5 Practical AI Techniques for Improving Customer ROI

Seventy-seven percent of marketers believe that real-time personalization is crucial, yet sixty percent struggle to personalize content in real time.

Personalizing content in the form of giving the right recommendations for each user, scheduling the content based on each user’s behavior, and then selecting the right communication channel require processing large amounts of data to understand customer preferences at a personal level. This strategy is accomplished most effectively and efficiently using artificial intelligence (AI) systems.

AI systems process large streams of data in real time and develop models that understand customer intents and preferences. For instance, AI can be used to score users on their likelihood to churn or purchase in the near term. It can understand a customer’s propensity toward various categories, balance content freshness with popularity, and recommend the next best content or product for every customer. It can also interpret the data to understand the optimal time and channel to engage each customer.

Here are the top five practical AI techniques for improving customer ROI:

1) User Intent Predictions

Modeling user actions on your website or app enables you to predict behaviors and attributes that are correlated with near term actions, like purchases or churn.

For instance, an e-commerce website with a 5% session conversion rate has 95% of all sessions abandoned. But all abandoned sessions are not alike. And it’s imperative that marketers understand the true composition of this group. Typically, some are serious, likely buyers, while the rest are casual visitors who are just browsing your site. A predictive engine, can help you build a model that separates your likely buyers – that can often comprise 25% of this group – from the rest. These high intent users are typically more than 2 times more likely to respond to email messages than the average recipient and yield 7-12 times the ROI from paid advertising such as display retargeting.

2) User Affinity Predictions

Unlike user intent (which is near term), user affinity models give you an idea of the user’s persona and lifetime value. It is done by using the concepts of affinity marketing to categorize users into affinity groups based on their demographics, preferences and behavior.  Using these attributes, one segments customers by the product categories and brands they prefer, similarity in attributes to known customers such as preferences for certain authors or their preferred price bands for specific types of products.

3) Product and Content Recommendations

Once you know the right set of users to target (e.g., high value users with high purchase intent), you need to understand the right content or product selection for each user. Techniques such as collaborative filtering, which makes predictions about a customer’s preferences based on the preferences of similar customers, and unsupervised clustering, which uses data analysis algorithms to find “hidden” patterns or groupings in customer data sets, can help determine the right set of products for each user.

4) Personalized Promotions and Offers

Since promotions directly impact the bottom line, you should not only model who will be receptive to the promotions, but also drive a change in a user’s behavior by offering the promotion. The former is known as affinity modeling or response modeling, while the latter is known as uplift modeling. In uplift modeling, you try to find users who would not have transacted with you without an offer and—from among these users—find the ones who have a high likelihood of responding to your offer.

5) Creative and Model Optimization

When you have a set of “always on” running campaigns, it’s important to set aside a budget for exploring new ways to auto-optimize creative or data science models using a challenger and champion paradigm.

Traditional A/B testing models offer a quick path to finding an initial set of champions among creatives and data science models. You can go beyond that by using Bayesian optimization algorithms that test a new set of challengers against current winners. Auto optimization platforms can do this every time a new variant is added to the system by running these tests automatically to yield results that optimize return while minimizing variability. This is similar to the concept of efficient frontier in portfolio theory.

You Need an AI-Powered Marketing Platform with Access to Customer Data

Marketers can unlock the full potential of AI by using a platform that gets advanced access to your customer data and applies these AI techniques to improve your marketing campaigns throughout the buyer’s journey. Learn more about the Blueshift Platform here.

7 Key Programs Online Retailers Must Launch to Drive Customer Engagement

7 Key Programs Online Retailers Must Launch to Drive Customer Engagement

Engaging today’s customers is harder than ever. According to a recent Marketo report, the majority of buyers expect all their interactions with a brand to be personalized — and they think that marketers can enable this if and only if they have a deeper understanding of their unique needs.

Many forward-thinking online retailers have already revamped their customer engagement strategies to cater to customer demand for personalized engagement. Take Amazon, for example – the e-commerce giant keeps winning on customer satisfaction because of its powerful individual customization strategies.

To be able to engage like Amazon does, you need to determine the right methods to employ to drive successful and sustainable customer engagement.

7 marketing programs that ignite customer engagement:

#1 Abandoned-Browse Recommendations Using Collaborative Filtering

The goal is to engage a prospective buyer who dropped by your online store — someone who searched for and browsed some tops and dresses, for example — but left without buying anything. Using collaborative filtering, you should email her within a day to show her similar tops and dresses that other customers liked. This will encourage her to go back to your online store to browse some more and, hopefully, make a purchase.

#2 Post-purchase Recommendations Using Collaborative Filtering

Never stop engaging a new customer. Encourage her to buy again. Collaborative filtering allows you to make recommendations based on an item she already purchased. Within three days of her receiving her shipment, show her similar or related products that she might be interested in based on product choices of other purchasers of the same item.

#3 Category Affinity

Offer your shoppers choices from a category of products or services that interest them. For example, you might have noticed that a shopper keeps searching for nursing wear and related items. You can take advantage of this affinity by recommending popular content from this category. Send her weekly updates to keep her engaged.

#4 New and Trending Content from Relevant Categories

Another proven strategy to engage customers is to notify them of newly added items and items that are converting the best in their areas of interest. Send them weekly “what’s hot” emails to showcase new and trending products that they might want to purchase. This strategy can also be effective for welcoming and engaging new buyers.

#5 Re-Targeting

What if your buyers recently searched and added some items to their cart but did not complete their purchase? Within three days of their activity, send them push notifications or emails to encourage them to complete their purchase. Show them again the items they wished to buy and similar items from the same category that might interest them.

#6 Replenishment Reminders

For products that get re-purchased regularly, such as staples and grocery items, remind customers to “buy it again.” Employ purchase frequency analysis to determine when to best send buyers replenishment reminders. Amazon does this very effectively for a range of items from groceries to batteries

#7 Content Update Alerts

Your buyers, particularly those who subscribe to “back in stock” alerts, would love to know about any changes in price and availability of their “favorite” items. Are they back in stock? Is their favorite item on sale? How many items are left? Alert them in real time as content updates happen.

Grow Your Audience using AI

These seven key programs will help you do more than just sell your products — they will help you grow your customer base and build loyalty. However, you need to leverage the right technology platform to be able to implement these programs effectively. Learn more about using Blueshift’s solution for retail and e-commerce and how other companies have benefitted from this.

To learn more, check out our comprehensive guide here for everything you need to know about the subject.


Top 3 Reasons wy Mrters Struggle with Data. Learn what they are from Blueshift and learn how to overcome them.

Why Do Marketers Struggle with Data?

Our recent study shows that activating data is crucial to successful artificial intelligence (AI). However, 92% of the companies we surveyed said that they struggle with one or more of either data access, unification or analysis that prevents them from making better use of their customer data (see figure 1).

Old news to you? Perhaps, but the ever-increasing importance of data in marketing will ensure that those who cannot harness their data are going to be left behind. The problem becomes clearer when we drill down into each of these areas to better understand the situation and look for potential solutions.

Figure 1: Top challenges faced by  marketers

Data Access

Data access is the first and most fundamental step to working with data. It gives marketers the ability to use data from multiple sources in near real-time. Today’s marketers need quick access to data stored or generated from multiple systems both inside and outside their organizations. Here are some examples:

  • Customer record data from the CRM system
  • Purchase data from e-commerce and traditional point of sale systems
  • Campaign response history from the marketing system
  • Customer service data from the support system
  • Visitor and browsing data from the website
  • Ad clicks from advertising networks and social media networks
  • Social media activity and interactions from those systems

With so many sources of data, it’s not surprising that 46% of all marketers consider data access one of their top challenges. Besides the numerous sources of data, two other factors impede data access:

  1. Systems that are monolithic and focused on specific tasks but not designed for easy data access. Legacy systems, in particular, were designed to work as fully contained units where information extraction was episodic and infrequent rather than continuous.
  2. Organizational processes that emphasize guarding the data and limiting access to it rather than sharing it. Usually IT departments serve as guardians and gatekeepers to data access.  In our study, marketers who were given advanced access to data were 1.6x more likely to be using a majority of their customer data over those who had to go through IT to get access to this data.

All of this results in balkanized data that is hard to get to and difficult to access in real-time.  Solutions such as data warehouses and data lakes were designed for a “store and analyze” approach which is ineffective in today’s real-time environment.

Data Unification

Data unification refers to the ability to bring together data from multiple sources and connect them together. A unified view of customer data is a good example of this because it enables marketers to better engage with customers by offering them personalized experiences on not only their profiles (gender, preferences, purchases, etc.) but also their behavior (website visits, catalog views, mobile phone activity, current geographical location and social media activity).

McKinsey & Company refer to this as a 3D-360° customer view (see Figure 2) that comprises data from many sources that get updated often and needs to be used in almost real-time to see its value.  41% of marketers in our study considered this to be one of their top challenges.

Figure 2: 3D-360° View of Customer

Data Analysis

Analysis is perhaps the hardest of these functions so it’s probably no surprise that 54% of all respondents in our study considered this to be a top challenge.  As a 20-year data-driven marketer, I attribute the data analysis challenge to three things: lack of a strategic approach to the analysis, missing data analysis skillset in many marketing teams, and deployment of the wrong technology.

It all starts with strategy.  Very often marketing teams jump into data analysis expecting to dive into the data in the hopes of an answer revealing itself.  And very often, such efforts take a lot of effort and don’t produce the results expected.  It’s the “dumpster diving” of data analysis. Instead, it is far more effective to start with one or more hypotheses, frame the questions you want to answer and then start the analysis.

There’s no substitute for having the right people. Data-driven marketing has been the domain of a relatively small number of people and good talent is hard to find.  This leads to marketing leaders confusing people who know how to create reports and run operations with those who can effectively frame the problems they are trying to solve and use analytical techniques to solve those problems

Use the appropriate technology to solve the problem rather than using the latest technology hoping to get the answers.  This starts with setting up the problem correctly, identifying use cases and then using the right technology that collects and helps analyze the data.

McKinsey & Company go beyond these three general areas to identify several other more specific issues in their recent report, Ten Red Flags Signaling Your Analytics Program Will Fail.

The Way Forward

Data has been called the “fuel of the new economy.” And just like during the oil rush of the late 19th century, everyone’s drilling. Smart organizations will take a thoughtful and fresh approach to getting the most out of their customer data by:

  • Opening up access to the data so marketers can utilize most of their customer data
  • Providing a unified view of the data by dynamically accessing the data from multiple systems rather than trying to consolidate everything in a single system
  • Analyzing the data by first developing a strategy and process for developing hypotheses and testing each hypothesis by asking the right questions.

Learn how to overcome your customer data struggles in this new independent research, “Activating Customer Data for AI-Powered Marketing”. Get your copy now.

Download Blueshift's report on the state of AI, Marketing, and Customer Data - lots of marketing stats

Latest independent research commissioned by Blueshift finds Marketers see the huge potential of AI, but the data remains a challenge

Marketers see the huge potential of AI, but the data remains a challenge


Download Blueshift's report on the state of AI, Marketing, and Customer Data - lots of marketing stats

Since the advent of the personal computer, no technology has captivated businesses in the way artificial intelligence (AI) has. Its influence is expected to span industries and job functions — and marketing is right in the thick of it.  

To learn and quantify what marketers are doing with AI today, understand their challenges, and hear their aspirations, we teamed up with TechValidate (a SurveyMonkey company) to study the use of AI in marketing. We surveyed 200 marketing executives and practitioners from 198 business-to-consumer companies to understand their current experiences with AI technologies and their plans for the future. The results are in our report: Activating Customer Data for AI Powered Marketing. Here are some highlights.

The use of AI in marketing remains rudimentary, but interest is high

Though more than 80% of marketers are using some form of AI, few have deployed advanced capabilities. Today’s AI use is largely focused on audience expansion and targeting using techniques like “lookalike expansion” — finding new audiences by targeting those with traits similar to existing audiences — on the large advertising networks.

AI in Marketing

Use of AI in Marketing Today

However, only 6% of all respondents were using collaborative filtering (automated predictions about user interests) and predictive modeling techniques (forecasting outcomes) and only 16% used advanced segmentation technologies like predictive affinities on their own data to market more effectively and precisely to customers.

Advanced AI Techniques

Use of advanced AI techniques

Of all marketers surveyed, 64% want to expand their use of AI in both experimental trials as well as production campaigns within the next 12 months.

Marketer-controlled access to data is key to unlocking AI potential

We were surprised to learn that marketer-controlled advanced access to data plays an important role in the sophisticated use of AI. We found that marketers who control their own data access, without having to go through IT, activate and use more of their customer data to drive campaigns. Those who had such access and control were 1.6 times more likely to be using a majority of their customer data in their AI-driven campaigns.   

Proportion of marketers who are using most of their customer data to run AI-driven campaigns

In addition, with advanced access to more of their customer data, marketers were two to three times more likely to use sophisticated AI techniques like collaborative filtering and predictive modeling.  

Most marketers are still struggling with customer data 

Effectively using a company’s own customer data (first-party data) for AI-driven marketing campaigns is perhaps the biggest hurdle facing marketers. Almost all respondents (92%) in the study identified one or more of three factors — access, unification, or analysis — as a major challenge, resulting in a majority of marketers using less than 50 percent of their own customer data.

Top challenges hindering data use

Activation of customer data correlates with company revenue

We asked respondents about their companies’ revenue performance in the most recent fiscal year and looked at patterns in their marketing as it related to the use of AI. We were surprised to see that companies in which marketers activated and used more than 75% of their customer data were 1.4 times more likely to have exceeded revenue targets than those that had not.

Respondents that exceeded revenue goals

The results of the study are enlightening. They show that AI is inextricably tied to the activation of data, which itself is connected to several other aspects of data in an enterprise. The report’s data and insights are organized into seven key findings along with four recommendations that every business-to-consumer marketer looking to use AI should read.

Download Blueshift's report on the state of AI, Marketing, and Customer Data - lots of marketing stats

Get the full report here

Blueshift's AI-powered marketing platform solves for The 3 Is of AI in Marketing: More Than Just Intelligence

The 3 “I’s” of AI in Marketing: More Than Just Intelligence

Watch this rare webinar with Analysts from Forrester Research and VentureBeat hosted by Blueshift about getting the most out of your customer data with AI

This is the third and last post of our 3-part series on the topics covered in our joint webinar with Forrester Research and VentureBeat on AI-Powered Marketing. In part 1, we focused on the primary problem facing many marketers: the explosive amount of customer data being generated and an inability for marketers to use much of it. This has led many to a search for new solutions and technologies, such as AI, to help process this marketing data in real-time to create and execute more effective campaigns. However, Rusty Warner of Forrester Research, advises us in part 2 to not rush into a decision without first considering three crucial elements – strategy, organization and technology – when planning the adoption of AI. We now end the series by summarizing the webinar discussion and Q&A session on what one should look for in an AI technology for marketing.


Don’t Lose Sight of the Problem to be Solved

In any technology evaluation, it’s important to always keep in mind the problems you are trying to solve and how best the technology will solve those problems. Our webinar attendees answered several questions that indicated that they had three major overarching problems that they are expecting an AI system to solve:

  1. Generating copious amounts of behavioral data about their customers, but using very little of it.
  2. Unable to unify the customer data and therefore unable to gain insights from this data and effectively use it for marketing campaigns.
  3. Long delays in getting insights from their customer data, and even then, these insights have very little information about marketing channels


Data Stats

Marketers have a data problem


Any AI technology has to solve these three overarching problems that revolve around organizing, accessing and acting on your customer data in real-time.


Identity, Insights & Intelligent Orchestration

Any AI technology for marketing can only be effective if it solves the three problems around identity, insights, and intelligent orchestration.


In order to effectively use customer data to inform our marketing campaigns, we have to first gather, organize and tie this data to the identity of specific individuals. An AI system must have the ability to track a customer or prospect’s interactions across multiple channels and stitch together behavioral data gathered from browser cookies, device identifiers, email addresses and customer IDs to develop a comprehensive view of the activities of a specific individual.

This is analogous to a person knowing things about another individual and committing that to memory for use in future interactions.

360-degree user profiles

360-degree Customer View


Once individuals are identified and their behavioral data organized, the system should also be able to take this behavioral information and combine it with information about product attributes, historical data and other external data such as location and time to develop insights that can drive marketing activities to that individual. These insights could take the form of predictive scores that determine the likelihood of a person purchasing a particular product or a churn score that indicates the likelihood of a customer ceasing to use your product or service. These scores are developed using a variety of signals such as decrease in product use, frequency of interaction, type of interaction, changes in patterns of interaction, etc.

This is analogous to the cognitive powers of the human brain to combine the information it has about an individual in its memory with external data to develop insights and opinions about that individual.

Churn Score img

Churn Score for a customer

Intelligent Orchestration

After developing insights into each customer based on all the factors discussed above, the AI system should also enable the marketer to develop segments of customers and execute multi-channel campaigns that act upon these insights to segments of customers. In our example above, customers with a high churn score could be invited to speak with a customer service representative to try to better understand their change in behavior.

A typical multi-channel campaign should execute across email, web, social media (such as Facebook) mobile messaging and notifications, and direct mail.

And in today’s world, all these three activities need to happen continuously and in real-time.

Multi-channel customer journey orchestration with Blueshift's AI-powered marketing platform

Multi-channel journey orchestration



The three blog posts in this series (Part 1 here and Part 2 here) give you a more complete view of the challenges facing business-to-consumer marketers today and the potential for AI to help solve some of these issues, a blueprint for developing a strategy to incorporate AI in your marketing, and key attributes to look for when evaluating AI technologies for marketing.

For More Information

Read more about AI powered marketing in our resources section.

Watch this rare webinar with Analysts from Forrester Research and VentureBeat hosted by Blueshift about getting the most out of your customer data with AI


AI-Powered Marketing – It’s About the Data “Cupid!”

Marketing is a fertile petri dish for the use of artificial intelligence (AI).  In fact, a recent study by Forrester Research found that 46% of respondents said that marketing and sales were the leading groups inside their companies evaluating the investment and adoption in AI.  In a recent webinar with analyst Rusty Warner from Forrester Research, Stewart Rogers of VentureBeat and Vijay Chittoor of Blueshift, we discussed these and other developments in the world of AI as they relate to marketing.  This is part 1 of a 3-part series about the topics discussed.

Customer data is exploding and marketers need help

Anyone who’s been in the trenches of online marketing in the last few years has seen the explosion of data first-hand. We’ve gone from having to work with just demographic data to the plethora of online data including online browsing activity tracked through cookies, opens and clicks of emails, mobile app activities, social media engagement, intent to purchase and purchase data and so much more. There is a treasure trove of information in all this data, but marketers are overwhelmed. 85% of them are unable to extract value from their data according to a study by Econsultancy and 59% of those attending our webinar told us that they are using less than 25% of their marketing data.

Not surprising to most marketers, there is a growing rift between the amount of customer data being generated and the capacity for traditional marketing techniques –largely powered by human analysis– to process this data. Rusty Warner refers to this phenomenon as “exceeding the human cognitive capacity” because of the complexity, volume and velocity of information.

Effective use of AI can narrow this human cognitive challenge by processing these large volumes of data quickly and use machine learning to recognize patterns in the data and predict what to do next based on the past behavior of similar audiences. While this sounds pretty logical and straightforward, it’s not as easy as buying a black-box AI system and dropping it in.  

The decision to deploy an AI system to improve marketing performance is a big one that should to be given the same level of planning and preparation that was given to deploying a CRM system (your system of record) or your marketing automation system (your system of engagement). The AI powered system will become your system of intelligence that must work closely with these other systems to improve results.

In the next part of this series, we will outline Forrester’s recommendations for planning, organizing, and deploying AI for your marketing.

webinar ai powered marketing and put your customer data to work with blueshift featuring forrester and venturebeat