5 cold thinking on digital advertising monitoring
The rapid development of the mobile Internet, fragmentation, digital advertising surge, the market urgently needs more perfect online advertising data monitoring and performance measurement standards. Recently, Nielsen, in partnership with Tencent and others, has moved directly into the country with companies such as Facebook and Verizon Wireless, launching digital Ad Ratings, DAR monitoring services. For the first time, DAR will provide Chinese digital advertisers with a more comprehensive, cross-platform/device-heavy metric that can be compared to TV ratings and reach their target audience, Nielsen said.
The author believes that the opening of third-party data, such as strong account system, is a very good trend. At a time when data and technology are driving growth, it's valuable both for advertisers' vital interests and for driving the growth of the entire digital marketing community. In view of the current development of the digital advertising industry, DAR's specific model, as well as the market for monitoring technology to introduce the initial impulse, the author carried out a lot of rational research and analysis, a few cold thinking to share with you:
Currently, global digital advertising ratings are based primarily on iGRP (Internet Gross Rating Points), which is the total Internet viewer/gross review. Measure digital advertising, iGRP, with gross reviews. Domestic mainstream third parties such as second hand, AdMaster, etc. have earlier supported iGRP, Reach measurement, and has accounted for more than 90% of the domestic monitoring market share.
“The concept of "big samples" was popular in the monitoring industry four or five years ago, but it has been vague. Statistically speaking, as the sample size increases, the error of the sample will gradually decrease, and when the sample size is greater than 1000, the error will be less than 5%, and when the sample size is greater than 3000, the error will be less than 1%. However, even if the sample size of more than 3000 is further expanded, the statistical error will be infinitely close to zero.
The author has studied a lot of statistics for this purpose, as well as some of the current situation in the industry. The use of samples to count ratings has been around for decades, ensuring small errors in data with limited data granularity, which is recognized throughout the industry. Even in the age of Internet advertising, the use of sample data to push and advertise audience structure has proved scientific and effective, and is widely accepted by the industry.
The problem with small samples is that as user behavior fragments, the sample size in fine-grained units is not sufficient to support the analysis. In response, Internet monitoring companies have addressed this by increasing the sample size to hundreds of thousands or even millions, while ensuring that most ad delivery can reach a sufficient number of samples to push the user structure of individual ad items, which is sufficient to address business needs.
An ideal solution to a small sample is to dissodge from the sample and measure it using a full amount of data, known as a census. But it's just an ideal situation, with no company in the country or even globally claiming to have all the user data, and the use of data from one or more media or vendors alone still has more limitations. The more realistic means is to get a relatively complete media audience structure by through the account system of most mainstream media.
To ensure the neutrality and authenticity of the data provided to customers, third parties insist on Panel, which has its own set-up and quality control (including million-level samples from third parties monitored by the Internet and 10,000-level samples from traditional view, all built by third parties themselves and controlled by themselves). Nielsen DAR monitoring services rely on the media to achieve this problem, which will have a lot of uncontrollable problems, which is also worth thinking about.
The information in the third-party sample needs to be detailed to the ID level, i.e. it needs to know the information of each device and everyone. The study found that DAR mode can not obtain ID-level information, can obtain the user structure after the media side aggregation. However, deviations in the aggregation process, or the actual sample size of the aggregation, etc., do not enable the third-party monitoring company's independent control, neutrality and precision are difficult to control.
At the same time, the user base of the media also has limitations, which will lead to a large deviation between the media sample structure and the Chinese of Chinese netizens and the structure of the mouth.
It is well known that advertising is delivered through IP for geo-targeted analysis, which is an internationally recognized guideline. In China, due to policy routing reasons, resulting in different rooms to see the same user's IP is not the same, but also will lead to media identification of the user source domain and third party identification of the user source domain is not consistent. The average error rate caused by IP instability can exceed 30%, and in extreme cases even 90%. Therefore, under the unified standard requirements of the China-Guangzhou Association, domestic technical third parties tend to use the same source room way to reduce errors and improve the reliability of data.
At the same time, in order to improve the safety of monitoring, SDK monitoring is also recognized by the industry as a way of monitoring. Through SDK, you can collect more accurate advertising exposure time, hardware device ID, etc. to help customers with data anti-cheating. At the same time, offline advertising also needs to be monitored through the SDK.
Dar mode also faces significant challenges in addressing these two points. The author found that Nielsen's main research and development and operations are abroad, there is no advertising monitoring SDK, so can not be deployed homor room to solve IP bias. The former will lead to data security damage, the latter will lead to large IP differences, relying only on the media to transmit regional information, easy to lead to the lack of neutrality.
While ad monitoring is important as a post-test for media delivery, brands and ad agencies need more than just monitoring, but data to help throughout the delivery process, such as traffic anti-cheating. False traffic, geography, and the identification of plays can all help customers reduce significant losses, which are also needed to think about the future of DAR services.
To sum up, digital advertising effectiveness monitoring and evaluation is a technical work, the need for practical and continuous technical polishing and practice. In the data opening industry trend, third-party monitoring companies only solid technology can encourage and meet the future more data opening and application, for the benefit of the industry.
Go to "Discovery" - "Take a look" browse "Friends are watching"
sent to have a look