How do you use ChatGPT for a JTBD analysis?

Quick Answer: You can use ChatGPT to run a Jobs-to-Be-Done (JTBD) analysis by feeding it customer interview transcripts and using structured prompts to extract functional, emotional, and social jobs. This turns raw qualitative data into a clear picture of what your customers are actually hiring your product to do, without weeks of manual synthesis.

Running a JTBD analysis manually is slow. You gather 10-15 customer interviews, spend days reading transcripts, and still end up with a spreadsheet full of half-formed insights that nobody acts on.

ChatGPT changes that workflow. With the right prompts, you can synthesise interview transcripts into structured JTBD outputs in a fraction of the time. This guide walks you through exactly how to do it, step by step, with prompts you can copy and use today.

By the end, you will know how to extract functional, emotional, and social jobs from raw interview data, identify the forces driving switching behaviour, and turn those insights into messaging and product decisions.

What Is Jobs-to-Be-Done and Why Does It Matter for B2B SaaS?

Jobs-to-Be-Done is a framework for understanding customer motivation. The core idea is that customers do not buy products because of features. They "hire" products to make progress on a specific job they need to get done.

For B2B SaaS, this distinction matters enormously. Your buyers are not purchasing software. They are hiring it to solve a business problem, reduce a professional risk, or look competent in front of their leadership team. When you understand the actual job, your positioning, messaging, and product roadmap all get sharper.

A JTBD analysis typically identifies three dimensions of every job:

• Functional job: The practical task the customer needs to accomplish ("I need to report pipeline accuracy to my CFO every week")
• Emotional job: How the customer wants to feel (or avoid feeling) in the process ("I need to feel confident the numbers are right before I present them")
• Social job: How the customer wants to be perceived by others ("I want my team to see me as someone who has finance under control")

The challenge is that customers rarely state these jobs directly. They surface through interview language, and spotting them requires careful reading across multiple transcripts. That is exactly where ChatGPT earns its place in this workflow.

What You Need Before You Start

Before you open ChatGPT, get these inputs ready:

• At least 3-5 customer interview transcripts (more is better, but 3 is workable)
• Transcripts in plain text format (copy from Otter.ai, Grain, or your note-taking tool)
• Basic customer context: role, company size, and whether they are a new customer, churned customer, or long-term user

The quality of your JTBD output depends entirely on the quality of your interview data. If your transcripts are thin, the analysis will be thin. Good JTBD interviews ask customers to walk through the moment they first realised they had a problem, what they tried before, and what made them finally decide to act. If you do not have that depth in your transcripts, run a few new interviews before starting this process.

Step 1: Prepare Your Transcript for Analysis

ChatGPT works best when you give it clean, labelled input. Before pasting a transcript, do a quick clean-up:

1. Remove filler words and crosstalk if your transcript tool has not already done so
2. Label the speaker clearly (e.g. "Interviewer:" and "Respondent:")
3. Add a one-line context note at the top: role, company type, and customer status

Then open a new ChatGPT conversation and use this setup prompt:

Prompt:

"I am going to share a customer interview transcript with you. Your job is to analyse it using the Jobs-to-Be-Done framework. Do not summarise the transcript. Instead, identify the functional, emotional, and social jobs the customer is describing. Look for the language they use around struggle, progress, and desired outcomes. Wait for me to paste the transcript before you begin."

This primes ChatGPT to approach the transcript analytically rather than just summarising what was said. Summarisation is the default behaviour and it is not what you want here.

Step 2: Extract the Three Job Dimensions

Once you have pasted the transcript, run this extraction prompt:

Prompt:

"Based on this transcript, identify:

1. The functional job: what practical outcome is this customer trying to achieve?
2. The emotional job: how do they want to feel, or what feeling are they trying to avoid?
3. The social job: how do they want to be perceived by colleagues, managers, or stakeholders?

For each job, quote the specific line or phrase from the transcript that supports your identification. Do not invent jobs that are not evidenced in the transcript."

The instruction to quote supporting evidence is important. It stops ChatGPT from hallucinating plausible-sounding jobs that have no basis in what the customer actually said. Every job it identifies should be traceable back to a real line.

Step 3: Identify the Four Forces of Switching

The JTBD framework includes a model called the Four Forces, which explains why customers switch from one solution to another (or why they stay put). These forces are:

• Push: The frustrations with the current situation that are driving them away
• Pull: The attraction of the new solution
• Anxiety: The fears and doubts about making the switch
• Inertia: The habits and attachments keeping them in place

For SaaS positioning and sales messaging, this is valuable. Use this prompt after the job extraction step:

Prompt:

"Now analyse the same transcript through the lens of the Four Forces of switching:

1. Push: what frustrations or problems were driving this customer away from their previous solution?
2. Pull: what attracted them to a new solution or to our product specifically?
3. Anxiety: what doubts or fears did they express about making a change?
4. Inertia: what habits, relationships, or sunk costs were keeping them in place?

Again, quote the transcript to support each force you identify."

This prompt is particularly useful for sales teams. The push and pull forces map directly to objection handling and value proposition language. If you are refining go-to-market messaging after this kind of research, it can also be worth reviewing specialist B2B SaaS copywriters or comparing proven B2B SaaS inbound marketing agencies that know how to translate customer language into demand generation.

Step 4: Synthesise Across Multiple Transcripts

Single-interview analysis gives you one data point. The real value comes from pattern recognition across multiple interviews. This is where manual synthesis usually breaks down, and where ChatGPT saves the most time.

Run steps 1-3 for each transcript in separate conversations, then bring the outputs together in a new conversation for synthesis.

Prompt:

"I am going to share JTBD analysis outputs from [X] customer interviews. Each output includes functional, emotional, and social jobs, plus the Four Forces of switching.

Your task is to:

1. Identify the jobs that appear across multiple interviews (these are your primary jobs)
2. Identify jobs that appear only once or twice (these may be segment-specific or edge cases)
3. Identify any contradictions or tensions between what different customers said
4. Produce a ranked summary: the top 3 functional jobs, top 2 emotional jobs, and top 2 social jobs based on frequency and strength of language used

Here are the analysis outputs: [paste outputs]"

The ranked summary becomes your working JTBD map. It is the document you bring into positioning workshops, messaging reviews, and product prioritisation conversations.

Step 5: Translate Jobs into Messaging and Product Inputs

A JTBD analysis only earns its value when it changes something. Use these follow-up prompts to turn the synthesis into usable outputs.

For positioning and messaging:

"Based on this JTBD map, write three positioning statements for [product name]. Each statement should speak directly to one of the top functional jobs and address the primary emotional job. Use the language customers used in the interviews, not marketing language."

For homepage or landing page copy:

"Using the top functional and emotional jobs from this analysis, write a hero section headline and subheadline for [product name]. The headline should name the job. The subheadline should address the anxiety or inertia that stops people from switching."

For product roadmap input:

"Review this JTBD map and identify any functional jobs that our product does not currently address well. List them as potential feature or improvement areas, with the customer language that signals the unmet need."

Each of these prompts takes the raw analysis and converts it into something a specific team can act on immediately. If those insights are feeding into a site refresh, working with one of the best B2B SaaS web design agencies or vetted B2B SaaS webdesign UX experts can help ensure the message carries through into the actual user journey.

Common Mistakes to Avoid

Skipping the evidence instruction. If you do not ask ChatGPT to quote the transcript, it will generate plausible-sounding jobs with no grounding in real data. Always require evidence.

Using transcripts from bad interviews. JTBD analysis depends on customers talking about their decision process, their struggles, and their desired outcomes. If your transcripts are mostly feature feedback or satisfaction ratings, the analysis will not surface meaningful jobs.

Treating the output as final. ChatGPT is a synthesis tool, not a researcher. Review its outputs with someone who was in the interviews. Correct anything that misrepresents what the customer meant.

Running everything in one long conversation. ChatGPT loses context over long conversations. Keep each transcript analysis in its own conversation, then bring outputs together for synthesis.

How SaaS Hackers Uses This Workflow

At SaaS Hackers, this approach sits inside a broader research framework for B2B SaaS positioning. The JTBD synthesis step typically takes 2-3 hours across 8-10 transcripts, compared to 2-3 days of manual analysis. The output feeds directly into the messaging architecture work we do with clients.

The most consistent finding: the emotional and social jobs are almost always more powerful than the functional ones in driving purchase decisions, and they are almost always missing from SaaS messaging. ChatGPT, used properly, makes them visible.

FAQs

What is the best way to use ChatGPT for a Jobs-to-Be-Done analysis?

The best approach is to paste individual customer interview transcripts into ChatGPT with structured prompts that ask it to identify functional, emotional, and social jobs, with direct quotes as evidence. Run each transcript separately, then use a synthesis prompt to find patterns across all interviews. This produces a ranked JTBD map in hours rather than days.

How many customer interviews do I need for a JTBD analysis with ChatGPT?

Three interviews is the minimum for any pattern recognition to be meaningful. Five to eight interviews will surface the primary jobs reliably. Beyond ten, you are typically seeing diminishing returns unless you are segmenting by customer type, company size, or use case.

Can ChatGPT replace customer interviews for JTBD research?

No. ChatGPT synthesises data you give it. It cannot generate genuine customer insight from nothing. The quality of your JTBD analysis depends entirely on the quality of the interviews you conduct. ChatGPT accelerates the analysis step, not the research step.

What is the difference between functional, emotional, and social jobs in JTBD?

Functional jobs are the practical tasks a customer needs to accomplish. Emotional jobs describe how they want to feel (or avoid feeling) while doing it. Social jobs describe how they want to be perceived by others. In B2B SaaS, all three are present in every buying decision, but most products only address the functional job in their messaging. Teams that want to operationalise those insights across acquisition channels often end up comparing B2B SaaS SEO agencies, B2B SaaS content marketing agencies, or broader B2B SaaS digital strategy agencies.

Is the Jobs-to-Be-Done framework useful for early-stage SaaS products?

Yes, and arguably more so than for mature products. Early-stage SaaS teams often build based on assumed needs rather than real ones. Running a JTBD analysis on even 3-5 interviews before building a feature or writing positioning copy reduces the risk of building something customers will not hire your product to do. If you need outside support at that stage, it can help to review specialist B2B SaaS start-up agencies or browse the wider top agencies directory for relevant expertise.