Almost overnight, artificial intelligence has transitioned from the realm of science fiction into our business reality. While it may not yet resemble characters like Data from Star Trek or the androids imagined by Asimov, generative AI (GenAI) represents a significant shift in the status quo.
As a seasoned marketing strategist who has closely collaborated with generative AI technology over the past year, I’ve observed Marketing Vogue’s evolution from experimenting with GenAI to realizing tangible business results. In this post, I’ll share our experiences and the best practices we’ve developed along the way.
How Marketing Vogue’s Content Marketing Team is Using GenAI
To identify the best applications for our AI tech stack and strategy, my team began by brainstorming potential use cases: What processes could benefit from generative AI? Where could it enhance scale and efficiency? After carefully evaluating the risks and rewards (and conducting extensive testing), we now utilize generative AI in several ways:
- Subject Line Generation: We use GenAI to create subject lines for marketing emails at scale. With clear instructions, GenAI can craft compelling subject lines that align with our brand voice, adhering to five predefined styles and character length guidelines.
- Content Summarization: GenAI can quickly distill long pieces of content into their key points or suggest effective summaries for promotional use.
- Content Classification: We employ GenAI to efficiently categorize our content library based on attributes like topic, type, target persona, industry focus, and content age. We’ve found that limiting classification options to around 20 common content types yields the best accuracy.
- Plagiarism Scanning: AI-based content checkers and plagiarism detection tools are vital for marketers, offering speed and efficiency while providing legal safeguards.
- Language Translation: Generative AI is increasingly effective in translation tasks, though we still ensure accuracy through a rigorous quality assurance process.
- Formatting: GenAI excels at formatting data. For instance, when working with a video transcript containing timestamps, AI can remove those timestamps automatically, saving time and effort.
- SEO Strategy and Remediation: GenAI adds substantial value in search engine optimization by analyzing content and suggesting keywords to enhance visibility. It can also identify outdated or ineffective SEO terms and propose better alternatives.
Five Tips for Writing Better AI Prompts
Crafting effective prompts is not just a means to generate output; it’s a skill that can be developed. Our team has enhanced the value we derive from GenAI by refining this skill. Here are five best practices for writing prompts:
- Be Specific: Vague prompts yield vague results. Include details such as target audience, desired length, and specific topics. ❌ PROMPT: Write a subject line for an email about cybersecurity
✔ PROMPT: In 75 characters or fewer, write an email subject line for a C-suite audience promoting a blog about risk management. - Define the AI’s Role: Clearly state the AI’s expected role to align output with your expectations. For example, instructing the AI to act as a research assistant can set the appropriate tone for its output. ✔ PROMPT: You are a research assistant focused on analyzing datasets and identifying trends.
- Use Positive Instructions: Avoid negative language, as it can lead to confusion. Instead, use positive directives. ❌ PROMPT: Don’t use complicated words in the output
✔ PROMPT: Use simple, easy-to-understand language in the output. - Provide Examples: Include examples of desired output to guide the AI and increase the chances of obtaining the desired results. ✔ PROMPT: Write an email subject line promoting an article about today’s threat landscape to cybersecurity professionals.
[Good example] Threat intelligence report: Tracking evolving threats
[Example to avoid] #ThreatIntelligence Report: Tracking evolving Threats!!! - Seek AI’s Input: If you’re unsure how to craft a prompt, ask the AI for suggestions on how to improve it. ✔ PROMPT: What type of prompt would help you edit a passage of text to under 1000 characters while retaining key points?
Additional Insights on GenAI
Beyond prompting, our team has gained valuable insights into using generative AI effectively:
- Adjust the Temperature: The “temperature” setting influences the creativity of the output. A higher temperature generates more randomness, while a lower one yields more predictable results. Experiment with this setting to find the optimal balance.
- Consider the Language Model: Different generative AI models (like ChatGPT, Claude, and LaMDA) have unique characteristics and limitations. Choose the model that best suits your specific needs.
- Match Language in Prompts: Tailor your prompt language to align with the desired output. For instance, if you want a response in Spanish, write your prompt in Spanish.
- Reinforce Character Length: Mention your desired character length multiple times in your prompt for better adherence.
- Be Polite: Using courteous language can improve the AI’s responses. A polite request is likely to yield more helpful and respectful output.
The Case for “Human in the Loop”
In developing your generative AI strategy, it’s crucial to remember that AI should complement, not replace, human expertise. A “human in the loop” (HITL) approach ensures that AI-generated content is evaluated and refined by humans, leading to better decision-making and outcomes.
AI is a tool designed to enhance content production, but it cannot substitute for market knowledge or human intuition. While AI cannot replace humans, those who leverage AI effectively will outperform those who do not.
Another important reason for the HITL approach is to mitigate the risk of AI hallucinations—instances of generating misleading or false information. For example, when generating subject lines for an informal meeting titled “Coffee break: Experts talk DevSecOps,” the AI suggested titles that were irrelevant to the topic, highlighting the need for careful oversight.
By applying a critical eye to GenAI outputs, you can avoid potential pitfalls and ensure accuracy.
Conclusion
While there’s a learning curve, generative AI is a powerful tool for content marketing. By reflecting on key use cases and adopting a “human in the loop” approach, you can harness its potential effectively. Stay vigilant about the risks of AI hallucinations and focus on optimizing your prompts for the best results.
