User:Alr5845/Artificial intelligence marketing

Artificial intelligence marketing
Artificial intelligence marketing is a type of marketing that uses databases, machine learning, and algorithms to comprehend data points to directly market a consumer. It Is a method to leverage customers information to anticipate the customer's next buy. Artificial intelligence marketing uses these tools to connect the data points of a specific consumer to customize marketing to them. The main reason of using this type of marketing is to anticipate the customers next purchase. The algorithms companies used must be scalable, efficient, and detailed in order to get accurate results.

The main goal of artificial intelligence marketing is to market to a specific customer with speed. The messages and advertisements they receive will be in real time and directed towards them. This type of marketing augments the marketing team to be faster and more personalized to communicate with customers. There are a few components to artificial intelligence marketing. These components include behavioral targeting, collect, reason, and act principles, and machine learning. All of these techniques combined allow for companies to zone into the customer's needs and future spending.

Behavioral Targeting
Behavioral targeting refers to reaching out to a prospect or customer with a communication based on implicit or explicit behavior shown. It is the understanding of behaviors is facilitated by marketing technology platforms such as web analytics, mobile analytics, social media analytics and trigger based marketing platforms. Artificial intelligence marketing provides a set of tools and techniques that enable behavioral targeting.

To improve the efficiency of behavioral targeting, machine learning is used. Also, to prevent human bias in targeting customers based on behaviors and do this at scale, artificial intelligence technologies are used. The most advanced form of behavioral targeting with the help of artificial intelligence is called Algorithmic Marketing.

There are two types of behavioral targeting; onsite and network. Onsite behavioral targeting is described as advertisements or promotions that are used to target the consumer based on the products they have bought or website they have visited. Information is gathered on the customer from their past experiences with the company. This technique is used to fast track the experience for consumers. Advertisers and companies can engage consumers in a new way by displaying recommended products that are specific to that consumer.

In network behavioral targeting, advertisements are assumed for the audience. The audience is categorized by the decisions they make that resonate with their demonstrated behavior. This information is most likely used to get a customer back and retarget them to the company. All this information is put into an algorithm that can make more assumptions about the person such as age, demographics, gender, and possible future buys. This data normally comes from the advertiser's website which is then used to personalize the experience for the customer. For example, if a person checks on makeup and women's fashion websites, one can assume that the user is a female.

The main component to behavioral marketing is the gathering of information and data. This is done through a DMP which are responsible for gathering, storing, and organizing the data found on consumers. The data can come from multiple sources such as, apps, websites, IP addresses, clicks, number of visits, content read, previous purchases, etc.

Behavioral Targeting Process

 * 1) Collection and Analysis of Data: Data on a user is collected from a number of different sources as stated above but all of this is done through "tracking pixels" or third-party cookies. The data is analyzed to collect information on a demographic of people.
 * 2) Segmentation: The users are segmented into groups by their online behavior. Most of these groups or segments are based on likes and interests. One example of this is if one person visits websites for dogs versus another person looking at cat websites. The first person would be put into the dog grouping, while the other person would be segmented into the cat group.
 * 3) Application of Data: Advertisements are then made to match each segment. The advertisements are more relevant and effective for each group they are targeting which increases the likelihood that the response from the group will be positive.

Contextual Targeting vs Behavioral Targeting
Contextual targeting does not use information about the user. It takes into account and uses contact of ads to an advantage. Behavioral targeting allows targeting to exact individuals based on their interests. Contextual targeting is much easier to achieve because behavioral targeting requires much more sufficient information.

Concerns
General Data Protection Regulation is in effect since 2016 in the EU and is the basis of rights and freedom from personal data and information including ePrivacy. This makes behavioral targeting more difficult with the stricter rules. It may even make marketers change their marketing techniques completely and move away from behavioral targeting. True contextual targeting is in compliance with these rules and regulations, but it does not allow the collection of personal data. A few examples of this is AccuWeather partnering with a contextual advertising automation company to provide real time ad personalization algorithms.

As mentioned in the behavioral targeting article :"'Many online users & advocacy groups are concerned about privacy issues around doing this type of targeting. This is an area that the behavioral targeting industry is trying to minimize through education, advocacy & product constraints to keep all information non-personally identifiable or to use opt-in and permission from end-users (permission marketing).'"

Artificial intelligence marketing takes a lot of training and time to implement. One of the challenges with implementing this kind of technology is that it requires a ton of data from customers. It takes time for the algorithm to have the high quality.

Collect, reason, act
Artificial intelligence marketing principle is based on the perception-reasoning-action cycle one find in cognitive science. In marketing context this cycle is adapted to form the collect, reason and act cycle.

Collect
This term relates to all activities which aims at capturing customer or prospect data. Whether taken online or offline these data are then saved into customer or prospect databases.

Reason
This is the part where data is transformed into information and eventually intelligence or insight. This is the section where artificial intelligence and machine learning in particular have a key role to play.

Act
With the intelligence gathered from the reason step above one can then act. In marketing context act would be some sort of communications that would attempt to influence a prospect or customer purchase decision using incentive driven message

Again artificial intelligence has a role to play in this stage as well. Ultimately in an unsupervised model the machine would take the decision and act accordingly to the information it receives at the collect stage.

Machine Learning
Machine learning is concerned with the design and development of algorithms and techniques that allow computers to "learn". It the ability for a machine to automatically learn and improve whatever experience they are providing without changing the algorithm within it. Machines learn by absorbing large amounts of data and looking for patterns within that dataset. The machines then use the information to make better decisions. The main goal of machine learning is to delete human assistance. It can be used to frame behavioral targeting.

There are four types of machine learning: supervised machine learning algorithms, unsupervised machine algorithms, semi-supervised machine learning algorithms, and reinforcement machine learning algorithms.

Supervised Machine Learning
Supervised machine learning is when the machine applies past and new data to predict future events. Humans are provide the machine with data and inputs to show correct outputs. from this process, computers will be able to find patterns and learn from them. The algorithm learns about each output decision and bases future decisions off of that. The machine will be able to identify errors in the data and correct them with a new output.

Unsupervised Machine Learning
Unsupervised machine learning is when the information that is given to the machine is not classified or labeled. The algorithm does not give a correct output value like supervised does, but it does report inferences it makes from the dataset. The descriptive model is most common in this kind of machine learning. The machine uses the inputted data find rules and patterns to come up with inferences about the customers.

Semi-Supervised Machine Learning
Semi-supervised machine learning is between the two listed above. It normally consists of a small amount of labeled data and a large amount of unlabeled data. It uses this algorithm and amount of data to improve accuracy. This type of algorithm requires relevant sources to train it. The cost of supervised machine learning is too high. The data used in semi-supervised machine learning has information about group parameters, not individual information.

Reinforcement Machine Learning
Reinforcement machine learning algorithms is a method that uses interactive learning to produce output and find the inefficiencies. It is a branch of artificial intelligence. The main characteristics of this algorithm is trial and error. The machine automatically knows what the behavior of the data is to maximize the performance.