30/04/2024

How to accelerate hyper-personalisation in marketing with generative AI

Authors: Darije Ramljak & Srini Rudrabhatla

In the dynamic marketing landscape, the rise of generative AI has opened up a multitude of possibilities. At the forefront of this technological advancement is hyper-personalisation, which allows brands to tailor their marketing efforts to the individual customer. By using generative AI to create hyper-personalised content at scale, companies can not only optimise their internal processes but also transform the customer experience. In this article, we provide insights and guidance on the use of generative AI in the field of hyper-personalisation in marketing and showcase the ground-breaking work IBM iX has done with our client MOL.

In today’s fast-paced business landscape, companies are under pressure to stay ahead of the curve, and the strategic adoption of innovative technologies has become a key differentiator between leaders and followers.

Among the array of transformative technologies, generative AI has emerged as a true game-changer, thanks to its remarkable capabilities and human-like behaviour. This technology holds immense promise, enabling automation in various business functions that were previously thought to be unattainable. The widespread availability of these powerful tools is a significant milestone for early-stage transformative technology.

What sets generative AI apart is its ability to generate original, conversational content using large language models (LLMs). These models can seamlessly integrate corporate guidelines and inputs, producing compliant content that aligns with brand requirements, tone, and voice. While generative AI offers substantial benefits on its own, its value is amplified when combined with traditional AI, unlocking even greater potential for businesses.

When to use traditional AI capabilities:

Predictive/ Prescriptive
Structured data analysis, predictions, forecasting etc.

Directed Conversational AI
Deterministic dialogue flows for structured conversational AI

Computer Vision AI
Machine vision for object and anomaly detection

Process Automation
Robotic process automation, process reengineering and optimization

When to use generative AI capabilities:

Summarization
of documents such as user manuals, asset notes, financial reports, etc.

Conversational Search
supporting SOP, troubleshooting instructions, etc.

Content creation
including personas, user stories, images, personalized UI, marketing copy, email/social responses etc.

Code creation
eg. Code co-pilot, code conversion, create technical documentation, test cases etc.

Generative AI in marketing

Generative AI has the potential to revolutionise marketing by addressing its inherent challenges, making it one of the most dynamic fields for this transformative technology. By combining traditional and generative AI capabilities, businesses can deliver personalised, real-time experiences that customers crave. This technology empowers Chief Marketing Officers (CMOs) to gather insights and act on them swiftly, enabling the production and deployment of tailored content necessary for hyper-personalisation at scale. Innovative tools are emerging in the marketing landscape that can enhance the process of creating captivating content and campaigns.

These applications of generative AI are wide-ranging, including the creation and optimisation of creative assets, enhancing analytics, and improving targeting accuracy. By generating personalised content at scale, generative AI not only helps businesses streamline their internal processes but also elevates the customer experience to unprecedented levels. As a result, the integration of generative AI in marketing strategies can support businesses in refining their approach and delivering more impactful customer interactions.

Some key use cases for implementing generative AI in marketing include:

Enhanced Customer Segmentation & Targeting

Analysing customer data to refine audience segmentation and targeting, leveraging behavioural and contextual data for more effective marketing campaigns.
Customer Lifetime Value Prediction

Assessing customer data and behaviour patterns to predict the customer's lifetime value (CLV), enabling strategic decision-making and resource allocation.
Data-driven Content Planning

Examining (social) media data, trends, and audience behaviour to develop informed social media content strategies, tailoring formats and posting frequency accordingly.
Automated Content Generation

Creating content such as digital ads, blog articles, social media posts, and email campaigns with minimal human intervention, maximising efficiency and consistency.
Personalised Content Curation

Recommending relevant content and offers to customers by considering their individual preferences and interests, enhancing engagement and satisfaction.
A/B Testing Optimisation

Streamlining A/B testing by auto-generating content variations, including ad copy, landing pages, and calls to action, to identify the most effective options.
Social Media Listening & Insights

Analysing social media conversations to gain valuable insights into brand mentions, customer sentiments, and emerging trends, informing marketing strategies and decisions.
SEO Optimisation

Fine-tuning search engine keyword strategies based on performance data and competitor analysis, maximising visibility and driving organic traffic.

How to accelerate the journey to marketing hyper-personalisation

It’s crucial to leverage generative AI effectively. This can be achieved by focusing on three essential building blocks that work together to create a seamless and efficient process, tailored to meet the unique needs of each marketer.

These vital building blocks for achieving hyper-personalized marketing success encompass a Branded Foundation Model, a Marketer’s Workbench, and Integrations with Campaign Management.

IBM Consulting Advantage for Marketing – AI Academy

Branded Foundation Model

The first building block, the Branded Foundation Model, involves preparing your data to customise your model, ensuring on-brand and accurate content generation and insights. This process involves several key steps and required capabilities:

Source of truth: Identifying the necessary brand, product, customer, and campaign data sources needed to create a reliable foundation for your model.
Asset tagging: Generating metadata for photos, videos, and other assets, making content indexable, searchable, and accessible to the models in downstream marketing processes.
Consistent taxonomy and metadata: Utilising a controlled vocabulary to tag and structure content, ensuring that the models deliver results based on the most current and accurate data.

Once the data is prepared, you can select, train, tune, and maintain your customized model through prompt engineering, fine-tuning, and data embedding preparation. This process involves optimising natural language instructions, adapting the model to specific use cases, and providing context and memory to enhance performance. Finally, deploying, optimising, and monitoring generative AI applications ensure the smooth operation of your marketing strategies.

Marketer’s Workbench

The second building block, the Marketer’s Workbench, provides a customised interface with a pre-configured family of conversational AI assistants designed to support marketers across various roles and tasks. These AI assistants can be divided into two main categories:

Creator Assistants: These assistants generate content and automate actions across the marketing content supply chain, streamlining the process and reducing manual efforts.
Insight Assistants: These assistants support the most common marketing requests, freeing up marketers’ time to focus on higher-value tasks and strategic planning.

Generative AI Marketer’s workbench assistants

Integrations with Campaign Management

The third building block, Integrations with Campaign Management, ensures seamless connections between generative AI and work management or marketing platforms. By integrating these systems, marketers can harness the full potential of generative AI while maintaining a unified and efficient workflow. This seamless integration enables marketing teams to access AI-generated content, insights, and recommendations directly within their existing campaign management tools, streamlining communication and collaboration. As a result, this empowers marketers to create more impactful campaigns with the help of generative AI while still maintaining full control over their strategies and execution.

Customer use case: The MOL marketing personalisation generative AI pilot

MOL is an integrated oil and gas company with 25.000 employees and over 2400 service stations in 10 countries. A few years ago, MOL started with a transformation from a traditional fuel retailer to a digitally driven consumer goods retailer and integrated mobility service provider.

With the new loyalty programme, they have planned to offer customers a personalised and convenient experience, thereby encouraging them to choose MOL service stations over competitors more frequently. The new loyalty and customer engagement platform has been designed and built in partnership with IBM on a variety of Salesforce technologies like the Loyalty Cloud, CRM, Experience Cloud, Marketing Cloud, Data Cloud and MuleSoft for integration.

Now, with over 5 million loyalty shopping transactions every month, MOL has a lot of collected data about their customers and we continued with the MOL group to build sophisticated customer management and data capabilities over the years. Our next pivotal shift is to move MOL towards hyper-personalisation with generative AI to drive higher customer margins with cross-sell and up-sell their hero products.

In a nutshell, MOL wanted to move from a standard, rule-based marketing concept, where all criteria for segmentation are pre-defined and the content was written in advance and uploaded into Marketing Cloud, to a behaviour-based model, where customer signals would be automatically harmonized into a persona profile with relevant hyper-personalised offers generated and sent to customers.

To be able to connect and harmonize all the data from Marketing Cloud, Experience Cloud, Loyalty, CRM, mobile app as well as 3rd party data into the customer profiles that made sense, we started with the implementation of Data Cloud as a foundational capability for generative AI. We then analysed the behavioural signals from the customers with a hyper-personalisation engine.

In the the latest generative AI pilot, we tackled the needs of marketers on languages for prompts and new data sets to enhance the customer profiles and dynamically A/B test content, subject lines on auto-generated personas per country. This enabled MOL to safely utilize generative AI capabilities powered Salesforce Data Cloud & IBM watsonx.ai, for campaign managers to generate hyper-personalised emails, push notifications in the local language in 1 click.

Marketing hyper-personalisation with generative AI pilot use case overview

The pilot solution focused on turning Data Cloud and Databricks customer data into micro-segment persona with feature engineering, we built a new Salesforce Marketer Workbench to interact with creative capabilities of generative AI and used IBM watsonx.ai platform to manage large-language models supporting Eastern European languages.

The initial strategic segmentation that was employed in the early stages of Loyalty platform evolution added significant value, driving personalized campaigns with redemption rates and ROI which are multiple times higher than those of general campaigns, ultimately leading to increased customer satisfaction.

Predictive AI and generative AI are now taking this to the next level. Our pilot program is currently focusing on coffee buyers in Hungary, but our platform has the potential to expand into new countries seamlessly. The exciting part of the generative AI pilot is that it sets a foundation for future scalability across more markets.

+ 24 %

Voucher redemption rate and product sales

x 10

Marketeer efficiency (mid-term projection)

From 10s to 100s

Market segmentation increase

Business results from our pilot using e-mail and push notifications content created by generative AI

By 2025, MOL’s goal is to onboard 5 million customers onto this platform, with personalised messaging tailored not only by country but also by individual preferences, resulting in potentially hundreds or even thousands of unique segments.