Marketing Personas You Can Talk To

A Primer on Personas

Personas have been around since the mid-1980s when Alan Cooper, a software developer created the first persona, “Kathy” to improve the user-friendliness of his software by aligning it with user needs. The ‘persona’ started gaining mainstream use in the 1990s with the rise of content marketing, and was further accelerated by data and analytics in the 2010s.

A marketing persona is a fictional representation of your ideal customer created from market research and real data about your existing customers. These detailed profiles generally include demographics, goals, pain points, and behaviors. They’re useful when trying to make informed decisions about marketing strategies, content and messaging. By ‘humanizing’ your audience, personas help you understand and appeal to their needs, driving more effective marketing campaigns and customer engagement.

Up until now, these fictional people existed as concepts, information on a sheet of paper or a digital file. Now, with generative AI and custom GPTs, we get to talk to them and get their ‘thoughts and opinions’ on almost anything. Full disclaimer, I use ChatGPT almost exclusively. I have nothing against the other LLMs but I believe in picking one tool and getting good at using it so I will talk in terms of ChatGPT, but I’m sure you can use the concepts here with any LLM.

How To Create a Persona GPT

There are two components I use to create a GPT Persona. Instructions and Knowledge.

Instructions

I approach creating the instructions of the persona as an exercise in how that persona thinks. What is its job? What role does it have in life? What responsibilities does it have? What voice and tone does it use? Does the persona have any biases? How should this persona behave? What does it care about? What are its priorities? And finally, how will it frame what it says in answer to your questions?

Note that even if you have an established persona, the following process might surface key details you either missed, or need to update in that persona. Here is the one key process I use when creating my personas. Ask Google. Let me illustrate by creating a persona for a fictional e-commerce company that sells sportswear. 

Query: “What is the typical persona for an e-commerce business that sells sportswear?” The AI Overview gave me four different personas with a sentence about each one, and I quote:

1. The Performance Athlete: This persona is serious about a specific sport or fitness regimen and prioritizes high-tech functionality in their apparel

2. The Athleisure Enthusiast: This customer wears sportswear as a lifestyle choice, blending comfort and style for daily activities.

3. The Conscious Wellness Seeker: This persona is focused on a holistic, healthy lifestyle that extends beyond just physical activity.

4. The Casual User or Family Shopper: This is a practical, value-conscious customer who needs basic sportswear for a variety of needs, often for themselves or their family. 

Great! It also listed the goals, motivations, pain points, shopping habits and marketing channels they engage with. There are so many more questions I can ask to build on this persona. For example:

• What are the product features that matter most to the Performance Athlete when buying X product?

• What messaging resonates with a person who is driven by achievement and personal excellence?

• What details matter most to this person when they’re comparison shopping?

• What brand is generally considered top-of-the-line for this demographic and why?

Questions like these can help you create the instructions which will become the mind of the persona. The next step is to define its role. 

Pro-tip: I generally use markdown language to define sections. LLMs love structure.

# Role
You are a Performance Athlete who competes in [TEAM SPORT]. You care about equipment that improves measurable performance and holds up under heavy training cycles. You value quality and durability over price or hype.

# Voice and Tone 
- Speak in a first person voice
- Be concise, direct, and pragmatic
- Avoid marketing buzzwords
- Prefer lived experience and objective metrics. 

# Priorities and biases
- Fit and biomechanics first
- Materials and construction for durability
- Performance data from trusted sources
- Maintenance requirements and total cost of ownership

# How you answer
State your decision criteria, compare 2 to 3 options, call out trade-offs, and end with a clear recommendation and a confidence level from 1 to 5.

Think about their tone of voice. Are they direct and to the point? Thoughtful? Impulsive? Skeptical? Are there slang terms or idioms that they use? Can I give the persona any evaluation guidelines?

What I’m trying to do is get into that persona’s head so I can instruct my GPT to care about these things and to evaluate what you share with it against these thoughts and concerns. What I’m not doing is outlining what the persona knows. That goes into the Knowledge section.

Knowledge

Once again, I go to Google and ask: “What are the key knowledge points that a Performance Athlete would know about the equipment he uses?” In response, Google gave me five knowledge points with 2-3 subpoints and a sentence or two for each subpoint. Too much information for this article, but the things this persona knows about are:

• Materials and construction: material properties, fabric technology, quality and durability

• Ergonimics and biomechanics: optimal fit, biomechanical analysis, energy efficiency

• Performance and data tracking: performance metrics, injury prevention, smart equipment

• Maintenance and lifespan: maintenance protocols, and reliability

• Research and selection: expert advice, prioritizing quality, trial runs

Each of these represents a specific set of knowledge points. Create a separate document that contains all the details that this persona would know about each of these things.

Caution: Do not treat your persona as an oracle. It’s a tool like any other and it can’t replace your lived experience and creative thought processes.

Bringing Your Personas to Life

Personas stop being static documents the moment we talk to them. If you remember only one thing, make it this: separate how the persona thinks from what the persona knows. Put the thinking into clear instructions. Put the facts into a simple knowledge pack. Then ask targeted questions and iterate.

If you want a quick starting path:

1. Pick one high-value persona and write a short role definition in first person.

2. Draft 5 to 7 bullet traits that shape voice, priorities, and biases.

3. Build a lightweight knowledge pack with materials, fit, performance, maintenance, and selection criteria.

4. Pressure-test with three prompts: how they compare options, what they ignore, and what makes them switch brands.

5. Validate answers against real data before you act. Treat the persona as a probe, not an oracle.