Original article can be found here (source): Artificial Intelligence on Medium
How to improve the results of Social Ads with Artificial Intelligence
Those of us who work in the digital environment always look for new ways to achieve more and better results. The application of Artificial Intelligence is one of the greatest advances to date, and it is now when we begin to see what it is capable of bringing us.
To develop it better, I will try to explain its benefits in general first. Next, I will comment on some specific examples in which artificial intelligence can add value when carrying out advertising campaigns on social media , as we do at Adtuo .
Advantages of using Machine Learning
These algorithms are designed with the objective that the machine does what a human would do, trying to imitate it to a great extent, but in a much faster and more efficient way. The machine is primarily tasked with performing specific tasks, but being fully focused on them, exponentially maximizes the chances of achieving success with the highest efficiency. These machines are nothing more than computers, with their processors and hard drives, which although inspired by our brains, have 3 great advantages.
The first advantage is the storage capacity. Today, as much information as desired can be stored, unlike the human brain which, although we were able to take advantage of its full potential, has a limited capacity. According to the predictions of Moore’s law with computers, the opposite happens, since they can store almost infinite data in the cloud. It is only necessary to expand said capacity as required, this being a great ally when developing tasks that need to handle immense amounts of information.
Today, thanks to Big Data (which is nothing more than the combination of new databases, such as non-relational or graph databases and the same as always, with advanced storage technologies and management of huge amounts of content) we can store vast volumes of complex data, executing these operations and instructions with total normality.
The second difference is its high capacity to process information, due to the continuous development of better microprocessors with a greater number of cores. This results in more processing cycles per millisecond. According to the predictions of Moore’s law, every year, or year and a half, it is possible to double this power. With current microprocessors it may seem that we have reached a ceiling, but we will see how progress continues with quantum processors, which represent the next level jump in this field. So, we can affirm that we still have a long process ahead of us.
The third and final difference is the mathematics-based algorithm, which, although not something new, because it actually comes from the statistics of the 50s and the advances of the pioneer Alan Turing. Now this algorithm can be applied with the aforementioned technology, without limiting space and shrinking time, thanks to its increasing speed when processing. What gives us the necessary ingredients to achieve everything imaginable.
Artificial Intelligence and advertising
If we talk about advertising, the possible applications of artificial intelligence are several, although the most obvious is to improve results. That is, increase the number of results obtained by the same budget or reduce the cost to achieve them.
To do this, we must understand what works best, and what performs worse. This can be applied both to differentiate channels, and to understand which of them is the one that contributes the most value to the desired results. What would be the traditional attribution of marketing budgets, but within a specific environment such as advertising in social media.
An example of this would be understanding which ad works best within the main social networks of the moment, such as Facebook and Instagram, as well as which locations can be most useful for effective advertising, asking ourselves if we get more results in the news feed, in the lateral location of Facebook, or in the Stories.
Another element to analyze is which devices (computer, mobile or tablet) can perform better, depending on the ad format we have chosen: traditional publication, carousel, video or collection, etc.
As many of you may already know, the ability to segment audiences is one of the fundamental keys to advertising on social networks. Therefore, are especially important when working on optimization sociodemographic variables, such as: gender, age, location, socioeconomic level or studies, with special relevance being those that tell us about the interests and tastes of our future potential clients.
Finally there are the creativities, that is, the ads, which are the first element to make an impact. If the image or the accompanying texts are not attractive enough to encourage interaction, all the previously established segmentation will not do much good. And always keep in mind that the performance of the same ad will be completely different depending on who and where it is displayed, depending on the location, device, format, etc.
At this point, we cannot forget the objectives of the campaigns. It has nothing to do with being able to analyze the information for objectives such as: post engagement, attracting traffic, getting leads or conversions (whose actions occur outside of Facebook), than being able to analyze the number of sales achieved in an online store or the number of downloads achieved from an app. To pursue these objectives, we will need to have the Facebook pixel configured, along with a series of predefined standard conversion events, in order to record each action carried out within our website.
Needless to say that we cannot ignore the analysis of each of the measurement indicators such as cost per click, impressions, CTR, reach, interactions, etc. of each of the channels with which we work. The elements to take into account are many, and their combinations almost infinite.
Joining human … and artificial intelligence
The best thing about all this is that, although we can analyze these elements individually, and go manually one by one, seeing what can go wrong or what works, the reality is that, as expert as we are, we all have our specialty, of which we know much more than the rest of the areas and our understandable professional bias based on our background.
In this sense, the human will always be able to find aspects to improve, based on experience. However, if this experience is greater in one field than another, it will only be able to improve some elements. In addition, it is necessary to take into account the own limitations that become weaknesses in front of the machine, such as working hours with a maximum of hours to dedicate, attention to different customer campaigns and the normal work speed that people have.
On the other hand, when a machine has through programmed algorithms, all these predefined parameters, it is capable of learning and combining them, carrying out constant and simultaneous optimization. It is also capable of quickly reaching conclusions, which surely those who have programmed it could not even approach. It is impossible, in a human way, to have a total and real-time understanding of all the data involved, and of all the relationships of the elements connected to each other.
To conclude, it’s important to highlight that artificial intelligence is capable of realizing when an element of the set does not work and that, thanks to its implementation, it is possible to detect behaviors, signals and search patterns that allow campaigns with a high probability of success.
By constantly learning from your results, you can guide us in creating a new marketing strategy, in which we need to know which products or services work best depending on the time of year we are in, where our efforts should go, and what type of content can most attract our customers.
The good thing about all this, is that, once the machines do the ‘hard work’, or at least the most repetitive and worthless, it leaves us as marketing professionals the possibility of dedicating our human intelligence and efforts to what our dedication can provide more added value, such as defining the most appropriate strategy, better understanding our users and having the best ideas bringing real value to the business.
In short, the best results go through the balance between the potential and advantages of Artificial Intelligence and the experience and added value that human capacity can bring.
Article originally published on 11 July 2019 in Marketing4Ecommerce.com