$100 with audio app KoeKome and free traffic from BAS

The case study was provided by iMonetizeIt partner as part of “Case study Contest”

Network: iMonetizeIt
Offer: Dating Smartlink
Geo: Japan
Period: 03/02/2021-03/03/2021
Costs: $0 (no costs, except a proxy for a month for BAS is about $20)
Traffic source: koetomo apk
Installations: 100 000+

According to the classics of free traffic sources, firstly I checked all possible activities: likes, subscriptions and messages. I’ve chosen the messages, since the link in the description is not clickable.

We launch the MEMU emulator and use Charles Proxy to analyze all the necessary actions. The template will look something like this – registration, uploading a photo, men’s parsing, sending a message + voice.

Login:

Sending a message:

[[SCAN_AUTH_TOKEN]] – a token that is parsed from the login response and is necessary for sending a message

[[SCAN_MY_ID]] – my ID, which is parsed from the login response and is necessary for sending a message. It’s static, but changes when you use a new account.

Test of mail out in 1 flow:

Well, clearly, there are almost no visits from iOS devices, although Japan is like a big fan of Apple gadgets. The test showed that the admins started to fix something and the links don’t get to IOS.

Android is working properly, the link is clickable:

On iPhone, any clickable link is usually not shown or clicked, and it looks like a blank message:

I got traffic, I get conversions, but also we need to improve the template

Creating a parsing of men

We send a Get request for a profile of user and see in a response the line s** is gender, 2nd is a woman, 1st is a man. We create a regular with “if” and save the men in a separate txt file.

Regular expression, content of response + if(gender = 1) – save to the file.

This application doesn’t allow multiple users to log into one account and we won’t earn much in one flow, also users would be parsed for a long time. So you need to farm the database of accounts, make a registration as a separate template, register 50-100 profiles and quickly parse everyone with the database.

Registration:

We create a resource with randomization of “auth_token” and “device_id” using a string generator, there is no need to confirm email, so we create a resource with random email. Then analyzing registration:

This is how the “auth_token” and “device_id” generator looks like

After that, we translate emails into url encoding, make Hanami nickname randomizer + two random numbers + one random letter.

Registration is successful! With the help of “Does the string match the regular expression?” we parse the answer after registration and if the email is placed there, we save it to the database.

If the registration fails, there will be no email in the response and the email won’t be saved in the database.

Then we are making a special launch – 50 successfully registered accounts will be enough for us.

The same profiles were used for spam, so we will immediately upload the avatar. The photo is uploaded into the application via the Amazon cloud, we analyze the request, get the directory with our photo and simply implement two requests:

– Get request to the cloud

• Post request to upload photo

In the request, we won’t indicate the resource with the photo and simply take the name of the file from the cloud directory.

The registration is ready and the parser is also ready. We registered a database of 50+ accounts for parsing in multi flow. This is how the finished file looks like:

Check the parsing in a multi flow, our goal is the performance of each flow without crashes:

Conclusion—everything works. 55 flows successfully parse people by gender and save “men” only(“1” in the log).

The parser is launched and works well. We just need to think of links on iOS devices and send voice messages.

First, let’s find packs of female voice messages in Japanese.

The site https://samplefocus.com helped me find them.

I found a voice message in which a girl at home says “I like you” in Japanese.

Let’s analyze the request.

This is a way sending of audio looks like, but I can’t send a voice. That’s why in the MEMU settings, I specify a dedicated microphone.

After I start recording, I just bring my phone up to the microphone of my laptop and play the voice.

Checked it and listened to it on the device—everything is ok. We look at Charles, see the same picture as upload photo – the Amazon cloud.

Again, we make two requests:

• get to the Amazon cloud, to the directory of our voice recording
• Post to send a message with the name of this voice from Amazon

While we are waiting for parsing, let’s finish the template for the newsletter. Adding a token generator, device ID, etc. to it, just like in parsing. This is necessary for work of multi flows.

Also, we add parsing of user nicknames, in total they will get a message from the profile with a photo + creative with their nickname like “Tanaka, hello, I like you” + voice message.

The problem of links on iOS devices was solved by bold text:

In the highlighted message, there is an empty space after ください — the link is clickable, it will be displayed in Android and users can click it. In the end of the message, there is a bold link that iOS users will see.

Yes, Android users will see two links, but there is simply no better decision.

I added a test sending to my verification account, it’s a message every 1000 spammed ones in order to control the situation.

We launch the template, the log looks great – everything works good, show the ID, nickname and server response.

Notifications are sent to the test account:

Statistics per day:

A day later, the admins began to ban the proxy, blacklisted the domain and did something with the ID device (it doesn’t accept random ones now).

I don’t know if this traffic source is still work. But I think that beginners will pick up something in technical terms.