AI for marketing will rule the world of marketing. It is happening as you read those lines. Ask Google Ads & Facebook Ads.
The transition to using machines to decide best marketing practices is happening very fast. More than you may notice or know!
Artificial Intelligence (A.I.) Marketing is a new concept that leverages customer data and AI concepts to create content tailored for each customer. It is gaining momentum and more marketing experts will talk more about it.
AI Marketing simply improves the customer experience and increases conversions and sales. (Usually without human intervention!)
One of the many books that introduces AI for marketing is:
AI for Marketing and Product Innovation: Powerful New Tools for Predicting Trends, Connecting with Customers, and Closing Sales (Amazon link)
(AI is the computer science that emphasizes the creation of intelligent machines that work and react like humans. ML is the data analysis that automates analytical model building)
AI for Marketing and Product Innovation is a helpful guide to the world of AI and ML, written by 3 experts in big data, machine learning, AI, Marketing and Sales.
While the book gets technical about the technologies and their core functions, it also offers easy and insightful case studies.
AI for Marketing and Product Innovation contains 15 chapters and is certainly a fascinating read all throughout.
Here’s your summary of AI for Marketing and Product Innovation.
Chapter 1: Major Challenges Facing Marketers Today
The key point of marketing is to effectively reach, attract and retain consumers. These are inherent resource allocation challenges that marketers face today. Marketers shouldn’t quibble over whether to use AI, the decision should be about which AI solution / implementation to utilize. A.I. is a core component of marketing today.
The utilization of AI and ML (Machine Learning) algorithms help to predict outcomes and answer questions such as which demographic of website visitors is likely to buy that new line of socks? Reduce dimensionality, understand language, cluster and classify.
Marketers need to know how to use AI to their advantage by asking the right questions. Knowing what they are looking for and programming accordingly.
Marketers will have to be more data-centric, more precise when targeting people, and know which aspect of AI & ML to tap into.
AI is only as good as the domain knowledge embedded within. Choose your data wisely.
“Within a few years, it is predicted that most employers will require some degree of AI and ML proficiency. However, it is still a rare skill at this time, and marketing professionals who gain a working grasp of AI and ML will have a strong competitive edge.”
2- Introductory Concepts for Artificial Intelligence & Machine Learning for Marketing
The foundation of AI is rule-based systems with ML. Rule-based systems can contain logic and change the information embedded within it as it is commanded.
These systems are usually comprised of IF-THEN logic, an example would be that if the client is interested in this, then the client may be interested in that, too.
Rule-based systems will be very important for providing buying suggestions to users. ML systems are inherently probabilistic based on past data.
Other compelling foundational concepts include:
interference engines, heuristics, hierarchical learning,
stimulus response learning, chaining, verbal association,
discrimination learning, concept learning, rule-learning and problem-solving.
Additional foundational concepts in this field include the aspect of expert systems, big data, data cleansing and data addition.
It is important to understand that machine learning (ML) is an application of AI where the machine can learn and improve as it goes.
Marketing professionals will want to use ML in their day-to-day marketing.
It will allow them to put forth dynamic messages that change regularly to attract and convert customers thereby reducing costs and increasing effectiveness.
3- Predicting Using Big Data – Intuition Behind Neural Networks & Deep Learning
Computers are as competent as the creator, even more so if programmed correctly but even computers are not perfect.
They have difficulties doing simple things such as reading poor handwriting, carrying on a reasonable conversation.
Thankfully, this is changing with the use of neural networks. Neural networks are built similar to how neurons work in the human brain.
Thus making it simpler for computers to think as humans do.
AI training through artificial neural networks have grown to reinforce concepts in the “mind” of the AI machine.
This progress is evidenced through deep learning and the momentum of programs such as AlphaZero which show that the computer is learning, improving and figuring out different ways to become better at games such as Chess and Go!
In late 2017, the Alphabet-owned artificial intelligence research company DeepMind introduced AlphaZero, a single system that taught itself from scratch how to master the 3 games of: chess, shogi (Japanese chess), and Go, beating a world-champion program in each case. –Wikipedia
4- Segmenting Customers & Markets – Intuition Behind Clustering, Classification, & Language Analysis
Through supervised an unsupervised learning, AI can differentiate data, cluster the data and classify the data. This process is important because it makes the data make sense and valuable for marketers to use in their campaigns.
The intuition behind segmenting, classification and clustering can be a bit difficult to comprehend but the main point is that with the right data scientists and expert matter in the field, finding, relating and to differentiating between customers can be simpler with AI & ML.
5- Identifying What Matters Most. Intuition Behind Principal Components, Factors & Optimization
Concepts such as linear algebra, matrices are important in representing the transformations of data and movements in space. We also run into topics such as the Principal Components Analysis (PCA), which helps to understand things better by being smarter.
We also learn in this chapter that rule based logic is great until you introduce unknowns or variables that were not in the original data set.
Rule based logic is important.
It serves as the foundation and will be useful in a variety of situations but adding the layer of fuzzy logic is more important.
Fuzzy Logic helps the machine to deal with unknowns, or variables and inputs not in the data set. Fuzzy logic helps to interact with probabilistic situations.
Finally, it is important to be able to design AI to evolve to have similar plasticity as the brain and this is being done today through genetic algorithms.
Yet, the AI FOR Marketing book authors highly recommend programming in MATLAB.
Learn how to use MATLAB and its application packages for statistical analysis, machine learning, deep learning, and neural networks.
6- Core Algorithms of Artificial Intelligence & Machine Learning Relevant for Marketing
Core Algorithms of AI and ML are comprised of several key components These include supervised learning, unsupervised learning, and reinforcement learning.
Supervised learning is where the machine is essentially spoon-fed the instructions and the answers. One has to handhold the machine the entire time. Supervised models are very helpful in classification and regression problems.
Unsupervised learning includes:
association (think affinities in marketing),
clustering (division of data into usable groups), and
dimensionality reduction (distilling information to understand the variables that explain all the variance in the data).
Reinforcement learning is comprised of learning control, adaptive control, and optimal control. Reinforcement algorithms can be coupled with a Markov Decision Process and quantitative approaches to be utilized in practical applications such as predictive customer service.
A Markov decision process (MDP) is a discrete time stochastic control process. It provides a mathematical framework for modeling decision making in situations where outcomes are partly random and partly under the control of a decision maker. MDPs are useful for studying optimization problems solved via dynamic programming and reinforcement learning. -Wikipedia
7- Marketing & Innovation Data Sources & Cleanup of Data
Data should be collected with the consent of the consumer.
The supply chain of the data must have as much scrutiny as a supply chain in manufacturing.
Data sources of ML & AI for automated marketing can come from various places, a few of them are retail data, online sales data, social media data, loyalty card data, consumer financial data, voting and demographic data, economic index data, Google CPC data and focus group data, in addition to a slew of other sources.
Now that we know where the data is coming from, the next step is to understand the cleanup process. Unclean data yields wrong conclusions, unclean data is usually comprised of missing and random data, and missing but patterned data.
Understanding these problems and fixing them by completing consumer purchase data, filling in geospatial data, normalizing temporal scales across data, eliminating seasonality from the data, and normalizing data across different ranges will help to yield proper results.
8- Applications for Product Innovation
Product innovation with AI & ML will be much more streamlined.
There will be no need for trend clinics or focus groups, algorithms will be able to read between the lines of searches and content in various data sources, product innovation, pricing, and product value packaging will be done in an algorithmic fashion.
Marketing professionals can start with time-series data to view trends and then progress further by understanding non-conscious data such as music preferences, tv addictions, YouTube channels consumed, followed and liked in the wee hours of the night.
Then progress even further by looking at conscious data (tweets, Facebook posts, call center data, and consumer survey data).
The fact of the matter is that data sets are present to bring about powerful product innovations, the challenge is fine tuning the tools of ML & AI to produce powerful results.
A product innovation process through AI & ML will include a nine-step process:
1. identification of metaphors,
2. separate dominant,
3. emergent, fading, and past codes from metaphors,
4. product contexts in the non-conscious mind,
5. parsing of conscious contexts to extract concepts,
6. generation of product concept ideas based on combinations,
7. validation and prioritization of product concepts,
8. the creation of algorithmic feature and bundling options, category extensions and adjacency expansion,
9. luxury extension identification.
“The fact is, we trust algorithms to deliver what most of us want foremost in life: a human partner, whether that’s a friend or a lover. If algorithms can help guide us expertly through the romantic jungle, it’s a safe bet that they can power our search for innovations in product development and marketing!” -Book quote
9- Applications for Pricing Dynamics
The utilization of algorithms will pave the way for algorithmic pricing. This will include built in heuristics-based models for pricing, pricing dynamics with the integration of control system theory, dynamic pricing, real time volume estimators, real-time discounts based on protocols.
Pricing is based on a variety of factors, including macro and micro economic forces, consumer demand and psychology.
Pricing is an interesting mechanism because it delves into qualitative and quantitative aspects to come up with an optimal result.
Optimal pricing includes an aspect where one portrays the value proposition such that the pricing becomes negligible in the decision making.
Data inputs in dynamic pricing may include: weather changes, average price, competitor price, consumer demand, loyalty, immediacy of demand, social media output and consumer confidence. Machine learning will be of use in this regard due to its dynamic nature.
Read also: This Is Marketing book summary
10- Applications for Promotions & Offers
Promotions and discounts are important in marketing. Both provide the potential buyer with a sense of urgency to take advantage of an offer that will only last for a while.
ML and AI for marketing will be effective in promotions and offers when the appropriate amount of data, creative control, and the ability to execute is a key part of the picture.
For ML to be applied appropriately, the timing of the promotion, the path to purchase, algorithmic choice of language, context scoring, and dynamic pricing structures are all key aspects of the equation to consider.
Google AdWords and Facebook ads platforms (for example) already use artificial intelligence and machine learning to target people more likely to make the advertiser’s desired action.
Phrasee and Persado are some samples of online AI Marketing tools that apply AI to content and ads creation and email marketing. They claim to increase sales and profits of marketing campaigns managed by their services by ~50%.
Adext is an Audience Management as a Service (AMaaS) that uses AI & machine learning to handle and optimize online ads on platforms like Google AdWords and Facebook Ads. (At the time of writing this post, Adext was still in Open Beta.)
Adext tests different audiences for each ad to detect the audience most likely to make the desired action. In doing so, Adext identifies and learns which platforms are most profitable. Then, it channels the advertising budget toward them.
* * *
Read also: Marketing 4.0: Moving from Traditional to Digital Book Summary
Keywords: ai for marketing, ai marketing