5 Ways To Leverage AI Research In Marketing & Advertising

Last updated: 05-04-2021

Read original article here

5 Ways To Leverage AI Research In Marketing & Advertising

The Best of Applied Artificial Intelligence, Machine Learning, Automation, Bots, Chatbots
5 Ways To Leverage AI Research In Marketing & Advertising
April 23, 2020
by Mariya Yao
Today, Artificial Intelligence (AI) can drive effective solutions to lots of marketing challenges. It can help you optimize your advertising budget, personalize your customers’ experience, suggest the most accurate attribution model, boost your marketing analytics, and even generate original brand content, including images and advertising slogans.
However, AI is evolving so fast these days that even the most advanced marketing teams can find it challenging to follow all the research advances in the field and fail to keep track of all possible business applications for the novel AI approaches.
If you’re an intermediate or advanced practitioner, we’ve created premium education to help you and your team master recent breakthroughs in applied AI for marketing. Click here to purchase our AI in marketing research summaries . 
In this series we focus on five key challenges for enterprise marketers:
State-of-the-art Approaches to Marketing Attribution
Before prospects become your customers, they usually interact with you via multiple touchpoints – they click on your posts in social media, find your website via Google search, go back to your website later, subscribe to your newsletters. So when they finally convert, how should you attribute this conversion and which of the touchpoints should you credit? The answers to these questions have a detrimental impact on your future advertising strategy.
Actually, there are quite a lot of approaches to user’s attribution – you can give all the credits to the last click or to the first click or divide the credits equally between all the touchpoints. These are just some of the most popular basic attribution models. However, the latest research focuses on more effective, data-driven approaches to attribution modeling.
For your convenience, we’ve summarized the most interesting breakthroughs in marketing attribution :
Researchers from a major Chinese university introduce a novel approach to multi-touch attribution that borrows from survival analysis and uses hazard rate when modeling the effect of an ad exposure upon the conversion.
The research team from a large media investment group addresses the attribution problem by decomposing and allocating marginal contributions to the coefficient of determination of regression models.
Another group of researchers introduces the way to capture sequential user patterns with a recurrent neural network.
The research team from a major advertising platform proposes a way to incorporate attribution modeling into the bidding strategy for more efficient bidding on advertisement platforms.
One of the leading Chinese eCommerce firms introduces a causally-driven approach to multi-touch attribution using a recurrent neural network.
The research team from a leading tech company offers a novel LSTM-based approach to solving the attribution problem.
Finally, the authors from the well-known research institute take a systematic approach towards measuring attribution in online advertising and introduce a new metric for attribution that is theoretically grounded and can be easily computed.
How to Optimize Your Marketing & Advertising Campaigns
As ads are getting very expensive and effective marketing channels are becoming more and more crowded, optimization of marketing and advertising expenses turns into the number one priority for most marketers. Fortunately, artificial intelligence offers lots of working solutions for optimizing marketing campaigns.
To help you drive the efficiency of your marketing efforts, we’ve summarized a number of state-of-the-art approaches to optimizing advertising campaigns :
The joint group of researchers from several top tech companies shows how an algorithm that won several Kaggle competitions can accurately predict click-through and conversion rates in a real-world production system.
A leading e-commerce giant introduces an algorithm that captures a user’s interests as well as the dynamics of these interests to improve the accuracy of click-through rate predictions in online advertising.
The researchers from a major university propose a novel approach to click-through rate prediction that enjoys the power of deep neural networks but with much lower computational costs.
Another research team from a tech giant suggests a novel approach to optimizing marketing campaigns by drawing upon causal inference, uplift modeling, and multi-armed bandits.
The famous tech leader addresses the key challenges of sponsored search by introducing a complex advertising system that is based on various machine learning techniques.
The authors from another well-known tech company address the problem of attracting prospective customers with online display advertising.
A group of China-based researchers investigates the best way to assign the right ad to the right user, while the number of ad slots and their locations is changing over time.
And finally, a leading eCommerce company from China suggests a unified framework for marketing budget allocation in online business.
Visual product discovery (example from Pinterest)
Techniques for Personalizing Your Customer Experience
Customer experience personalization is an important driver of customer satisfaction as well as customer lifetime value for the company. That’s why top performing businesses take the problem of implementing a really effective recommender system very seriously. There are a number of existing approaches to providing valuable recommendations to the customers, including collaborative filtering, clustering algorithms, deep neural networks, and others. The task of the marketing team is to choose the approach that will be the best fit based on the company’s needs and available data.
To help marketers stay aware of the latest research for better-informed decision-making, we’ve summarized the key research advances in AI-driven personalization from the top tech companies and research institutions:
The joint group of researchers from tech giants introduces a novel approach to collaborative filtering with variational autoencoders.
The research team from a leading e-commerce company shows how to stabilize reinforcement learning algorithms so that they could be used for building online recommendation systems in a real-world setting.
Another group of researchers stresses the importance of explaining the recommendations to the users and demonstrate the positive effects of such explanations.
The research team from a French insurance company suggests an approach to modeling the customer online journey with convolutional and recurrent neural networks, enabling the generation of realistic scenarios of consumer behavior.
The joint group of researchers from European universities introduces a comprehensive survey of the sequence-aware recommender systems.
The researchers from a leading social media platform show how personalization can be achieved by combining perspectives from recommendation systems and predictive analytics.
Finally, the authors from a well-known research institute address market basket prediction by considering the complete shopping histories of all available customers.
Rebecca Minkoff’s smart dressing-room
Key AI Research Advances for Improving Marketing Analytics
Knowing your customers and understanding their interactions with your product is fundamental for building effective marketing campaigns. No doubts you spend lots of time and efforts on market research to identify the key segments of your customers and learn their opinion about different aspects of your product.
However, you should know that there are many AI-driven tools that can automate your marketing activities and significantly improve your marketing analytics and insights. For your convenience, we have summarized several research papers that cover the latest advances in sentiment analysis, customer clustering and capturing information from social media images :
The researchers from a major Chinese university show how to improve targeted aspect-based sentiment analysis by incorporating commonsense knowledge into the deep learning model.
The research from a top university demonstrates that convolutional neural networks can be very accurate and efficient in aspect-based sentiment analysis.
Another group of researchers introduce a novel progressive self-supervised attention learning approach for aspect-level sentiment classification.
The joint group of researchers from Germany introduces a very interesting approach to image captioning with a specific focus on marketing needs.
The paper from a major research institution demonstrates the state-of-the-art approach to clustering.
The authors from a famous tech giant introduce their approach to predicting customers’ future lifetime value.
Finally, the paper from another research team demonstrates how Bayesian networks can assist in reducing the number of questions in the market research questionnaire.
Generating Marketing Content With Adversarial Learning
In the last part of our AI for Marketing series, we are featuring the most important research advances related to generating images , videos , and text for marketing and advertising.
Do you need a virtual model for your advertising campaign? Do you want to get some ideas for a new logo? Or what about generating videos from the captions provided in the marketing brief? All of these tasks have working AI-driven solutions!
To help you stay informed about the latest research breakthroughs in automated content generation, we are summarizing a number of exciting research papers covering Generative Adversarial Networks (GANs) applied in marketing:
Researchers from Belgium introduce a novel approach to synthesizing person images in arbitrary poses.
A well-known company working in hospitality service uses GANs to generate successful product listings.
The research team from a top tech company explains how GANs can help you create videos from captions.
Another research team introduces an interesting approach to generating realistic images that match a given text description.
The researchers from a well-known tech giant show how the recent advances in self-supervised and semi-supervised learning can be leveraged to significantly reduce the amount of labeled data required for natural image generation.
The authors from a leading eCommerce platform  introduce their approach to hyper-personalization of homepage banners.
And finally, researchers from Netherlands explore the ways to generate logos conditioned on color.
Leverage Leading AI Research In Marketing & Advertising
Our AI for Marketing series provides you with the latest research in marketing automation. We have carefully curated and summarized the key research papers for AI-driven marketing solutions to help you:
build a data-driven marketing strategy;
explore new ways of creating optimized advertising campaigns;
personalize your customer experience;
improve your marketing analytics and insights;
generate your marketing content using AI.


Read the rest of this article here