Other Data Issues

1. eCPM related

(1) How is the eCPM returned by the networks determined? Why is the eCPM price so low?

The price of eCPM depends on whether the advertising resources of the networks match the traffic, and their judgment of user value.

 

(2) Why do different products on the same network, or different ad sources of the same type of products, have different fill rates and bid price?

It is normal for the same network to have different code fill rates for different products or products of the same type.

The budget and resources of the networks, product quality, user quality, reserve price setting, user conversion effect on the advertisement, etc. will all affect the fill rate. So the fill rate and bid price will be different.

 

(3) Does click rate directly affect eCPM?

There is a certain correlation between the two, but not absolutely. eCPM is determined by the advertising conversion rate of the networks.

 

(4) How to increase eCPM and revenue?

 

2. Impression Related

(1) Why is the impression rate so low?

You can troubleshoot placement impression rate issues based on the Taku background report - Funnel Report (click to jump: https://app.takuad.com/m/report/funnel ).

Focus on the following funnel report, the number of times per capita in the funnel process by placement dimension, and the troubleshooting ideas.

 

(2) Does low impression rate affect revenue?

Impression rate calculation formula: Impression rate = impression/load of filled ads.

The level of impression rate is affected by the request strategy, repeated request strategy, the number of network connected, and the number of advertising sources. A decline in impression rates does not necessarily result in a decrease in impressions per user.

However, when the display rate is too low, you need to pay attention. It may be judged as low-quality traffic by the network, thereby lowering the bid and reducing ad filling. It ultimately affects the eCPM of ad settlement and the ad fill rate at the time of ad request.

Therefore, it is recommended to try to keep the display rate in a high range, not less than 10%.

 

(3) Why do low-priced ad sources have higher impressions than high-priced ad sources?

The number of impressions of an ad source is related to both the ad fill rate and eCPM, and is not only affected by eCPM.

For example, only two ad sources A and B are configured under the placement of the application, the fill rate of A is 1%, and the eCPM of A is 1000.

B’s fill rate is 100% and its eCPM is 1.

A's eCPM is higher than B's.

The application has 100 ad display opportunities. When requesting ads 100 times, because the fill rate of ad source A is 1%, ad source A only successfully fills 1 time, and gets 1 display due to the higher price; at this time, the revenue is 1 display out of 1,000 displays, which is 1 yuan.

Ad source B was filled every time. Except for one impression opportunity that was obtained by A, the rest were displayed, a total of 99 times.

It can be seen that ad source B, which has a lower price, gets more impressions due to its higher fill rate.

 

(4) Taku’s advertising display logic

After the client initiates a traffic request, Taku SDK will request ads according to the waterfall flow sorting price from high to low. After the bidding ad source inquires about the price and obtains the price, it will be inserted into the waterfall flow and sorted together. Finally, the ad with the highest sorting price among all returned ads will be displayed.

 

(5) Why is the bidding winning rate high but the ad source impression rate low?

Tips: More information about header bidding >> Click to view

Taku's bidding logic is like this. Waterfall starts requesting and at the same time inquires about bidding.

Assume that the price returned by bidding at this time is 20, then it will participate in the sorting in the waterfall.

1. At this time, Taku SDK starts requesting from the highest layer of the waterfall header downwards. For example, the waterfall header layer is 25. Then, if the header layer 25 is returned, Taku will request the advertising source of 25. At this time, the Bidding platform will be defeated.

2. If the previous waterfall has not returned and has been requesting until the bidding of platform A with a price of 20 is returned, we will initiate a request for the bidding of platform A and request the advertisement of platform A for 20, but if we are requesting the price of platform A When it is 16, the previous high-priced advertisement 25 is returned, such as an advertisement worth $25, or any advertisement higher than 16 in the waterfall flow.

At this time, Platform A is considered the winner, but Taku SDK will choose to display ads with a higher price of $25 in the cache.

Brief summary: In fact, it is best to compare the display rate of ad sources with the same platform price in the conventional waterfall flow near the inquiry eCPM (that is, approximately equal to the inquiry eCPM).

For example, the inquiry eCPM of Bidding on platform A is 20, and the display rate is 20%. If the display rate of platform A's regular waterfall flow 20 layers is also 20%. Then, the display rate can be considered reasonable. (Note that this should not be compared with the display rate of platform B or platform C. It is also recommended not to compare with advertising sources whose prices are too different)

Note: Taku will optimize the statistical caliber in version v5.9.71 and above. Subsequently, the winning notification will be sent when the request is actually initiated. It is expected to increase the display rate of the Bidding platform after winning.

 

3. Permeability Related

(1) Why is the penetration rate of the same app so different between Android and iOS, with Android’s penetration rate being 50% and iOS’s being 80%?

First of all, please make it clear that even for the same app, the advertising call logic of Android and iOS is different. Even if the technical call logic is the same, the user quality and user behavior of Android and iOS are different. So there will definitely be differences in penetration rate. For example, the user splash penetration rate of a certain app A on Android is 50%, while the penetration rate on iOS is 80%.

Secondly, when looking at the penetration rate indicator, it is recommended to analyze each placement separately and analyze the penetration rate of each placement and the reasons for its high or low penetration rate.

For example, the splash penetration rate of Android users of a certain application A is 50%. To increase the penetration rate, it is recommended to first communicate with the developer's product team and client technology to understand in detail how the splash calling strategy is designed.

If Android app A is set up not to show ads to 50% of users, then this strategy is reasonable.

If the Android app A is set to display splash ads to all users, then the 50% penetration rate is indeed lower than the designed product strategy of nearly 100%. After verifying this strategy, there are two directions for investigation.

First, check user behavior and user quality.

Second, according to the Taku backend - Funnel report, filter the placement, and based on the funnel model, check the link of user loss and confirm at which step the user is lost, resulting in low penetration rate.

 

(2) Why is the permeability so low? What is the normal permeability?

First, please read and refer to 3.1

The penetration rate is primarily determined by the strategies involved in the product and the technology used to call for advertising display.

Generally speaking, from the perspective of increasing advertising revenue, the recommended penetration rate on the aggregation side is definitely as high as possible, but too high a penetration rate will affect the user experience.

Therefore, there is no 100% correct penetration rate standard. Developers are advised to design reasonable advertising displays based on their own application categories and advertising scenarios.

 

4. Ad Filling Related

(1) How to improve the fill rate of splash ads?

Since splash ads are ads that pop up when users open an app, it is very important to increase the speed of loading splash ads. Developers are often troubled by the low fill rate of splash ads. If you want to increase the fill rate of splash ads, you can adjust and optimize it as follows:

 

(2) What factors affect the decline in eCPM and fill rate?

The fill rate and eCPM of each network are determined by its own advertising delivery algorithm. Generally, there are two factors that affect the fill rate or eCPM of each layer of ad source:

Network budget reduction: If the ad budget for your app is reduced, the fill rate and ecpm may decrease (this is affected by the overall market of the network, and is generally higher during holidays and weekends)

[Solution]: There is no way to decide this. You can only achieve resource complementarity by connecting to multiple networks as much as possible; or communicate more with the networks to see if optimization is possible.

Decreased user quality : Poor user conversion results may cause networks to be unwilling to put more advertising resources, or higher-priced advertising resources, into your app, resulting in a decrease in fill rate and ecpm.

[Solution]: This can be improved mainly by improving the quality of users acquired on the delivery side; or by creating a new network placement ID to see if there is any improvement.

 

(3) What determines ad fill duration?

The ad filling time is related to platform materials, response speed, user device performance, user device network conditions, etc.

The filling time in each region is different. For example, the network condition in T1 region is generally better than that in T3 region, so the filling time in T1 region will be faster.

 

(4) Why is the fill number of an ad source always much larger than the impression number?

• Impression rate = number of impressions/number of fills. Ad filling is at the application level. As long as the user opens the app, they will initiate a request to all placements (ad preloading), and then the placements will be filled. Therefore, the number of fills will be affected by the number of network connected and the number of PIDs.
• Ad impression needs to be triggered by the user. The user must enter the advertising scene and trigger the ad before the ad will be displayed. Therefore, the number of fills of the ad source is always much greater than the number of impressions.

 

(5) How to improve the fill rate of placements?

Due to factors such as the network, the time it takes to fill an ad, and the bid of the network, the ad fill rate cannot reach 100%. Generally speaking, a placement fill rate of more than 85% is excellent. If you want to improve the placement fill rate, you can refer to the following methods:

 

5. Preload Related

(1) Recommended preloading timing for each ad type

The so-called preloading is to call the Load method of Taku in advance to cache the ads locally to avoid the situation where the ads are not loaded when they need to be displayed, thereby increasing the display success rate.

① Preloading is recommended for all ad types. (Starting from Taku V5.7.6, splash ads also support preloading)

② If there is a splash ad, it is recommended to load the splash first, and then load other ad types.

③ If there are many advertising scenes in the app, it is recommended to load the advertising scenes that are easily triggered by users first, and then load other advertising scenes.

④ If multiple Taku placements are preloaded at the same time, it may consume too many mobile phone resources and affect the user experience. It is recommended to arrange the preloading timing reasonably.

⑤ The loaded ads will have a cache validity period, which varies from one network to another, but is generally half an hour or one hour.

 

6. Waterfall Related

(1) If the waterfall is too long, will it cause the ad to return more slowly?

If the waterfall flow is too long, it will affect the return time. It is recommended to decide whether to optimize the waterfall flow length based on the ad ready rate.

If the ad ready rate decreases, you can appropriately reduce the number of waterfall layers, or place ad sources with shorter return times at the head of the waterfall.

 

(2) Why is the number of requests for Bidding less than the number of inquiries/tiered requests?

Tips: More information about header bidding >> Click to view

Taku header bidding principle: Each time an ad is requested, the header bidding ad source will be queried for price, and the returned price will be sorted with the price filled in when creating a regular ad source. Ad requests are made according to the sorted waterfall priority.

Since Bidding participates in the waterfall sorting of regular ad sources based on the price returned each time, if other high-priced ad sources in the waterfall have ads returned, the Bidding ad source will not be requested.

Therefore: Therefore: the number of [Inquiries] for bidding advertising sources is close to the number of [Ad Requests] for waterfall flow high-price tiering, and the number of [Requests] for bidding advertising sources is generally less than the number of [Requests] for waterfall flow high-price tiering.

Note: The reason why the number of Vungle Bidding inquiries and the number of header layer requests are very different is that Vungle Bidding will cache the ad for half an hour after returning it from a query. It will only send out another inquiry after it needs to be displayed or after there is no cache for half an hour.

 

7. Segment Related

(1) Will too many segments reduce the speed of ad requests?

If a user hits multiple segments, Taku SDK will return the highest priority segment for waterfall flow configuration, and the request speed will not be affected.

 

(2) If a traffic request does not hit the highest priority group, where will it flow to?

If the traffic does not hit the highest priority group, it will automatically flow to the lower-level segment and finally to the default group.

Therefore, it is emphasized again that the default group, as a fallback group, cannot be closed and a code position must be built to ensure that traffic is not wasted.