Menu

Mediation data

1. View the mediation (Waterfall) report data

Adjusting Waterfall needs to be done in combination with the actual data performance of each layer of ad sources. The Taku mediation page directly provides two sets of data: Taku SDK statistics and network report API data. There is no need to switch to the full report page to view and then switch back to make adjustments, which maximizes your convenience and saves time.

 

1.1 Report Time Zone

The data report supports modifying the report time zone. Currently, it supports UTC+8 China Standard Time, UTC+0 Greenwich Standard Time, and UTC-8 Pacific Standard Time. The default is UTC+8 time.

Taku SDK statistics and the following network data have been aggregated according to the selected time zone (Facebook, Mintegral, Applovin, UnityAds, Tapjoy), and the remaining network data are based on the time zone returned by the API.

Report time zone modification path: Taku Developer Backstage→[My Account]→[Account Information]→[Base information]→Report Time Zone

 

1.2 Data update time

The update time of different data metrics will be different, as follows:

(1) eCPM: eCPM is calculated by splitting the API revenue according to the impression ratio. Yesterday's data will be updated at 11:00 and 17:30 every day;

(2) Other Taku SDK statistics except eCPM: updated every 30 minutes;

(3) Report API data of ad platforms (we will pull network data of the previous day in multiple time periods), please refer to the update instructions of each network in the full report for details

(4) Latency Percentiles data updated the next day:

① UTC+8 China Standard Time: Updated daily at 6:00
② UTC+0 Greenwich Mean Time: Updated daily at 11:00
③ UTC-8 Pacific Standard Time: Updated daily at 20:00

 

1.3 Report Metrics

Custom metrics support checking the required data dimension content to impression in the aggregate management report.

Description of each indicator of the mediation report:

Report indicator Description
Priority The order in which Taku initiates requests to each ad source. Request in ascending order of Arabic numerals and descending order of priority.  Note: When the numbers are the same, the priority is the same. Each time a request is made, the priority of the ad source with the same number will be randomly requested.
S2S and C2S icons are Impressioned for bidding ad sources
Ad source The placement ID parameter of each network corresponds to a unique Taku ad source, which can be used for waterfall ad layering in mediation. >>More
Status Turn on or off the ad source. Blue is on, gray is off.
Operation Edit or delete the ad source.
Price Sorting price. This price is only used for priority sorting when requesting ads: for example, if ABCDE 5 layers of ad sources are added, these 5 layers of ad sources need to be sorted. This sorting will be sorted from high to low according to the filled-in price, and will not affect ad filling.  Note: ① If the sorting prices of multiple ad sources are the same, the priority is the same, and each request is made with a random priority; ② It is recommended that the price be as consistent as possible with the actual eCPM (i.e. eCPM API) value of the ad source, which can be used to estimate the real-time revenue and LTV of the day.
eCPM Taku calculates eCPM based on the total Revenue API( provided by Network's Reporting API ) and Taku statistics Impressions, calculated as ( Revenue API / Taku Impressions ) * 1000. Note: eCPM is delayed by 1 day.
eCPM API Taku calculates eCPM API based on the total Revenue and Impressions API ( both are provided by Network's Reporting API ), calculated as ( Revenue / Impressions API ) * 1000. Note: eCPM is delayed by 1 day.
Estimate eCPM Taku calculates Estimate eCPM based on the total Estimate Revenue and Taku statistics Impressions, calculated as ( Estimate Revenue / Taku Impressions ) * 1000. Note: 1. Estimated eCPM available on current day ; 2. Non-bidding ad sources are based on manually filled-in eCPM price, and bidding ad sources are based on real-time bid price ; 3. Meta is based on historical eCPM API
Estimate Revenue Bidding advertising sources use real-time advertising display prices to obtain estimated revenue, and non-bidding advertising sources use manually filled eCPM prices * Taku statistics display summary to obtain estimated revenue,Meta use Meta historical eCPM API* Taku statistics display to obtain estimated revenue.
Estimate Revenue Proportion Proportion of Estimate Revenue counted by Taku,Calculated as (Estimated revenue of the floor/Total estimated revenue of the segment)
Attempts/Bids Normal ad source is attempt, Header Bidding ad source is bid request. Header Bidding process and indicators
Bid Request Number of bid request from header bidding ad sources
Bid Response Bid response rate of header bidding ad sources(Bid Response/Bid Request*100%)
Bid Fill eCPM The average eCPM returned by the header bidding ad source
Request The number of attempts reported by Taku. The number of times an ad request has been conveyed from App to Taku. A given ad request can attempt multiple ad sources.
Normal ad source is attempt.
Header Bidding ad source is ad request. After the bidding ad source's request has a bidding response, the bidding ad source participates in waterfall sorting, and when the ad source's turn to request, the actual number of requests sent to networks.
Request API The number of ad request provided by networks' reporting API. Notice: Some networks don't provide this data
Fill Rate The rate at which an ad is served following a attempt to Ad source; calculated as (Attempt Fills / Attempts). After the bidding ad source's request has a bidding response, the bidding ad source participates in waterfall sorting. The rate at which an ad is served following a attempt to Header bidding ad source.
Fillrate API The fill rate provided by networks' reporting API. Notice: Some networks don't provide this data
Latency(sec.)

The filling time of the ad source indicates the average time from request to successful filling. The filling time of the placement indicates the average time from the loading of the placement to successful filling. Due to Unity Ads' self-filling behavior, the filling time cannot be accurately counted. The time-consuming data of Unity Ads's ad sources is not displayed temporarily. Starting from September 21, 2026, the filling time of S2S ad sources will include the bidding time, so the data will be higher than that under the previous counting method.

Impression Impression data has changed since 2023/3/7 00:00:00(UTC+8). When Taku get the callback of networks, it will be counted as an impression.More detail.
Impression Proportion Proportion of Impression counted by Taku,Calculated as (Impressions of the floor/Total Impressions of the Segment)
Click How many times the ads are clicked. Reported by Taku SDK. Some of the platforms do not support click callback so part of the click can not be tracked by Taku.
API Click The number of times an ad is clicked, provided by networks' reporting API
Click Proportion Proportion of Click counted by Taku,Calculated as (Clicks of the floor/Total Clicks of the Segment)
CTR Calculated as (Clicks / Impressions * 100%)
API CTR Click-through rate, calculated as (Clicks API / Impressions API)
Revenue API The total revenue provided by networks' reporting API
Revenue API Proportion Proportion of Revenue counted by Taku,Calculated as (API Revenue of the floor/Total API revenue of the Segment)
Impressions API The number of times an ad is shown, provided by networks' reporting API.
Impression API Proportion Impression API Proportion.
Impression Rate Calculated as (Impression / Fills)
Impression Gap The gap between Impression & Impression API, calculated as (Impression- Impression API)/ Impression API.
Click API Proportion Proportion of Click API counted by Taku,Calculated as (Click API of the floor/Total Click API of the Segment)
eCPR Revenue per thousand attempts, calculation formula: Estimated Revenue/Attempts *1000.

 

1.4 FAQ

(1) What is the function of latency?

Latency refers to the average time from initiating an ad request to the ad returning successfully. It can be used to view the loading time of each ad layer, and optimize the waterfall flow in time by combining the ad layer fill rate and ecpm.

Note:

① Long latency does not necessarily mean that the ad effect is poor, and it needs to be evaluated in combination with other data.

② Generally, video ads with a loading time of less than 10 seconds are not considered slow.

Previous
Header bidding
Next
Advanced Setting
Last modified: 2026-09-20Powered by