Enhancing Fraud Detection through Advanced Large Data Analysis

i-mobile Co.,Ltd. provides an advertising delivery system [i-mobile] to internet used on smartphones and PCs. Launching the foremost ad network on mobile devices in the industry and currently operating the most premier service in Japan, we look at the reason why i-mobile has decided to introduce Spider AF. Ad Platform Business HQ. We talked with i-mobile’s Chief Executive Officer, Mr. Kai.
Table of Contents
Ad Platform Business HQ Chief Executive Officer, Yasuhiro Kai

Q. Please tell us about your company’s background and business.

We operate ad networks for smartphones and PC. We possess a lot of advertising inventory from the biggest and most successful web companies out there as well as hold the largest delivery inventory in Japan with banner native advertising.

Q. Please tell us about the problems/issues you were having before you introduced Spider AF.

Before we introduced Spider AF, we were handling fraud detection for impressions, clicks, and installation during ad deliveries.

However, since we handle a large number of web companies numbering in the hundreds of thousands, there was a lot of work cost in doing fraud detection and efficiency was becoming a problem.

Q: Did fraud detection take a lot of time?

It did indeed.

Extracting the data, processing/analyzing/organizing, reporting, and then finally visualizing everything took about 3 days. The total number of work hours was more than 8 hours. Because there is a lot of data that needs to be extracted and because there are multiple focal points that need to be tested, it takes up time.

Q. What kind of expectations did you have for Spider AF prior to the introduction?

We were looking for a service that could detect fraud with efficiency and high accuracy.

Q. Among the services provided with Spider AF, what was the deciding factor that caught your attention and made you want to go with Spider AF?

The deciding factors were being able to analyze a large number of logs efficiently and the expectation that accuracy can be improved by learning from the fraud results.

We first gave it a try and introduced it to our company. It lived up to our company’s exact specifications and helped us out a lot.

With each repeated result, the fraud detection accuracy improved, and now it can detect fraudulent media with high accuracy.

Q. When you were introducing Spider AF, did you have any difficulties/concerns?

There weren’t any particular difficulties.

When we introduced Spider AF, we were concerned about knowing how necessary the accuracy of fraud detection and work/learning costs would be, but soon after those concerns went away.

It was because it was highly accurate, and even inexperienced members could use the dashboard and prepare a report.

Q: How did you overcome them?

It’s all thanks to the fact that the service adjusted to our expectations and was able to respond flexibly.

Q: How many people actually introduced Spider AF and gave it a try?

Including members who were involved in the fraud response project, technology, and sales members – 10 people used it.

Q: Did it solve the problems you had before (prior to Spider AF)?

It almost completely eliminated the work time from the technology side of things.

Q: How do you use Spider AF?

At i-mobile, we conduct fraud detection in real-time when an ad is distributed.

Efforts have been made to exclude those determined to be fraudulent at the time of impressions and clicks. We use Spider AF for detecting/patrolling against fraudulent activities on our companies registered web companies based on delivering results.

Q. We would like to ask you about the support you received before and after introducing Spider AF. How was Phybbit’s support (engineers/sales)

They were very kind and polite when giving answers to our concerns. It helped us out.

Q: Were there places that you wish there was a little bit more support?

As a request, being able to download the report from the dashboard, and I think that if the Japanese response improves, the usability will improve. I am expecting that kind of support in the future.

Q: If there was one thing you could tell those thinking of introducing Spider AF/companies who are considering Spider AF, what would it be?

Spider AF is a fraud detection system with high accuracy and flexibility. It allows you to customize the scoring criteria and corresponds to detailed requests.

From a cost perspective and since it is easy to introduce, I recommend this to any company thinking of fraud protection!

Q: Finally, could you tell us about your company’s thoughts on ad fraud and how you will tackle the problem of ad fraud in the future?

For gaining a reputation in providing correct advertising results as an ad network, preventing fraudulent use is indispensable. We will work as an overarching mission to build an environment where advertisers and web operators can use our services fairly and with peace of mind.

In terms of industry, I would like to work with other companies in concert to improve the quality of the ad network (including the restriction on advertising that may trigger false clicks).