Google Heart Framework Example
The Google HEART framework example below demonstrates the evaluation of two AI tools and their success at producing holiday-themed recipes. To illustrate this, 100 Home cooks in the US reacted to each AI experience and provided their feedback.
The provided template showcases a comparative study of two AI tools, Variation A: Chat GPT and Variation B: Google Bard, using the HEART framework.
A Summary of the Google Heart Framework Example
To begin with, here’s a summary of the Google HEART framework example:
- Goal: A helpful AI tool for home cooks without providing excessive information.
- Signal: Bard AI’s more concise recipe response led to less negative impressions and allowed users to enjoy the fun and innovation of the recipe. For instance, Chat GPT’s NPS is 21.57, with 142% alignment, while Google Bard’s NPS is 36.63, with 189% alignment.
- Metrics: Emotional reaction through Net Positive Alignment, and Net Promoter Score (NPS) to measure user satisfaction.
- Goal: Encourage users to use the command bar and its suggestions.
- Signal: Bard AI offers a more extensive category suggestion list, driving over 3x more users (39%) towards command suggestions than ChatGPT (12%)
- Metrics: The first-click interaction with the tool.
- Goal: New users should ideally use the provided recipe or tool more often.
- Signal: Chat GPT produces twice as many participants who want to type in a new command, indicating dissatisfaction with the current results.
- Metrics: Users’ actions and decisions after using the tool.
- Goal: Move users from neutrality to a positive sentiment about using the tool repeatedly.
- Signal: Over 10% of users feel they won’t use Chat GPT for recipes again, compared to only 3% for Bard.
- Metrics: Likelihood of users re-engaging with the tool post-first use.
- Task Success
- Goal: Users should successfully utilize the command bar and its suggestions.
- Signal: Bard AI has more spots for successful interaction via its suggested commands. For instance, Chat GPT has a 65% success rate for a specific command, while Google Bard’s rate is 81%.
- Metrics: Evaluate the number of users who successfully follow given goals/directives.
Additionally, the framework provides URLs at the bottom, linking to the raw data in Helio.
Heart Framework Example Takeaways
The Google HEART framework example depicted here offers a structured way to compare the user experience of two AI tools. In this example, Google Bard outperforms Chat GPT in several aspects based on the metrics provided.