Sending money app advertising launch (money transfers)

Conversions

The most important conversion for the client was a “successfully sent transfer.” But there were complications with its transmission. The fact is that registration took place within the application, but the transfer itself opened in a webview, and in fact, the conversion was recorded on the website, not in the application.

At the start, we didn't have this conversion, so we launched ads optimized for registration in the app and installation. Together with the client, we selected a list of target countries where it is legal to operate and where the app is available, and compiled a list of languages. For example, we launched ads in France in Ukrainian, knowing that most people would be sending money transfers to Ukraine.

Ad campaigns setup

So, we launched campaigns: we compiled a list of countries and languages based on technical restrictions on sending money transfers.

There were two campaigns — for installations and for registrations. And the target conversion we needed (successful transfer) was not transferred to the advertising accounts. Therefore, we manually compiled daily and weekly reports: how much we spent on advertising, how many installations and registrations we received. The client provided feedback on the number of transfers, and based on this, we adjusted the target price for registration and installation, adjusting campaign expenses.

We regularly updated the banners. We made themed banners for holidays, events, and national dates in both the sending and receiving countries. Sometimes we ran promotions with free transfer fees.

After about six months, working together with the client's developers, we managed to set up a “successfully sent transfer” conversion. From that moment on, we abandoned campaigns focused solely on installations and concentrated on two types:
- the first — with the goal of registration,
- the second — with the goal of deep conversion “successful transfer.”

With this campaign setup, successful translations yielded the maximum number of target users, but the cost per registration was high. Registration campaigns, on the other hand, yielded cheap registrations and volume, but the quality was lower (lower conversion to translation). Nevertheless, both types of campaigns made sense. As a result, we allocated 70% of the budget to “successful translation” campaigns and left 30% for campaigns aimed at “registration.”

We made a separate observation regarding social networks. First, we tried to reduce the attribution window to 1 day after the click, abandoning the standard Facebook window (7 days after the click and 1 day after the impression). The campaigns worked, but the results were worse. When we tested the standard attribution model, it performed better. It turned out that users needed time: they installed the app, registered, figured out the interface, and only made a transfer after a few days. In other words, on the first day, they simply did not have time to complete the target action. As a result, Facebook's standard attribution (7+1) helped us a lot — within 7 days, 99% of users either made a transfer or deleted the app, and we covered all the necessary conversions.

Key points

If your conversions are not the simplest (not just a click or an application), but more complex — especially those related to applications, installations, and registrations — it is better to use the standard attribution model recommended by Facebook.

In this case, the key issue was analytics and the correct transmission of conversions. For the first six months, we were unable to transmit the “successful transfer” event, so we optimized for registration, but the traffic was of low quality and we were unable to scale up. After the desired conversion appeared, the situation changed dramatically: we were able to increase the budget more than threefold and allocate 70% to campaigns optimized for “successful translation.”