that fraud is more of an issue affecting data quality and accuracy, we’ll continue to see more focus on techniques and tools that approach this key problem.
At the same time, ad fraud methods continue to mature in this cat-and-mouse game.
The challenge for us will be to shift our fraud prevention strategy before more budget is spent in the wrong channels — and starting with the approaches we’ve mentioned above here is a step in the right direction.