The role of AI prompts in personalised marketing

Prompt engineer Indhran Indhraseghar discusses the semantics of this ever-advancing technology and how it ensures publicity campaigns succeed

Ongoing developments in artificial intelligence continue to open possibilities in various fields, despite some saying the technology poses challenges such as lack of authenticity and originality. But speak with prompt engineer Indhran Indhraseghar and he will tell you that AI presents more opportunities than threats – for example in the world of marketing.

For the uninitiated, prompt engineering involves the process of structuring an instruction that can be interpreted and understood by a generative AI model. A prompt is natural language text describing the task that an AI should perform.

Here, Indhran answers questions you might have about AI prompts and the overall potential of this ever-evolving technology.

How does prompt engineering play a role in developing AI models for personalised marketing?

According to Stanford University’s Human-Centred Artificial Intelligence Institute, one leading streaming platform leveraged prompt engineering to create an AI model that analyses users’ viewing histories, preferences, and real-time interactions to generate tailored content suggestions.

This resulted in a 25% increase in user engagement and a 15% reduction in churn rate.

What are some specific challenges prompt engineers face with AI models for marketing campaigns or content generation?

One of the primary challenges is striking the right balance between specificity and flexibility. Prompts must be precise enough to guide the AI towards generating relevant and on-brand content, while also being adaptable to handle diverse user inputs, changing trends, and unexpected scenarios.

At one leading advertising agency, prompt engineers addressed this by developing an AI system that automatically adjusts prompts based on real-time user data, campaign performance metrics, and insights from creative teams.

This approach, as detailed in the Harvard Business Review, enabled the agency to deliver highly targeted ad campaigns that resonated with specific audience segments, leading to a 30% increase in click-through rates and a 20% boost in conversions.

Can you provide examples where prompt engineering improved the targeting or personalisation of marketing campaigns?

A notable case study involves the implementation of AI-driven email campaigns for a leading luxury fashion brand. By leveraging user purchase history, browsing behaviour, and demographic data, prompt engineers created dynamic prompts that generated highly targeted product recommendations and personalised email content.

This initiative resulted in a 45% increase in email open rates, a 30% boost in conversion rates, and a 20% uplift in average order value compared with generic email campaigns.

I also recently collaborated with a major media company to develop an AI system that generates personalised news digests using multimodal prompts that combine text, images, and video snippets, resulting in a 25% increase in user engagement and content consumption.

In the context of marketing and media, how important is human input and feedback in shaping the prompts?

Human input and feedback are crucial. At a leading advertising agency, we established a cross-functional team comprising prompt engineers, creative directors, data analysts, and client representatives.

This collaborative approach allowed us to incorporate domain expertise and user insights into prompt design, ensuring AI-generated content was not only engaging but also aligned with brand guidelines and messaging strategies.

During a recent influencer marketing campaign for a fashion brand, we used sentiment analysis and social-media engagement data to iteratively optimise prompts, resulting in a 35% increase in campaign reach and a 20% improvement in audience sentiment.

What measures do you take to ensure AI prompts mitigate biases and promote ethical considerations such as diversity and inclusivity?

First, we work closely with legal and compliance teams to ensure prompts do not perpetuate harmful biases or promote discriminatory practices. This involves carefully auditing training data for potential biases, implementing fairness metrics in model evaluation, and incorporating diversity and inclusivity guidelines into prompt creation.

Second, we have established an AI Ethics Advisory Board, made up of experts from various fields, including technology, law, social sciences, and civil-rights organisations. This board provides guidance and oversight on ethical AI development, ensuring our practices align with industry standards and best practices for responsible AI.

Finally, we prioritise transparency and explainability in our AI systems. We have developed techniques to provide clear explanations for how prompts influence content-generation and -recommendation decisions, empowering clients and users with greater understanding and control.

How do you envision prompt engineering evolving?

Key trends I foresee are:

• increased adoption of multimodal prompts that combine text, images, video, and other modalities to create more immersive and engaging user experiences;
• integration of real-time data streams from Internet of Things (IoT) sensors, wearable devices, and smart environments to enable even more granular and context-aware personalisation;
• emphasis on explainable AI techniques that provide transparency into how prompts influence content generation and recommendation decisions; and
• collaboration between prompt engineers and creative professionals to develop new forms of AI-assisted content creation and storytelling.

This story was first published on FMT