Image: Alexandra Koch from Pixabay
Microsoft has detailed a number of practical tips to advertisers on the use of AI inclusive prompt engineering for ad copywriting.
Microsoft research shows that inclusive advertising has a significant impact on purchase intent and brand perception among Gen Z consumers, with 70% more trusting of brands that represent diversity in ads.
The research shows that inclusivity in the ads boosts ad performance. Findings include:
- 76% are more likely to support brands they consider authentic in their advertising
- 49% stopped purchasing from a brand that didn’t represent their values
and strikingly for the bottom line,
- A 23-point lift in purchase intent for the most inclusive ad shown
Generative AI tools such as Microsoft's Copilot can help advertisers in drafting inclusive ad copy to create high quality ads that perform well.
AI is potentially a huge time and labour saver for advertisers, enabling the generation of creative assets at scale for a range of scenarios and audiences. Effective prompt engineering is a skill in itself, and Microsoft’s blog post aims to help advertisers in developing these skills.
MIcrosoft’s blueprint for a great prompt for ad copy looks like this:
- Goal - what you want to achieve in the task
- What role the AI will take on in the task, eg the ad copywriter
- The context for the product or service - benefits, industry insights etc
- Who the target audience is for the ad
- Relevant source information such as brand guidelines or market research
- The tone of the ad copy
Taking this blueprint as a base, you can now consider the insights from Microsoft’s research on inclusive advertising and add inclusive modifiers, turning inclusive advertising insights into prompt engineering modifiers.
Microsoft outlines 5 types of modifiers to build inclusive prompts:
1. Inclusive metaphors emphasising connection, openness and balance (meaning everyone is represented equally)
2. Inclusive words - 50 terms that signal inclusion according to the three concepts above
3. Inclusive feelings, characterized by the two main categories of joy and trust, which break down further into subcategories.
4. Dimensions of diversity, including generation, gender, ability and so on.
5. Inclusive product features such as accessibility and skin customisation.
Here’s the earlier prompt, updated with inclusive modifiers to show dimensions of diversity and inclusive product features.
And here’s an inclusive prompt considering some of the inclusive words in the table above:
Microsoft says you should consider tones and feelings in your ad copy. Tone is the voice of your brand, and feeling is the result in your target audience.
Brands with a high degree of emotional intelligence towards their customers will do better at getting this right and producing effective ads which connect with their audience and perform well.
Different customer groups will need a different approach, for example Gen X might respond better to a nostalgic, value-oriented, or relatable tone. You can use Copilot to help determine the correct tone for a particular audience.
The blog post ends with a reminder that responsible use of generative AI means you’re the pilot in this venture. Far from handing everything over to the AI, you remain accountable for the final version of your creative assets.
