Is the information flow promotion effect not good? Maybe you overlooked this!

Therefore, one of the basic skills of optimizers is to understand the monitoring principles of third-party monitoring. Not only can it better control data, but it can also effectively identify fraudulent traffic and media.

Let's start with a series of soul-searching questions, and let's feel it:

Okay, with these questions, let’s analyze them one by one.

1. The principle of last click attribution (lastclick)

How to determine which media an activation is attributed to? Most manufacturers and third parties now use the principle of attributing the last click before activation, i.e. lastclick, which is the most basic monitoring principle .

Here I would like to quote Professor Liu Peng’s article on his public account “Computational Advertising”: “Is Facebook’s advertising really that effective?” 》 to introduce you to "attribution" with a paragraph of explanation and a picture:

The principle of "attribution" is as follows: when a click occurs, in addition to jumping to the landing page, a click link is sent to the attribution monitoring party (usually a third-party product), and the user's ID is sent as a parameter; when a conversion occurs, a conversion link with the user ID is also sent; the attribution monitoring party matches the user IDs in the two processes to know which channel brought about a certain conversion.

The last click attribution principle tells us two things: First, media activation attribution is determined based on the click behavior. If you want to check whether there is a problem with media monitoring, you actually need to test whether the click data with correct parameters can be received normally after the ad is released. If the click data is lost, all ad displays will not be monitored for any effect.

Another thing is that activation data is attributed to the media that was last clicked before activation. When advertisers place advertisements on multiple media at the same time, users may be exposed to product advertisements in multiple dimensions, but a user will only be attributed to the media that was last clicked.

For example, there is an activation now. Before activation, you first click on the ad of media A, then click on the ad of media B, and then click on the ad of C. Finally, it is activated. It depends on which media's ad was clicked last before activation. The one in the example here will be activated for media C.

2. IP+UA fuzzy matching oridfa precise matching

There are two types of clicks for media transmission: one is the precise match of IDFA . IDFA is a 36-bit identification number, which is a relatively stable unique identification number currently used by iOS devices (it will only change when flashing the device or manually changing the ad tracking code in the system).

When the user of a device clicks on an advertisement on an advertising medium, the advertising medium will transmit the idfa of the device that clicked the advertisement to the monitoring party, and match the activated idfa with the idfa of the media click. If the idfas are consistent, the advertisement will be attributed to the media (of course, it must also comply with the rules of the last click).

This is currently the most accurate matching method, but since the iOS 10 update, Apple has launched an anti-tracking function. If the user clicks "Turn off ad tracking" on the iPhone, the idfa of these devices with tracking turned off will all be a string of "0000" identification numbers, and the idfa matching will become invalid (in addition to adding IP, UA, fingerprints and other captureable information for further matching, the author has not yet come across a more effective solution).

The other is the IP+UA parameter . That is, when the device number cannot be transmitted, the IP and UA (UA refers to the relevant information of the mobile device such as the mobile phone model, etc.) when the ad is clicked are matched with the IP and UA at the time of activation. If the IP+UA are consistent, it is attributed to the media. (This method is somewhat similar to the identification logic of Shangdun promoted by Baidu )

Of the two monitoring methods, IDFA is more accurate because the device's IDFA is relatively stable. Even if it is changed, it cannot be attributed to the media, and there will be no multiple attribution that may cause misjudgment of advertising. In contrast, IP+UA monitoring is not very accurate. As we all know, IP changes greatly and cannot be unique. If the same device cannot connect to the wifi or switches to the 4G network, the IP will change. In addition, the IPs of countless devices connected to the same wifi are the same. Therefore, IP+UA monitoring is relatively inaccurate.

If you have a choice, try to use IDFA monitoring. A lot of cheating traffic is generated under IP monitoring.

3. Click validity period (or click window period)

As mentioned in the first point above, the most basic principle of ad monitoring is the principle of the last click. So how long before activation should one click on the ad to be considered as activation brought by the ad? This concept is the click validity period.

The core logic of this concept is how long an advertisement affects users, and how long it takes for the user's activation behavior to be considered effective after clicking the advertisement .

Generally, the validity period of an IP address is set at 24 hours or 12 hours because people are constantly moving; the validity period of an IDFA address can be set relatively longer, usually 7 days, 28 days, 30 days, etc. Facebook's validity period is set to 28 days.

Of course, the longer the validity period is set, the better the advertising media’s effectiveness data will be. Therefore, before the advertisement goes online, the window period should be set as reasonably as possible. Otherwise, it will not only be factors such as materials, media, and click-through rate that will affect the advertising data.

I would also like to remind you that if you find that the data of a platform is particularly good, it does not necessarily mean that it is really effective. Maybe the validity period you set is too lenient to it...

4. Activation validity period (or activation window period)

As we all know, the life cycle of mobile APP users is relatively short. If a user uninstalls and reinstalls the APP, is he considered a new user? If not, is he considered a valid user? How do we define this? This leads to the concept of an activation validity period, that is, a user's uninstallation and reinstallation within a certain period of time is counted as only one validity period, and uninstallation and reinstallation beyond a certain period of time is counted as one validity period, and this time is the activation validity period .

Obviously, the shorter the activation period, the better the advertising effect may be, because multiple uninstalls and reinstalls within a short period of time will still be counted as valid. Most advertisers set the validity period to 90 days, some set it to 30 days, and some set it to a longer permanent period, which means that a user is only counted as valid when activated for the first time, and subsequent uninstalls and reinstalls are not counted as valid.

As shown in the above picture, if the validity period is 30 days, the device will be counted as valid 4 times during the period from March 1 to June 1; if the validity period is 90 days, it will only be counted as valid twice during this period; if the validity period is permanent, it will only be counted as valid once.

After clarifying the above four points, everyone will have a basic understanding of the basic principles of advertising monitoring. Then everything is ready, only the east wind is missing. The next thing to do is to identify fake traffic and fight against the black-hearted media!