SOLUTION
Advanced Measurement solutions provided by Kochava enabled an unbiased and data-driven assessment of the campaign’s incremental lift. To begin, the OTT/CTV platform provided over 2 million hashed primary IP addresses – 1.7M of which belonged to households in the ad exposure group, with the remaining 300K in holdout as a control group. A primary IP/household identity graph provided a 66.1% match rate against both groups.

To confirm there was no bias between the exposure and control groups, Kochava data science analyzed household devices within each group using proprietary device scoring, which measures attributable qualities like activity recency & frequency, geo visitation, etc. The device scores were nearly identical, confirming no bias.

IMPACT
With a 90% confidence interval, Kochava showed the campaign generated a significant incremental lift, resulting in over 4,800 purchases. Campaign exposure was also clearly shown to influence the timing of app installs with the greatest likelihood occurring within 4 hours of exposure to the ad. Ideal impression frequency for the household was 3-10, with less than three under-performing and more than 10 showing no benefit.

PROBLEM

A worldwide telecommunications and financial wire service company was concerned about performance anomalies in their mobile app acquisition campaigns. Certain key performance indicators (KPIs) seemed over inflated, and post-install engagement with important conversion events was less than desirable. For the marketing team, understanding their true return on ad spend (ROAS) was elusive and they were deeply concerned about ad fraud.

SOLUTION

The company hired Kochava to perform a fraud audit on their attribution data. The primary focus was to analyze click and install data attributed to all ad networks for two of the company’s apps on both iOS and Android platforms. The goal was to uncover any fraudulent activity impacting their campaigns. To determine fraudulent activity, Kochava leveraged their internally developed fraud detection algorithms to process the onboarded attribution data from the company. Kochava data science and client analytics teams performed a detailed analysis and consultation, delivering a comprehensive report of the findings.

IMPACT

The networks analyzed demonstrated varying levels of fraudulent activity ranging from fewer than 15% to over 85% identifiable fraud. Overall, 65% of all attributed installs for the time period were identified as fraudulent. The primary fraud tactic overwhelmingly utilized by the networks was click flooding. By flooding large numbers of fake clicks into the ecosystem, certain networks had an unfair advantage in gaining last-click attribution, not only stealing credit from other legitimate network partners, but also taking credit for truly organic installs. A significant volume of installs further exhibited behaviors that suggested they were manufactured and not real user installs. The company took swift action on the findings and their rates of fraud fell considerably as they stopped running with the networks with the highest rates of fraud.

SOLUTION
The gaming company enlisted Kochava to perform an independent fraud analysis on attribution data captured by their internal system for a specific app over the period of one month. To do this, Kochava used their internal fraud detection algorithms that focus on outlier detection and pattern identification. Click and attributed install data from the app publisher were used to establish baseline norms. Kochava data science and client analytics teams performed a detailed analysis and consultation, providing a comprehensive report of the findings.

IMPACT
The Kochava Fraud Audit found many measurable quantities of fraudulent transactions. Click flooding accounted for the largest volume of false attributions at over 60% of all installs analyzed, followed by time-to- install (TTI) outliers, and click-to-install geo outliers. As for clicks, over 90% of all the clicks analyzed were flagged as fraudulent. In total, over 190,000 misattributed installs were reported—driven by over 400 million fraudulent clicks. The company was able to take action on the findings and swiftly cleaned up their media mix.

SOLUTION
The DSP requested that Kochava perform an independent fraud audit for its client. Kochava worked with the client to onboard their raw user engagement and ad campaign data from their existing MMP. A thorough processing of all traffic against Kochava’s fraud detection and prevention solutions was executed. Kochava data science and client analytics teams performed a detailed analysis and consultation, providing a comprehensive report of the findings.

IMPACT
The Kochava Fraud Audit uncovered serious fraud violations by the newly added partners, with click flooding accounting for the largest stake of stolen conversions. Over 49% of the installs that the existing MMP attributed were matched to invalid clicks or impressions. A majority were organic installs that the client falsely paid for due to organic install sniping, while another large portion should’ve been rightfully attributed to their primary DSP’s campaigns. Based on the independent nature of the audit, the DSP’s traffic was also analyzed and surfaced only a miniscule percentage of fraud, validating their own efforts to keep their premium supply clean. The client leveraged the audit’s findings to secure over $55K in make-goods and is now re-evaluating ad partners and their existing MMP vendor.