Building Our AI Text Analysis App: Exploring JTBD with AI

?? Hello!

Welcome to another edition of our newsletter. Today, we’re diving into a fascinating topic: how AI can help us explore Jobs-to-be-Done (JTBD) and how this approach is shaping our AI Text Analysis app. Ready to see how we’re building this in public? Let’s jump in! ??

What Exactly Are Jobs To Be Done?

If you’re here, you probably already have a good grasp of JTBD. But just in case, here’s a quick refresher: JTBD is all about understanding the underlying needs and motivations driving your customers’ actions.

Think of it as figuring out what "jobs" your customers are trying to get done when they use your product or service. Pretty essential stuff, right?

So what?

Why does this matter? Understanding JTBD helps us design better products, craft more effective marketing messages, and ultimately meet our customers' needs more precisely.

It’s like having a roadmap to their pain points and desires, allowing us to offer solutions that genuinely resonate.

For our AI Text Analysis app, this means creating features that directly address the key tasks our users are trying to accomplish, making our tool more valuable and user-friendly.

Collecting JTBD

The Traditional Way vs. The AI Way

Traditionally, collecting JTBD involves lots of user interviews, surveys, and sifting through support tickets. It’s effective but can be quite the time sink ? and heavy on the budget ??.

Enter AI! Specifically, ChatGPT. This tool can help us generate JTBD insights quickly and cost-effectively. Imagine having a virtual assistant who can comb through data and pick out the key jobs your customers are trying to get done. Intrigued? Let’s see how it works.

Practical Application: Identifying JTBD for Our AI Text Analysis App

Using ChatGPT for JTBD

We’ve been using ChatGPT to identify JTBD for our AI Text Analysis app. Here’s a quick rundown of how we do it:

Create a Descriptive Prompt: Give ChatGPT as much context as possible about the product/service, users, goals, problems and motivations.

You can ask ChatGPT to ask you what questions it needs you to answer:

Supply the answers, as well as any other contextual information and run the Prompt to get the JTBD:

And you'll get the results like:

Let's get the minor jobs for the major job of 'identifying key areas for improvement':

How we use JTBD to build the UI

Understanding JTBD has been a game-changer for us in designing our reporting features. We can tie all functionality back to individual customer JTBD.

JBTD to UI

So we can take the JTBD and ask ChatGPT to create a UI outline

Returns a list of jobs like below

And we can ask ChatGPT to create a simple wireframe.

Challenge, Challenge, Challenge

This is a good start to what users may expect to see in order to achieve that minor job. We would use our existing knowledge of the industry to include these suggestions and combine them with our wider vision of the product and the other jobs required.

Here is a first draft of our sentiment dashboard for Hotel feedback:

Other Uses for JTBD Output

Versatility of Synthetic JTBD

JTBD insights aren’t just for UI. They’re incredibly versatile and can help with:

• Creating User Stories / Product Design
• Customer Understanding
• Product Development
• Customer Journey Mapping
• Customer Retention
• Marketing Strategy
• Service Design

Benefits and Limitations of AI in JTBD

Pros and Cons of AI-Driven JTBD

Let’s break down the pros and cons of using AI for JTBD:

Benefits:

• Speed and Accessibility: AI speeds up the process of analyzing and responding to user needs.
• Cost-Effectiveness: AI reduces costs and offers broader contextual insights.

Limitations:

• Data Dependency: AI’s effectiveness depends on the quality of data it’s fed.
• Lacks human interrogation: A good human will challenge and delve deep to understand the underlying motivations that AI may not provide
• Confirmation Bias and Accuracy: There’s a risk of AI generating incorrect information.

This final bullet point is likely the riskiest. ChatGPT is great at telling you want to hear, and extra validation is required to avoid this.

Always validate AI-generated insights with real people and apply common sense.

Are synthetic JTBDs as reliable as those collected by humans? Absolutely not!

Human data collection remains the gold standard and will continue to be so in the foreseeable future.

However, synthetic JTBDs are faster and more cost-effective. When paired with experience and knowledge, they can directly address customer pains.

They offer hypotheses that need validation with real users.

We always anchor our work to customer JTBDs to ensure we stay on the right track!

Conclusion and Feedback Request

Thanks for Reading! ??

Thanks for sticking with us through this exploration of AI and JTBD. We’re on an exciting journey, and we’re glad you're along for the ride.

In the future, we'll explore the possibility of using AI to extract JTBD from organization text data and classify survey responses back to JTBD.

We Want Your Feedback!

What do you think about using AI for JTBD? Have you tried something similar? We’d love to hear your thoughts and experiences. Comment below or reach out directly.

#TextAnalytics #CustomerFeedback #AIInsights #SurveyInsights #GenAI #JTBD