{"id":12918,"date":"2020-03-25T21:48:27","date_gmt":"2020-03-26T02:48:27","guid":{"rendered":"https:\/\/www.invespcro.com\/blog\/?p=12918"},"modified":"2024-01-12T15:36:19","modified_gmt":"2024-01-12T15:36:19","slug":"the-use-of-machine-learning-and-artificial-intelligence-in-conversion-optimization","status":"publish","type":"post","link":"https:\/\/www.invespcro.com\/blog\/the-use-of-machine-learning-and-artificial-intelligence-in-conversion-optimization\/","title":{"rendered":"The Use of Machine Learning and Artificial Intelligence in Conversion Rate Optimization"},"content":{"rendered":"<span class=\"span-reading-time rt-reading-time\" style=\"display: block;\"><span class=\"rt-label rt-prefix\">Reading Time: <\/span> <span class=\"rt-time\"> 14<\/span> <span class=\"rt-label rt-postfix\">minutes<\/span><\/span><p><i><span style=\"font-weight: 400;\">Disclaimer: This section is a TL;DR of the main article and it\u2019s for you if you\u2019re not interested in reading the whole article. On the other hand, if you want to read the full blog, just scroll down and you\u2019ll see the introduction.<\/span><\/i><\/p>\n<ul>\n<li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">In his famous book Machine Learning, Tom Mitchell defined ML as \u201cthe study of computer algorithms that improve automatically through experience.\u201d\u00a0<\/span><\/li>\n<li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">The algorithm learns automatically in a loop, but humans must establish rules for this learning process.<\/span><\/li>\n<li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">When compared to standard statistical modeling, machine learning is time-saving because the data modeler needs to make the improvement her\/himself, while ML algorithm uses a lot of data to make accurate decisions.<\/span><\/li>\n<li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">Being able to write down the function and final prediction for ML is what differentiates it from artificial intelligence. In AI the function is not known and often impossible to write down in an explicit mathematical way.<\/span><\/li>\n<\/ul>\n<p><b>Using Machine Learning in Digital Marketing<\/b><\/p>\n<ul>\n<li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">It can be used to analyze customer reviews in an eCommerce store. These reviews are a bunch of unordered text files that need to be categorized.<\/span><\/li>\n<li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">Another example is social media feeds for sentiment analysis.<\/span><\/li>\n<li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">For cluster analysis (to search for subgroups with common baseline patterns).<\/span><\/li>\n<\/ul>\n<p><b>Artificial Intelligence Defined<\/b><\/p>\n<ul>\n<li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">It\u2018s the science and engineering of making computers behave in ways that until recently we thought required human intelligence.<\/span><\/li>\n<li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">When we say AI nowadays, people think of Siri, Alexa and other \u201cintelligent\u201d assistants. These digital assistants can perform some basic tasks for you.<\/span><\/li>\n<li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">The term artificial intelligence was first used in 1956 by a group of researchers including Allen Newell and<\/span><a href=\"https:\/\/www.cmu.edu\/simon\/what-is-simon\/history.html\"> <span style=\"font-weight: 400;\">Herbert A. Simon<\/span><\/a><span style=\"font-weight: 400;\">.<\/span><\/li>\n<\/ul>\n<p><b>Artificial Neural Networks<\/b><\/p>\n<ul>\n<li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">These networks try to imitate the human neural networks that exist in our brains. It\u2019s the class of algorithms that model the connections between the neurons.<\/span><\/li>\n<li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">Neural networks are popular and widely used in medical research, especially in bioinformatics for DNA sequence prediction, gene expression profiles classification, etc.<\/span><\/li>\n<\/ul>\n<p><b>Deep Learning Defined<\/b><\/p>\n<ul>\n<li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">It\u2019s a specific approach used for building and training neural networks.<\/span><\/li>\n<li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">The input data is passed through a series of nonlinearities or nonlinear transformations.<\/span><\/li>\n<li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">Deep learning is actually a part of machine learning.<\/span><\/li>\n<\/ul>\n<p><b>Deep Learning Applied in CRO<\/b><\/p>\n<ul>\n<li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">An intelligent chatbot that actually uses AI can help you in a fast and efficient way to provide better customer support or makes the sale process much shorter.\u00a0<\/span><\/li>\n<li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">Ecommerce stores can implement the \u201cpeople who bought this also bought\u201d feature through product tagging \u2013 however, deep learning can provide a much more interesting data set.\u00a0<\/span><\/li>\n<\/ul>\n<p><b>The Trap of Personalization<\/b><\/p>\n<ul>\n<li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">There is a trap of too much personalization in ads, searches, social media content, etc.\u00a0<\/span><\/li>\n<li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">It is called a \u201cfilter bubble\u201d effect or the \u201cecho chamber\u201d described recently by<\/span><a href=\"https:\/\/arxiv.org\/abs\/1902.10730\"> <span style=\"font-weight: 400;\">Jiang et al<\/span><\/a><span style=\"font-weight: 400;\">.\u00a0<\/span><\/li>\n<\/ul>\n<p><b>Evolutionary Algorithms in Experimentation<\/b><\/p>\n<ul>\n<li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">There is much more about AI usage in conversion rate optimization than just chatbots, personalized ads, or profiles.\u00a0<\/span><\/li>\n<li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">There are five main stages in the evolutionary algorithm: (choosing) Population, Evaluation via Fitness Function, Selection, Crossover, and Mutation. In the first step, we choose a population characterized by a particular set of variables (genes).\u00a0<\/span><\/li>\n<\/ul>\n<p><b>Conclusion<\/b><\/p>\n<ul>\n<li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">Genetic algorithms are a very flexible tool. You can specify more than one fitness function for each segment you want to find a winner in. And each segment can have different fitness function criteria.<\/span><\/li>\n<\/ul>\n<p><span style=\"font-weight: 400;\">\u00a0<\/span><\/p>\n<p style=\"text-align: left;\"><b>Here&#8217;s A Longer And More Detailed Version Of The Article.<\/b><\/p>\n<hr \/>\n<p><span style=\"font-weight: 400;\">Can we use <a href=\"https:\/\/www.invespcro.com\/blog\/chatgpt-for-cro\/\">Artificial Intelligence<\/a> (AI) to improve the user experience on a website and thus impact the site\u2019s conversion rate?<\/span><\/p>\n<p><span style=\"font-weight: 400;\">It seems that many products try to mention that they use AI as a unique feature that makes them different from other competitors.\u00a0\u00a0\u00a0<\/span><\/p>\n<p><span style=\"font-weight: 400;\">However, many AI methods are always confounded with machine learning methods.\u00a0<\/span><!--more--><\/p>\n<h2><span style=\"font-weight: 400;\">What is Machine Learning?<\/span><\/h2>\n<p><span style=\"font-weight: 400;\">In his famous book Machine Learning, Tom Mitchell defined ML as \u201cthe study of computer algorithms that improve automatically through experience.\u201d\u00a0<\/span><\/p>\n<p><span style=\"font-weight: 400;\">The phrase \u201cimprove automatically through experience\u201d is a crucial point since that is what we call \u201clearning.\u201d Obviously, \u201cmachine\u201d refers to \u201ccomputer algorithms.\u201d\u00a0<\/span><\/p>\n<p><span style=\"font-weight: 400;\">Who is learning then? We or the algorithm?\u00a0<\/span><\/p>\n<p><span style=\"font-weight: 400;\">The algorithm learns automatically in a loop, but we must establish rules for this learning process. This automatic \u201clearning\u201d is therefore controlled by us, but we do not have to improve the learning models manually.\u00a0<\/span><\/p>\n<p><span style=\"font-weight: 400;\">This means that ML is convenient as compared to standard statistical modeling when a data modeler needs to make the improvement by her\/himself. It is definitely time-saving, mainly when the ML algorithm uses a lot of data to make accurate decisions.\u00a0<\/span><\/p>\n<h3>What are the examples of the typical machine learning process?<\/h3>\n<p><span style=\"font-weight: 400;\">Let\u2019s take an example from the medical field, where ML helps solve many classification problems. ML is used to assess MRI images to decide whether the patient has a disease or not.\u00a0<\/span><\/p>\n<p><span style=\"font-weight: 400;\">The ML algorithm must begin with an initial \u201clearning set.\u201d This is a set of MRIs with a diagnosis made by the physician. The algorithm is \u201ctrained\u201d initially through by using the initial learning set. Afterward, the algorithm will classify a \u201ctest set\u201d of new images.\u00a0<\/span><\/p>\n<p><span style=\"font-weight: 400;\">The training process is based on exposing the algorithm to a set of images with known diagnosis.\u00a0<\/span><\/p>\n<p><span style=\"font-weight: 400;\">During this phase, the analyst builds a model (a function) to describe the relationship between the image characteristics and the diagnosis outcome. This model specifies the <\/span><b>potential<\/b><span style=\"font-weight: 400;\"> image features are the <\/span><b>likely<\/b><span style=\"font-weight: 400;\"> candidates for meaningful predictors. The algorithm will test the images to see if it can establish that relationship.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">In machine learning, we know the final predictors in the model, and we know what the function is. So you can plug-in the predictors into the model function and obtain the expected outcome (or the probability of that outcome).<\/span><b>\u00a0<\/b><\/p>\n<p><span style=\"font-weight: 400;\">Once this function is built and its parameters are estimated, we can check the accuracy of the algorithm prediction. In other words, we can check how well the algorithm is trained. If the algorithm is well written, then it will search for similar patterns in the test and learning sets &#8211; and make the classification based on those similarities.\u00a0<\/span><\/p>\n<p><span style=\"font-weight: 400;\">In other words, the algorithm will look for the MRI that is the most similar to the one it is evaluating and assign its class into that MRI. In this example, the algorithm learns only once, unless we give it another training set.\u00a0<\/span><\/p>\n<p><b>Note on reinforcement learning &#8211;\u00a0<\/b><\/p>\n<p><span style=\"font-weight: 400;\">Some algorithms are constantly learning. These are called reinforcement learning. In the reinforcement process, our classification algorithm will look at the accuracy of its own prediction and use it for further learning. How? Penalizing wrong decisions and awarding the good ones.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">As you can see, with ML, we know the function and the final predictors &#8211; and we typically write down the function in an explicit mathematical way.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">That is what differentiates ML from artificial intelligence (AI): in AI, the function is not known and often not possible to be written down in an explicit mathematical way.<\/span><\/p>\n<h2><span style=\"font-weight: 400;\">Using machine learning (ML) in digital marketing<\/span><\/h2>\n<p><span style=\"font-weight: 400;\">There are many classification problems in <a href=\"https:\/\/automatetowin.com\/\">digital marketing<\/a> where ML can be handy. Let\u2019s take an example of analyzing customer <\/span><a href=\"https:\/\/www.researchgate.net\/publication\/328306943_A_Comparison_of_Machine_Learning_Algorithms_in_Opinion_Polarity_Classification_of_Customer_Reviews\"><span style=\"font-weight: 400;\">reviews in an ecommerce store<\/span><\/a><span style=\"font-weight: 400;\">. These reviews are a bunch of unordered text files that need to be categorized.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">Another example is social media feeds (Facebook, Twitter, etc.) &#8211; These are powerful sources of data, that provide a valuable opinion about a given product, brand, website, or a tool. In statistics, analyzing what people share is referred to as Sentiment Analysis, and it was used for many years for survey analysis.\u00a0<\/span><\/p>\n<p><span style=\"font-weight: 400;\">As a result of the large amount of data that social media provides, the demand for Sentiment Analysis increased tremendously in the last few years.\u00a0<\/span><\/p>\n<p><span style=\"font-weight: 400;\">Back in the late nineties, the \u201cbig data problem\u201d referred more to science, such as genetics or psychics. Nowadays, the \u201cbig data problem\u201d has the face of \u201cFacebook\u201d?.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">Of course, there are also other problems that ML can solve apart from classification related problems.\u00a0<\/span><\/p>\n<p><span style=\"font-weight: 400;\">When we do not have the classes defined a <\/span><i><span style=\"font-weight: 400;\">priori<\/span><\/i><span style=\"font-weight: 400;\">, we might be interested in finding the groups of similar features.\u00a0<\/span><\/p>\n<p><span style=\"font-weight: 400;\">For example, we do not want to merely classify our visitors to \u201cbought\u201d or \u201cdid not buy\u201d &#8211; we might want to dig deeper. In this case, ML is used for cluster analysis. ML is used to search for subgroups with common baseline patterns, regardless of any \u201cbought\/did not buy\u201d type of response. ML could be used to classify visitors as \u201clikely to buy\u201d or \u201cnot likely to buy.\u201d Imagine a visitor coming to your site, and ML tells you that the visitor is likely to buy &#8211; what kind of offer do you show them? Do you need to show them any offers?<\/span><\/p>\n<p><span style=\"font-weight: 400;\">How about estimating how much a visitor is likely to spend. This is modeled using regression modeling &#8211; which is just a linear function that represents the relation between the amount spent by the visitor and a set of predictors. Those predictors can be strictly related to visitor characteristics such as age or the region. Still, there could also be other information included, such as the type of device he\/she uses or time (day) of the transaction, etc..\u00a0<\/span><\/p>\n<p><span style=\"font-weight: 400;\">All those models can learn and be automated using ML.<\/span><\/p>\n<h2><span style=\"font-weight: 400;\">Artificial intelligence<\/span><\/h2>\n<p><span style=\"font-weight: 400;\">Now that you have become more familiar with machine learning, allow me to introduce you to AI. <\/span><\/p>\n<p><span style=\"font-weight: 400;\">According to Andrew Moore,\u00a0<\/span><\/p>\n<blockquote><p><span style=\"font-weight: 400;\">Artificial intelligence is the science and engineering of making computers behave in ways that, until recently, we thought required human intelligence.<\/span><\/p><\/blockquote>\n<p><span style=\"font-weight: 400;\">Another definition is:<\/span><\/p>\n<blockquote><p><span style=\"font-weight: 400;\">The simulation of human intelligence processes by machines, especially computer systems<\/span><\/p><\/blockquote>\n<p><span style=\"font-weight: 400;\">In other words, AI is all that comes naturally to human beings like communication through conversation, drawing conclusions, and making decisions. When we say AI nowadays, people think of Siri, Alexa, and other \u201cintelligent\u201d assistants. These digital assistants can search for some information for you, book a flight for a hairdresser, say a joke, or tell you the weather forecast. <\/span><\/p>\n<p><span style=\"font-weight: 400;\">The term \u201cartificial intelligence\u201d was first used in 1956 by a group of researchers, including Allen Newell and <\/span><a href=\"https:\/\/www.cmu.edu\/simon\/what-is-simon\/history.html\"><span style=\"font-weight: 400;\">Herbert A. Simon<\/span><\/a><span style=\"font-weight: 400;\">.\u00a0<\/span><\/p>\n<p><span style=\"font-weight: 400;\">Since then, the definition of AI changed as more and more developments in the field take place.\u00a0<\/span><\/p>\n<p><span style=\"font-weight: 400;\">What was referred to as AI in the past is not what AI is today.\u00a0<\/span><\/p>\n<p><span style=\"font-weight: 400;\">We notice that more companies claim that they use AI in their products. This was not the case twenty years ago. When IBM developed the famous Deep Blue, many insisted that it was not using artificial intelligence, while it actually did.\u00a0<\/span><\/p>\n<p><span style=\"font-weight: 400;\">Everything has changed after deep learning methods, and neural networks became popular.\u00a0<\/span><\/p>\n<h3><b>Artificial neural networks<\/b><\/h3>\n<p><span style=\"font-weight: 400;\">Let us start from the neural networks (NN), or we should rather say artificial neural networks. As we might guess, these networks try to mimic the human neural networks that exist in our brains. Therefore, it is the class of algorithms that model the connections between the neurons.\u00a0<\/span><\/p>\n<p><span style=\"font-weight: 400;\">These algorithms are derived from biological systems. They consist of a network of nonlinear information processing elements called \u201cneurons\u201d that are arranged in layers and executed in parallel.\u00a0<\/span><\/p>\n<p><span style=\"font-weight: 400;\">This is called \u201cthe topology of a neural network.\u201d\u00a0<\/span><\/p>\n<p><span style=\"font-weight: 400;\">The interconnections between the neurons are the synapse (or weights), just like in our body. We can train them with supervised algorithms. In the supervised training, the network knows the inputs and compares its actual outputs against the expected one.\u00a0<\/span><\/p>\n<p><span style=\"font-weight: 400;\">Kirshtein <\/span><a href=\"https:\/\/www.sciencedirect.com\/topics\/neuroscience\/neural-networks\"><span style=\"font-weight: 400;\">suggests<\/span><\/a><span style=\"font-weight: 400;\">:<\/span><\/p>\n<blockquote><p><span style=\"font-weight: 400;\">Errors are then propagated back through the network, and the weights that control the network are adjusted based on these errors. This process is repeated until the errors are minimized. It means that the same set of data is processed many times as the weights between the layers of the network are refined during the training of the network. This supervised learning algorithm is often referred to as a back-propagation algorithm<\/span><\/p><\/blockquote>\n<p><span style=\"font-weight: 400;\">Neural networks are popular and widely used in medical research, especially in bioinformatics for DNA sequence prediction, gene expression profiles classification, and analysis of gene expression patterns.<\/span><\/p>\n<p id=\"ukLHNjb\"><img decoding=\"async\" class=\"alignnone size-full wp-image-12921 \" src=\"https:\/\/www.invespcro.com\/blog\/images\/blog-images\/img_5e7c16338e875.png\" alt=\"\" \/><\/p>\n<h2><span style=\"font-weight: 400;\">Deep learning<\/span><\/h2>\n<p><span style=\"font-weight: 400;\">Deep learning is a specific approach used for building and training neural networks.\u00a0<\/span><\/p>\n<p><span style=\"font-weight: 400;\">The input data is passed through a series of nonlinearities or nonlinear transformations.\u00a0<\/span><\/p>\n<p><span style=\"font-weight: 400;\">In contrast, in most modern machine learning algorithms, the input can only go only a few layers of subroutine calls. The key thing here is the word layers.\u00a0<\/span><\/p>\n<p><span style=\"font-weight: 400;\">Following <\/span><a href=\"https:\/\/en.wikipedia.org\/wiki\/Geoffrey_Hinton\"><span style=\"font-weight: 400;\">Geoffrey Hinton<\/span><\/a><span style=\"font-weight: 400;\">, <\/span><a href=\"https:\/\/www.evl.uic.edu\/creativecoding\/courses\/cs523\/slides\/week3\/DeepLearning_LeCun.pdf\"><span style=\"font-weight: 400;\">Deep learning<\/span><\/a><span style=\"font-weight: 400;\">:<\/span><\/p>\n<blockquote><p><span style=\"font-weight: 400;\">the layers themselves are composed of multiple processing layers to learn the representations of data with multiple levels of abstraction. (\u2026) These layers of features are NOT designed by human engineers: they are learned from data using a general-purpose learning procedure\u201d. Therefore, <\/span><b>we cannot control the structure of deep learning architecture<\/b><span style=\"font-weight: 400;\">. Completely black box.<\/span><\/p><\/blockquote>\n<p><span style=\"font-weight: 400;\">Deep learning is actually a part of ML. Both deep learning and neural networks are not new concepts. They both existed for a long time, but people just didn\u2019t know how to train such networks. During the last 10 years, algorithmic improvements and the advances in hardware allow you to run small deep learning models on your laptop.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">The difference in the performance of those algorithms does not get worse as more and more data are given as input, which is the case for the traditional algorithms.<\/span><\/p>\n<h3>Deep learning in CRO<\/h3>\n<p><span style=\"font-weight: 400;\">Let\u2019s come back to the world of CRO and digital marketing. How can deep learning methods be used there?\u00a0<\/span><\/p>\n<p><span style=\"font-weight: 400;\">An intelligent chatbot that actually uses AI can help you in a fast and efficient way to provide better customer support or makes the sale process much shorter.\u00a0<\/span><\/p>\n<p><span style=\"font-weight: 400;\">Another way of using deep learning is through the massive amount of customer data through their personalized profile, personalized search, personalized clicks, and product suggestions. Ecommerce stores can implement \u201cpeople who bought this also bought\u201d feature through product tagging &#8211; however, deep learning can provide a much more interesting data set.\u00a0<\/span><\/p>\n<p><span style=\"font-weight: 400;\">Of course, both of these examples are not new. Chatbots existed long before; the same applies to google translators or digital assistants. But the quality of their performance left much to be desired before using deep learning methods.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">Speech and face recognition, image classification, and natural language processing helped these products take really great leaps forward.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">Nowadays, everyone claims to use AI, and the word \u201cAI\u201d helps to sell everything. While in reality, they usually use a variant of machine learning technology\u2026.<\/span><\/p>\n<p id=\"KVGvvMU\"><img decoding=\"async\" class=\"alignnone size-full wp-image-12922 \" src=\"https:\/\/www.invespcro.com\/blog\/images\/blog-images\/img_5e7c1655e5f98.png\" alt=\"\" \/><\/p>\n<h3>The trap of personalization<\/h3>\n<p><span style=\"font-weight: 400;\">There is a trap of too much personalization in ads, searches, social media contents, etc.\u00a0<\/span><\/p>\n<p><span style=\"font-weight: 400;\">It is called a &#8220;filter bubble&#8221; effect or the &#8220;echo chamber&#8221; described recently by <\/span><a href=\"https:\/\/arxiv.org\/abs\/1902.10730\"><span style=\"font-weight: 400;\">Jiang et al<\/span><\/a><span style=\"font-weight: 400;\">.\u00a0<\/span><\/p>\n<p><span style=\"font-weight: 400;\">In practice, it means that the user is continuously exposed to the same content defined by his\/her own previous choices. That narrows the potential products of interest &#8211; and in the long term, will have very adverse effects on future sales.\u00a0<\/span><\/p>\n<p><span style=\"font-weight: 400;\">It is also very annoying for customers because they see the same kind of ads or contents all the time. Their interest is not around only kids all the time just because they ordered the product for kids in the past.\u00a0<\/span><\/p>\n<p><span style=\"font-weight: 400;\">Have you watched a couple of videos on Facebook and noticed that your video feed is full of videos on the same topic?<\/span><\/p>\n<p><span style=\"font-weight: 400;\">The term of the &#8220;filter bubble&#8221; was actually first <a href=\"https:\/\/books.google.com.tr\/books\/about\/The_Filter_Bubble.html?id=-FWO0puw3nYC&amp;redir_esc=y\">coined<\/a> <\/span><span style=\"font-weight: 400;\">by Pariser. His vision was pretty scary:\u00a0<\/span><\/p>\n<blockquote><p><span style=\"font-weight: 400;\">Imagine a world where all the news you see is defined by your salary, where you live, and who your friends are. Imagine a world where you never discover new ideas. And where you can&#8217;t have secrets.<\/span><\/p><\/blockquote>\n<p><span style=\"font-weight: 400;\">But there is hope. Jiang et al conclude:\u00a0<\/span><\/p>\n<blockquote><p><span style=\"font-weight: 400;\">The best remedies against system degeneracy we found are continuous random exploration.<\/span><\/p><\/blockquote>\n<p><span style=\"font-weight: 400;\">According to the simulation results for the system dynamics, a user has to be continuously exposed to new content\/products &#8220;at least linearly.&#8221;<\/span><\/p>\n<h2><span style=\"font-weight: 400;\">Evolutionary Algorithms in Experimentation <\/span><b>\u00a0<\/b><\/h2>\n<p><span style=\"font-weight: 400;\">There is much more about AI usage in conversion rate optimization than just chatbots, personalized ads, or profiles.\u00a0<\/span><\/p>\n<p><span style=\"font-weight: 400;\">The power of science was always in its interdisciplinary character. So, CRO benefits and uses biology concepts such as evolutionary algorithms. These algorithms mimic the Darwinian natural selection process.\u00a0<\/span><\/p>\n<p><span style=\"font-weight: 400;\">There&#8217;s nothing better than the mechanism behind our evolution looking for constant improvement so that we are better and better adapted to the changing environmental conditions.\u00a0<\/span><\/p>\n<p><span style=\"font-weight: 400;\">As a result, &#8220;good genes&#8221; are becoming more prevalent, and &#8220;bad genes&#8221; are eliminated (or switched off).\u00a0<\/span><\/p>\n<p><span style=\"font-weight: 400;\">The same happens to bad or good algorithms.\u00a0<\/span><\/p>\n<p><span style=\"font-weight: 400;\">This is conducted via the so-called &#8220;fitness function,&#8221; which plays the role of an assessment criterion. The fittest individuals from a population survive to produce offspring. Their children inherit the parent&#8217;s characteristics, and these good features are shared by the next generation. Every generation will be better than the previous one and therefore have a higher chance of surviving. In the end, a generation with the fittest individuals continues to survive.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">There are five main stages in the evolutionary algorithm: (choosing) Population, Evaluation via Fitness Function, Selection, Crossover, and Mutation. In the first step, we choose a population characterized by a particular set of variables (genes).\u00a0<\/span><\/p>\n<p><b>Example: Suppose we want to consider 6 changes in website design.<\/b><\/p>\n<p><span style=\"font-weight: 400;\">Think of each gene as a single change in the website design. Genes are coded using binary 0\/1 notation (two possible alleles).\u00a0<\/span><\/p>\n<p><span style=\"font-weight: 400;\">In CRO, setting 1 will be a presence of a change in the website, and setting of 0 denotes an absence of that change in the site design. The string of genes is called a chromosome, and these are all the changes we want to consider (or not) in website design.\u00a0<\/span><\/p>\n<p><span style=\"font-weight: 400;\">Therefore the length of a chromosome is the number of all changes in website design we want to test.<\/span><b> In our case, the length would be 6.\u00a0<\/b><\/p>\n<p><span style=\"font-weight: 400;\">The gene that is responsible for a single change has a position on the chromosome. The gene has an allele that is either 0 or 1 representing the absence or presence of a change.<\/span><b>\u00a0<\/b><\/p>\n<p><span style=\"font-weight: 400;\">A single chromosome is one of the possible solutions to a problem. It is a combination of all changes in the website (present or not). The population is, therefore, a set of those solutions (combinations), a set of chromosomes.\u00a0<\/span><\/p>\n<p><span style=\"font-weight: 400;\">Each individual (chromosome) is evaluated using a fitness function. This function is a criterion that measures how good your solution is. In short, the fitness function is a function we want to maximize. Therefore, it should give low values for bad solutions and high values for good ones<\/span><b>.\u00a0<\/b><\/p>\n<p><span style=\"font-weight: 400;\">In a CRO setting, the fitness function could be just a conversion rate in a given time interval.\u00a0<\/span><\/p>\n<p><span style=\"font-weight: 400;\">But we can think of more specific criteria. The fitness function could be a conversion rate in a given time interval only for a particular subpopulation of all traffic etc. Based on the fitness function, each solution gets a fitness score. That score is a probability for the individual to be selected for reproduction.\u00a0<\/span><\/p>\n<p><span style=\"font-weight: 400;\">In the Selection stage, the individuals (combinations with the highest conversion rate) are chosen to be parents for the next generation. That process happens based on the probabilities calculated in the previous step using the fitness scores.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">In the crossover stage, the parents&#8217; genetic material mixes:<\/span><\/p>\n<p id=\"aLUUouT\"><img decoding=\"async\" class=\"alignnone size-full wp-image-12924 \" src=\"https:\/\/www.invespcro.com\/blog\/images\/blog-images\/img_5e7c168f89e99.png\" alt=\"\" \/><\/p>\n<p><span style=\"font-weight: 400;\">That means that a child inherits some part from one parent and the rest from the other. As a result, new possible solutions are created which have some changes in the website design present and some absent, similarly to their parents.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">Finally, in the mutation step, one or single genes might change from 0 to 1 or vice versa in an offspring. That means that with some probability, some changes would be switched off or on in a given combination. The mutation might or might not occur. We control it via the likelihood of mutation parameter.\u00a0<\/span><\/p>\n<p><span style=\"font-weight: 400;\">The population size is fixed. That means that the individual designs (possible solutions) with the worst fitness characteristic die, making the space for the newborn. That means that the combinations with the lowest conversion rates are eliminated. In every iteration, all steps are repeated to obtain a better generation (combinations with a higher conversion rate) than the previous one.\u00a0<\/span><\/p>\n<p><span style=\"font-weight: 400;\">If the next generation is not significantly different from the previous one, the process stops. We say that the algorithm reached convergence, and there is no sense to continue it since the algorithm produced the solution to our problem.<\/span><\/p>\n<h3><b>Can we not just simply test all combinations?\u00a0<\/b><\/h3>\n<p><span style=\"font-weight: 400;\">In a given example, there are 2^6 all combinations, which is not much (64). That might not be much for larger sites &#8211; you will probably need 2,000,000 visitors a month (at 1% conversion rate).\u00a0<\/span><\/p>\n<p><span style=\"font-weight: 400;\">But when we consider more changes on the website, the number of all possible solutions might be huge. For 20 changes, it is 2^20=1048576. The genetic algorithm allows us not to test all of them in a brutal force\u201d manner but searching for the best solution in a more efficient way.<\/span><\/p>\n<h2><b>Conclusion: ML or even AI will NOT do our job<\/b><\/h2>\n<p><span style=\"font-weight: 400;\">Genetic algorithms are a very flexible tool. You can specify more than one fitness function for each segment you want to find a winner in. And each segment can have different fitness function criteria. But even if a genetic algorithm can help us test many combinations of the existing variations of the website, this is our job to formulate those variations (genes). It will not evaluate the genes that are not given or invent a new one. You have to specify the fitness function(s) and final segments.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">The power of gene algorithms in disciplines such as image recognition is in its genes specification. Because genes could represent tiny areas of the images such as pixels, we do not need to specify the specific parts of the image to look at, based on the experts\u2019 knowledge (however, that might speed up the algorithm considerably). We might let the algorithm do all the jobs, and it will probably invent some relevant new areas, which were not considered before. To achieve the same effect in CRO, we need to then \u201cpixelate\u201d the problem so that the genes are as tiny changes as possible. Then that algorithm can build the solution out of those bricks pixels<\/span><\/p>\n<p><span style=\"font-weight: 400;\">Of course, \u201cthere is no free lunch,\u201d and working with only pixels in image recognition leads to very slow convergence. Therefore, people invent the whole method of efficient image segmentation. So each gene is instead not a pixel but a polygon &#8211; which can be circles, squares, rectangles, etc. Also, each gene encodes codes the color, position, or size of a polygon.\u00a0<\/span><\/p>\n<p id=\"cgGuybb\"><img decoding=\"async\" class=\"alignnone size-full wp-image-12919 \" src=\"https:\/\/www.invespcro.com\/blog\/images\/blog-images\/img_5e7c15816f9b1.png\" alt=\"\" \/><\/p>\n<p><span style=\"font-weight: 400;\">You can <a href=\"https:\/\/alteredqualia.com\/visualization\/evolve\/\">read more technical details<\/a> (and some cool pictures of Mona Lisa \ud83d\ude42 )\u00a0<\/span><\/p>\n<p id=\"JZDnFzO\"><img decoding=\"async\" class=\"alignnone size-full wp-image-12920 \" src=\"https:\/\/www.invespcro.com\/blog\/images\/blog-images\/img_5e7c15cd221f1.png\" alt=\"\" \/><\/p>\n<p><span style=\"font-weight: 400;\">Do we have the same problem in CRO? Very unlikely, unless one wants to test every single word appearing on the website. So there has to be a balance between the number of potential variations to test and the time to get the results. The fact that we have the genetic algorithm now does not mean we need to test \u201ceverything.\u201d The AI methods were invented to help experts, not to replace them.<\/span><\/p>\n","protected":false},"excerpt":{"rendered":"<p><span class=\"span-reading-time rt-reading-time\" style=\"display: block;\"><span class=\"rt-label rt-prefix\">Reading Time: <\/span> <span class=\"rt-time\"> 14<\/span> <span class=\"rt-label rt-postfix\">minutes<\/span><\/span>Disclaimer: This section is a TL;DR of the main article and it\u2019s for you if you\u2019re not interested in reading the whole article. On the other hand, if you want to read the full blog, just scroll down and you\u2019ll see the introduction. In his famous book Machine Learning, Tom Mitchell defined ML as \u201cthe [&hellip;]<\/p>\n","protected":false},"author":54,"featured_media":12950,"comment_status":"open","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[36],"tags":[],"class_list":["post-12918","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-cro"],"_links":{"self":[{"href":"https:\/\/www.invespcro.com\/blog\/wp-json\/wp\/v2\/posts\/12918","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.invespcro.com\/blog\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.invespcro.com\/blog\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.invespcro.com\/blog\/wp-json\/wp\/v2\/users\/54"}],"replies":[{"embeddable":true,"href":"https:\/\/www.invespcro.com\/blog\/wp-json\/wp\/v2\/comments?post=12918"}],"version-history":[{"count":1,"href":"https:\/\/www.invespcro.com\/blog\/wp-json\/wp\/v2\/posts\/12918\/revisions"}],"predecessor-version":[{"id":97940,"href":"https:\/\/www.invespcro.com\/blog\/wp-json\/wp\/v2\/posts\/12918\/revisions\/97940"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.invespcro.com\/blog\/wp-json\/wp\/v2\/media\/12950"}],"wp:attachment":[{"href":"https:\/\/www.invespcro.com\/blog\/wp-json\/wp\/v2\/media?parent=12918"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.invespcro.com\/blog\/wp-json\/wp\/v2\/categories?post=12918"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.invespcro.com\/blog\/wp-json\/wp\/v2\/tags?post=12918"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}