AI and big data technologies has in recent years permeated every sector and industry from manufacturing, the internet economy to other traditional businesses. It seems as if AI has taken over the market overnight; and the sector in which it has been better monetized is none other than programmatic advertising.
As big data and AI gradually become prominent in brand marketing, it is indeed a force to be reckoned with in the industry. However it takes an expert in the field to fully harness the potential of AI. iPinYou founder and CEO Grace Huang is arguably one of the most well known figure in this area. This programmatic advertising maven gives this publication an insight into her world over an interview spanning a few hours.
Step-by-step: The building blocks of programmatic advertising
The robust development of programmatic advertising currently did not just appear out of nowhere. Since entering China in 2012, programmatic advertising has come a long way from its baby step as a concept on paper to be the most valuable and fast-moving component in advertising. According to Huang, the four essential stages in the development of programmatic advertising are:
First stage: New kid in the block
Being one of the latest method in advertising, programmatic advertising has from the very beginning attracted attention from various parties in the industry. However, the market then did not have a good grasp of the programmatic concept and was far from knowledgeable of the inner workings with the technology. Programmatic advertising had yet to gain wide acceptance among the industry players.
Second stage: Improvisation
On the surface, programmatic advertising seems to be part of media planning. However, its implementation requires an ecosystem that serves as an anchor to its objectives. While brand marketers have looked to programmatic as a platform to carry out their media strategies, the prerequisite to programmatic is the varied number of publishers on its system. This is important for a multidimensional approach that will be backed by the data report later.
Third stage: Getting on the bandwagon
Certain companies may be motivated by profit to slap programmatic on whatever operations they are running, be it buying and reselling traffic, engaging in practices that are in conflict of interest to their reputation and clients. Such practices are not uncommon but it is certainly not sustainable.
Fourth stage: Back to basics
First and foremost, technology and data are at the heart of programmatic, principally speaking. Since a data and technology backed strategy remains the USP (unique selling point) of programmatic advertising, the focus should be on technology for now.
However, in reality, programmatic advertising is not really a playing field for any interested party. Before venturing into programmatic, you have to ask yourself whether you are a tech vendor or resources vendor. Do you have the required technical suite and does it meet the required standards? What kind of system are at your disposal and what are its capabilities? Do you have data and how it is backended? What is the algorithm that you use, and what are its potential in strategy plannning? Being certain of your technical capabilities is the first step in getting back to basics.
Secondly, it is crucial to allign your technical capabilities with the media strategy. A common occurrence in the market previously was that ad tech companies were well equipped with advanced technology but brand marketers do not know how to make good use of it. Fortunately, the situation has improved slightly with more and more brand marketers willing to tap into the technology under an open and mature system.
We must know that CTR can be improved many times over even with small changes in the strategy. How can we achieve the full potential of programmatic advertising? No doubt it is through a well-planned media strategy to reap profitability from technology and make full use of data available in the market.
Furthermore, conflict of interest is a serious problem in programmatic advertising. It was very acute in the industry previously with publishers being involved in ad serving and data vendors monitoring and verificating the data themselves. That inevitably lead to issues when big data comes into play. Conflict of interest certainly will give rise to fraud. Therefore, reassessing the situation is a crucial step in getting back to the basics of technology.
Behind the interest of each party is transparency. Trust issues will rear its ugly head when all the parties involved are not forthright with their processes. That was the main problem faced by the clients in 2016. As an ad tech vendor, iPinYou has been at the forefront of the transparency initiative since last year to help restore trust among brand marketers who have lost their faith in the whole system due to the lack of transparency. We can’t afford to let any offending company off the hook because brand marketers will lose faith on the whole system and the consequences are immense.
Levelling up: Reconstructing the AI revolution
Disorderly competition seems to be rite of passage for the development of the Chinese market. The scenario was not limited to programmatic advertising but experienced by other red-hot fields such as the early stages of ecommerce, barter trading, P2P, to even the recent bike sharing startups. Nevertheless, Huang believes that disorderly competition will sieve out the weaker players and is unavoidable for the market to progress further.
At the same time, Huang has her own view on the application of AI. “ In reality, only AI in (programmatic) advertising is the real AI, it cannot be compared with the AI in a game of chess where only wining or losing matters. Programmatic advertising involves ad serving under very clear factors. This is why a big data and A1 oriented digital marketing strategy needs a lot of investment. We don’t want to waste resources due to a misguided A1 system in strategy planning.
Having said that, AI and algorithm are not entirely new concepts. However, its implementation and results are better represented in digital marketing and programmatic advertising over the past few years.
As it has been often said, with power comes responsibility. iPinYou has been at the forefornt of big data and AI in the past few years. Being at this enviable position, iPinYou has taken up the challenge to turn the huge trove of unstructured data into useful information that can be use in programmatic advertising as audience profile. It is done through the compilation of inumereous tiny details to contruct a multidimensional and rich audience profile which when combined with other information and algorithm will result in the final strategy.
In 2015, iPinYou upgraded its technical suite for its overall system after spending over 1 billion RMB in investment. The huge injection of funds and responsibility that comes with being purveyor of transparency is not an easy task that can be undertaken by any company. iPinYou has the foresight to plan ahead to bolster its capabilities when so many others are blindly chasing the profit because technology is the most useful tool that any company could have.
Asked on how she came up with the decision, Huang said: “iPinYou fully understands what can be done with technology, how to make good use of big data and A1 to glean useful information from the unstructured data and utilize it strategy planning. We only have one goal: to do our very best in data mining.
Sincerity rewarded: Going all out in building feedback mechanism
In managing our expectations, we have to be aware that advertising is not an industry to reap benefits from a short-term effort. There are so many players in the market, there will be those who accept reasonable demands from clients and those that offer ridiculous prices far below market standards. They may have good results to boast off but the rewards will be short-lived and not those afforded by big data and AI.
Huang stresses that what sets iPinYou apart from others is its emphasis on strategy planning.
Simply speaking, AI is not too difficult but what’s at heart of the system is its feedback mechanism. For example, the application AI in chess is made possible by feeding data to the system and deep learning. It is the same with big data and AI, strategy planning in advertising should be adjusted according to reactions from the target audience. Better ROI after ad serving is positive reinforcement from the system.
While each sector would have its own media strategy and KPI such as CTR, download rate, coverage target, conversion rate, only brands that have a well-planned media strategy and clear-cut KPIs would have a comprehensive feedback mechanism. These brands are well on track to engage in programmatic advertising.
Big data thrives on its amount—the bigger, the better and more accurate. Niche brands may have a smaller user base and hence fewer data and feedback. Brands that have more data would benefit from a richer database, likewise.
At the same time, Huang believed that big data and AI would greatly benefit the marketing of traditional brands. For example:
Setting things straight: The flow of big data
It must be brought to attention that the market has had a false assumption of the value of big data over the recent years, not the mention the various inaccurate definitions of big data. However, iPinYou has been fairly strict in defining what constitutes big data, both from an objective and scientific perspective. “Big data is derived from a huge trove of data accumulated via various sources that can be processed into useful resources.”
Let’s look at the terminology involved: complex, unregulated, extensive and varied. This also implies that it is hard for a single sector to own big data, and each brand should not count on having its own big data that is valuable. What is most important is to have some data.
The second half of the argument looks at the different values generated from the integration and processing of data. Huang gave an analogy to better explain the proposition: “Let’s take a male consumer as an example. If Company A knows that he has bought a pair of shoes today, while Company B knows that he has watched a movie and Company C knows that he has bought a bouquet of flowers and lastly Company D is aware that he has purchased a greeting card. By putting all this information together, we can easily assume that this consumer is likely to have just celebrated Valentine’s Day, a birthday or an anniversary. But if we just look at the data separately, we could not have deduced his purchasing behavior from just the sales of the shoes alone. This is why data is useless if we don’t look at it as a whole.
It is the same with marketing. Rather than treating data as its own property, it is more important for brands to have the ability to manage data. A vibrant data flow would be more valuable than otherwise, few would have treated tap water from their homes as valuable resources, would they? Big data is like water, it is only valuable while it is flowing. We cannot expect it to be valuable when it stays stagnant. “
This is precisely the reason for industries to maintain a healthy flow of data in order to reap benefits from data itself. iPinYou’s role is to reactivate the data, so as to create its market value. More often than not, brands do not know what to do with the data they own. For example, iPinYou has worked with China Telecom recently to study and utilize its data for marketing purposes. If we don’t dig deep for the data and use it, the huge amount of data generated will just be wasted away like flowing water. This is the biggest misconception of big data among the brands.
Since big data is the precursor of AI, it is nearly possible for a single industry to have its own big data or even AI capabilities. What they can do is to look for tech partners with AI platforms to enhance their technical capabilities.
We can find new information by tinkering with engines, while social media helps consumers to look for their friends. What iPinYou wants to do is to help brands reach their target audience. Connecting information with people has been the cornerstone of iPinYou since its founding more than 10 years go. The objective lives on even as the market and our lives become more fragmented.