A series of articles “Smart techniques in online advertising”
Part 1. AI knows us better than we do
This is it!
Just look around, analyze global tendencies and never stop asking yourself questions “Why?” and “What is going to happen?” A world progress snowball has just started to accelerate along a slope and I pray that we have enough reaction to go round any obstacles. In any case, let’s leave philosophy for the best minds of humanity and just analyze one business direction and one field of science that have mixed some time ago and impend to change too much of a daily routine for an ordinary man in next 10–20 years.
You probably dawned upon a field of science I’m going to talk about from the topic of this article. This weird pair of words is “Data Science” and combines lots of sub-terms like Data Mining, Big Data, Machine Learning, etc.
Still a primary quantity of people doesn’t have a clue on what Data Science is. Every day people make photos using a face auto grasp option, start search saying “Ok, Google!” and watch movies online, basing on recommendations of a website, they search at, etc. Very few of them are curious, what is behind the curtain, making their lives easier. The fact that data science algorithms organically fit our lives and imperceptibly add extra comfort is the key reason for a rapid development of this branch.
Data science affects on all business spheres, but one of them is in the face of total remapping. What are the very first associations for a typical person, who hears a word “Advertising”? I guess that not too pleasant. Colorful posters in metro and along streets; happy housewives using another detergent at TV; irritating pop-up banners at websites and finally enthusiastic and a little bit crazy voices calling at your mobile and persuading you to buy insurance / take a loan etc. Advertising activity brings lots of profit to all parties that participate in process, but also lots of irritation to the ordinary people, who are felt as “goods” at a market. Data science is aimed to change advertising completely.
With a constant increase of new information that our brains process each day, people become immune to tons of advertisements, paying attention only to the necessary one. If showing each person ten times less advertisements, the probability that overloaded eye or ear would respond to a new info increases proportionally. It is worth mentioning that the main purpose of advertising is a sales assistance. As a second step, for big companies, it is also brand awareness, but in any case, a sale is primary. If only companies knew names and addresses of people, who need their goods or service, who strive them, who will pay for them right here and right now! In that case there would be no need in advertising at all!
So, what is a role of data science for advertising? Data science kills advertising… in a good sense☺. It transforms advertising into a scientific sales technique. Nowadays, metrics to measure advertisers’ skills and campaign results is not a quantity of billboards and posters, not in square miles of territory covered by shouting posters, but in customer’s awareness and goods/services sold. Data science itself is not an initiator of that transformation — people are. Data science is only an instrument that accelerates an unstoppable world progress.
Ok, but how does data science actually affect advertising? Every one of us considers himself to be unique. This is true to some extent. Everybody has his unique set of traits of character, appearance, principles and outlook on life, but our behavior and wishes are predictable. If a person enters a food shop, he or she wishes to buy something to eat or drink. Dressy look garment? Then probably, a person will look for something sweet, a cake or candies. Woman? Women more than men like chocolate, then it is likely to be some chocolate cake. An old woman? Granny? It is forbidden to eat much of a sugar for old people, so cake should be light and definitely without hard nuts inside. Imagine yourself to be an advertising agent with 10 different meal offers, standing in front of a shop, deciding what to offer or advertise to each client. A client may not even understand his wish, but his appearance and behavior will give you a clue.
Data science is a source of complex algorithms, even virtual robots that imitate human’s decision making processes that are based on many-years experience. Scientists feed such algorithms with enormous amount of data, divided into two main segments: lots of features and one result. Appearance, age, sex, cloths of a person and his behavior are features and goods he finally bought — result. Scientists “train” such virtual robots to serve many spheres of business and ordinary life and advertising, mostly online advertising is one of the highest priorities.
With a help of such artificial intelligence, it becomes much easier to find a real client in a crowd and not waste time and strength on advertising every single person your product. Moreover, advertisers can target those users, who even do not know about their inner wish to react positively on advertising. Should be mentioned that on the opposite side of the process, companies that pay money for promotion and advertising of their product/service have lots of benefits, too. They begin to understand an audience segment with a positive reaction to their goods/service and this helps much in marketing and up sales.
Key advantages of using machine learning in online advertising are:
- Prediction,
- Recommendation and
- Visualization of audience segments
Technical explanation of each one will be described in Part 2.
