Gather your consumer data

Before creating the GPT, you need some consumer data you can feed into it. You can gather the data from various sources: surveys, social media analytics, website metrics, customer reviews, and purchase history, to name a few.

Comprehensive and relevant data like this will provides insight into your target audience's demographics, preferences, behaviours, and interactions with your brand - this builds a strong and effective marketing analyst GPT.

Pull together any of the below types of data:

Surveys

• This includes info about your customer demographics (age, gender, location), product preferences, satisfaction levels, and any open-ended feedback.
• How to find it? Use survey tools (Google Forms, SurveyMonkey, Typeform etc) to distribute surveys to your customer base via email, social media, or your website, then collate the results.

Social media stats

• Here we’re talking audience demographics, engagement rates (likes, comments, shares), top-performing posts, and sentiment analysis.
• Where to find it? Use built-in analytics tools on social media platforms (e.g., Facebook Insights, Twitter Analytics, Instagram Insights) or third-party tools like Hootsuite or Sprout Social.

Website metrics

• Gather your website traffic, page views, bounce rates, conversion rates, and user behaviour flow.
• Where to find it? Web analytics tools like Google Analytics or Plausible.

Customer reviews

• Product or service feedback, sentiment analysis, and common themes or issues mentioned by customers.
• How to find it? Collect reviews and testimonials from e-commerce platforms (e.g., Amazon, Etsy), third-party review sites (e.g., Yelp, TripAdvisor), customer emails, and feedback collected by your customer service team.

Purchase history

• Products purchased, frequency of purchases, average order value, and customer lifetime value.
• Where to find it? Export sales data from your e-commerce platform (e.g., Shopify, WooCommerce) or customer relationship management (CRM) system (e.g., Salesforce, HubSpot).

Prepare the data

To prepare this data for use in your marketing analyst GPT, ensure that it is:

• Consolidated into a single, structured format (e.g., CSV or Excel file)
• Cleaned and preprocessed to remove any duplicates, irrelevant information, or sensitive personal data
• Labeled and categorised appropriately for easy analysis and interpretation by the GPT

‍

Create your GPT

Log in to ChatGPT and click ‘Explore GPTs’ on the left-hand side of the screen. Click +Create in the top-right corner and you’ll be taken to the build screen.

On the left, there are two options: ‘Create’ and ‘Configure.’ Make sure ‘Create’ is selected.

In the text box below, type a brief description of your marketing analyst GPT, e.g. "A custom GPT that analyses consumer data, helps design data-driven marketing campaigns, and generates engaging social media content." Press "Enter" to submit it.

You can customise the name ChatGPT offers or keep it, and keep or change the profile picture it generates.

ChatGPT will then work with you to refine the GPT according to its purpose. In our case, we want the GPT to offer insights from data and suggestions actions and strategies. So let’s tell it that.

Prompt:

The marketing analyst GPT should have a dual focus:
1. Analysing consumer data to identify trends, patterns, and key insights that can inform marketing decisions.
2. Leveraging those insights to suggest actionable strategies and ideas for marketing campaigns and social media content.By offering both data-driven insights and practical recommendations, the GPT will be a valuable tool for marketers looking to optimise their efforts and drive better results.

This is a good question. Important we answer it fully too, so that when we interact with the GPT, it gives us quality responses. We want our GPT to focus on providing valuable, actionable insights while maintaining a professional and objective tone. It should emphasise the use of relevant consumer data and only make recommendations that are supported by the available data.

Prompt:

It should:
[List things to adhere to]

It should avoid:
[List things to avoid]

Next we’ll customise what the GPT should do when a request or prompt is unclear.

Prompt:

When faced with incomplete or ambiguous information in a request, the marketing analyst GPT should prioritise accuracy and clarity by asking for more details or clarification from the user. If the GPT makes educated guesses or assumptions based on limited information, it should clearly communicate these assumptions to the user and note any potential limitations or uncertainties in its output. However, the GPT should default to requesting additional information or clarification whenever possible to maintain the highest standards of data-driven decision-making. By taking this approach, the GPT will build trust with users by demonstrating a commitment to accuracy, transparency, and reliability in its insights and recommendations.

Now it’s asking what kind of personality the GPT should have. Let’s aim to strike a balance between professionalism and engagement, so it’s a nice experience for people when they’re using the GPT.

Prompt:

The GPT should maintain a clear and direct approach, but it should also incorporate elements of motivation and encouragement to help users feel supported and empowered in their data-driven marketing efforts.

ChatGPT thinks we’re done here, but the last step is upload all our data so that GPT can respond with context and insight.

Prompt:

[upload your data as attachments]
‍
I am uploading all of our current data. You'll need this to analyse our customers' purchasing patterns, preferences, and behaviours comprehensively. This data includes:

- Purchase history data: This will help you identify common product combinations, order frequencies, and customer buying habits.
- Customer survey data: Surveys provide insights into customer preferences, satisfaction levels, and areas for improvement.
Website metrics: This data will reveal user behaviour on our website, including popular pages, bounce rates, and conversion paths.
- Social media analytics: Insights from our social media platforms will help you understand our audience demographics, engagement patterns, and content performance.
- Customer reviews: Reviews from our website and third-party platforms offer valuable feedback on our products and services.

By analysing this comprehensive dataset, you'll be able to:
- Identify opportunities for cross-selling and bundling products based on common purchasing patterns.
- Develop targeted marketing strategies that align with our customers' preferences and behaviours.
- Optimise our website and social media content to better engage and convert our audience.Address areas for improvement in our products and services based on customer feedback.Provide data-driven recommendations to enhance our overall marketing efforts and drive business growth.

Please use this data to generate actionable insights and strategies that will help us better serve our customers and achieve our marketing goals.

Great! We’re done customising and now we can start testing our GPT to check it works.

Test-drive your GPT

Run some test prompts in the panel on the right side of the screen. Here are some examples to inspire you.

To identify strengths and weaknesses in your offerings

Based on your knowledge of our customer review data, what are the top two areas where our products or services excel, and what are the top two areas that need improvement?

To get insights into your most engaged audience segments and their content preferences

Based on your knowledge of our social media analytics data, which demographic groups engage most with our content, and what types of content resonate best with each group?

To uncover shifts in user behaviour over time

Based on your knowledge of our website metrics data, comparing the past three months to the previous quarter - what trends or changes in user behaviour can you identify, and what recommendations do you have for optimising our website?

To identify cross-promotion opportunities

Based on your knowledge of our purchase history data, what are the most common product combinations purchased by our customers, and how can we use this information to create effective cross-selling or bundling strategies?

To uncover the key drivers behind customer purchasing behaviour

Analysing our survey data, what are the top three factors that influence our customers' purchasing decisions, and how can we align our marketing messages and content to address these factors effectively?

Use these prompts or your own to test your marketing analyst GPT's ability to generate valuable insights and recommendations across various aspects of your marketing efforts.

Here’s an example response:

Success! The GPT is using the knowledge and data it’s been given to make data-driven decisions and recommendations.

Publish and share your GPT

When you’ve finished customising your GPT, click ‘Create’ in the top-right corner and you’ll be presented with sharing options. Select your preference and click ‘Share’.

After the sharing link generates, click ‘View GPT’ to see your customised GPT in action. You can share this link with your team or use it yourself to interact with the GPT anytime.

Optimise and refine your GPT

Some tips to ensure your GPT stays up-to-date and effective at its job:

• Regularly update your GPT with new consumer data to keep its insights and recommendations relevant
• Provide feedback on the GPT's outputs to help it learn and improve over time
• Experiment with different prompts and parameters to fine-tune your GPT's performance