Mining Jobs-to-be-done with AI

Building things people want is a game of trial and error: build something and see if people actually use it. Getting product-market fit is the art of fitting a product into people’s real life context — their situations and circumstances, struggles and problems, hopes, dreams, and desires. The problem is this context is shrouded and even lying dormant.

Can we uncover context beforehand?

Architects of Jobs to be done (JTBD), Clayton Christensen and Bob Moesta have argued that there are ‘no new jobs’. For example, throughout time, we’ve always had the same job of sending private messages. We've used messengers and chariots, carrier pigeons, the telegraph, The Pony Express and mail, email, SMS, WhatsApp — all to do the same job!?

There are no new jobs

Background: A job-to-be-done (JTBD) is the progress people are trying to make, where they ‘hire’ products to make progress in their life context, moving out of ‘struggling moments’ to ‘desired outcomes’.

The context that causes us to send messages hasn't changed. What's changed is that we can now use messaging technology in WAY MORE context. Before, only a handful of people could send messages (like a Marathoner warning Sparta about a Persian invasion) and now, anyone can send messages at anytime, like a 'good morning' to a loved one so you feel connected when apart. The context that causes someone to connect with a loved one in a different place has always been there; technology just awakened it. (Side note: It’s provocative to retrospectively imagine how often we wanted to connect with loved ones when we were apart but couldn’t. And then reverse our gaze to the future: How much more connected do we want to be when apart but can’t yet?)

So. If there are 'no new jobs’, then what if we could uncover this existing context? If so, we could create a lexicon -- a canon of human motivation -- to guide us in how to build the future.?

Mining podcasts for context using AI

So much incredible context is embedded in stories in podcasts. This American Life for example has so poignantly captured the human experience in their storytelling. Ira Glass recently shared their origin story with Esther Perel : "We would just walk up to people waiting for the subway and start conversations with them. And then if it was any good, we would just ride with them to their destination until we found a story."

I took one of my favorite TAL episodes "Rest Stop" and had ChatGPT-4 extract jobs-to-be-done from two of the people they interviewed, using this prompt:?

You are a world class Jobs-to-be-done researcher and analyst. Below is a transcript from an interview. Please extract the jobs-to-be-done using this syntax: “When [insert 'struggling moment'], I need [insert 'solution features'], so that [insert 'desired outcome']”. Use the participant’s specific, verbatim language. Be concise. Be extremely careful to capture nuance.

Story 1: How a cashier (Clara Dragon) handles difficult situations

Listen / read the transcript

JTBD A: When dealing with aggressive or rude customers, I need to maintain a genuine smile and remain silent if possible, so that I can defuse the situation and possibly make the customer less belligerent, leading to a more positive interaction.

JTBD B: When customers come in fuming due to external stresses like traffic or heat, I need to expedite their service and express sympathy, so that I can minimize the time they spend in a state of agitation and prevent any potential confrontations.

Story 2: How a father (Dan) navigates life post-divorce ?

Listen / read the transcript

JTBD A: When I'm seeing less of my children due to a divorce, I need to make seeing them more regular, so that I can maintain a strong relationship with them despite the changes in our family structure.

JTBD B: When I want to address issues with my ex but realize we don't have that relationship anymore, I need alternative ways to communicate and resolve conflicts, so that we can co-parent effectively and maintain peace for the sake of our children.

JTBD C: When contemplating the next half of my life after significant personal upheavals, I need to figure out what to do with my future, so that I can find fulfillment and purpose moving forward.

Context frames the future

Do these AI-extracted jobs-to-be-done provide the context to create new products?

With the Cashier, it is interesting how so much of her job wasn't taking orders, but actually helping calm customers down and create orderliness. She had learned the power of a genuine smile to defuse belligerence, and to expedite service to get people back on the road. This story is 15 years old and many solutions have likely already happened; mobile and kiosk orders solve a lot of the issues. However, a lot of ideas are still sparked:

• What if the ordering experience could relax and decompress people?
• What if cashiers weren't actually cashiers, but facilitators, asking people how their day was going while they waited for their order??
• What if the turnaround time was made explicit in whatever the interface was?

With the Divorced Father, we uncover some incredible struggling moments:

• How to maintain a strong relationship with your kids?
• How to co-parent when you no longer have 'that' relationship with your ex-partner?
• How to heal the heartache and re-invent oneself?

Imagine the solutions that could be created and fitted to these context! If one were building a kind of therapy, parenting, or coaching app, this context would be gold.

The Lexicon

GPT-4's distillations are still cursory, and might always be. Really understanding context still requires human creativity to exact nuance and meaning from the plethora of signals in stories.

However, what if there was a library of all these AI-extracted jobs, organized by the kinds of circumstances we all find ourselves in? I think we'd have a much more insightful and profound understanding of the human experience, and the context made explicit to fashion solutions fit to help people make progress.