Jobs to be done (JTBD) is a way of understanding demand by asking what progress a customer is trying to make in a particular situation, rather than who they are or which features they want. People "hire" a product to get that job done and "fire" whatever they used before.
For conversion copy, the payoff is concrete: your headline names the progress, and your page answers the worries that stop people from switching.
The idea comes from innovation research, not CRO.
But it maps cleanly onto the copy decisions you test every week, and our database of A/B test results has a run of tests that show what happens when pages are rewritten around the customer's job.
The JTBD framework, in plain terms
The Christensen Institute defines a job as "the progress they're trying to make as they strive toward a goal or aspiration within particular circumstances," with functional, social, and emotional dimensions.
The circumstance matters as much as the goal.
The same person has different jobs at 7 a.m. and at 7 p.m.
The best-known example is the milkshake study Clayton Christensen taught at Harvard Business School.
A fast-food chain had asked people matching its milkshake-buyer profile what an ideal shake would be, acted on the answers, and saw no change in sales.
A researcher then watched the restaurant for a day and found that 40 percent of milkshakes were bought first thing in the morning by commuters ordering to go.
The job was "keep me busy and full on a long, boring drive," and the competition was bagels and doughnuts, not other shakes.
Treat that story as a teaching case, not data.
It is told secondhand and does not report a sales outcome for the changes.
What it illustrates is the shift: demographics may correlate with a purchase, but the job is what causes it.
Two schools you will run into
JTBD is not one method. Two camps use the same name for different tools, and it helps to know which one a guide is teaching.
| Outcome-Driven Innovation (Ulwick / Strategyn) | Switch / forces of progress (Moesta and Spiek) | |
|---|---|---|
| Core question | What steps make up the job, and how do customers measure success at each step? | Why did this person switch from their old way to a new one, and what nearly stopped them? |
| Main output | A job map and a list of desired outcome statements, then a survey scoring importance vs. satisfaction | A timeline of the purchase decision and a four-forces diagram |
| Research method | Interviews, then a quantitative survey of a large sample | Qualitative "switch" interviews with recent buyers |
| Best use in CRO | Choosing which benefits to lead with in a crowded category | Headlines, objection handling, and the language on the page |
Strategyn says Tony Ulwick developed the approach in 1990 and named the process Outcome-Driven Innovation in 1999. In that school, needs are measurable "desired outcomes," and Strategyn says a single market typically contains 50 to 150 of them.
That level of rigor suits product strategy. For a landing page, it is usually more research than you need.
The second school is the one most copywriters use.
Bob Moesta and Chris Spiek's forces of progress say a customer switches only when push plus pull outweighs anxiety plus habit:
- Push: the problem with the current situation that starts the search.
- Pull: the better life the customer pictures with the new option. Not your feature list.
- Anxiety: fear about the new choice ("What if it doesn't work?").
- Habit: the comfort of what they already do.
The forces give you a checklist for any page.
Does the headline name the push or the pull?
Does anything on the page reduce anxiety? Does anything make the switch feel small?
How to run a jobs to be done interview
The JTBD interview, often called a switch interview, is a slow, backward walk through one real purchase.
It is not a satisfaction survey or a feature wish list.
The goal is the story of a decision, told in the customer's words.
If you already run customer interviews, this is a specific script for one question: why did they buy?
Who to talk to
The Re-Wired Group's guide recommends recruiting people who switched in the last 30 to 90 days, because the details fade after that.
Include people who almost bought and didn't. Avoid recruiting only your happiest long-term customers; they remember the product, not the decision.
The same guide suggests 45 to 90 minutes per interview and says patterns often appear within 5 to 8 interviews for a given job.
That is a rule of thumb for qualitative research, not a sample-size calculation.
The interviews generate hypotheses; your A/B tests check them.
The interview script
- Set the frame. You are reconstructing what happened, like a documentary. There are no wrong answers and small details matter.
- Find the first thought. "When did you first start thinking about this?" It is often weeks or months before the purchase, triggered by a specific event.
- Walk the timeline. First thought, passive looking, an event, active looking, deciding, buying. Ask "what happened next?" and "where were you?" rather than "why?"
- Listen for the four forces. Don't ask about them directly. "The old one finally died" is push. "I'd seen a friend using it" is pull. "I wasn't sure it'd be worth it" is anxiety.
- Write down their exact words. Never feed them your vocabulary. A phrase like "I just got fed up" is raw material for a headline.
Two practical tips from the same guide: interview in pairs so one person can listen for forces, and deliberately misstate the sequence of events so the customer corrects you with the real story.
When you can't get interviews
Many stores can't book a dozen 60-minute calls.
A cheaper substitute is a post-purchase survey with open questions aimed at the forces: "What almost stopped you from buying today?"
(anxiety) and "What were you using before?" (habit).
Our guide to survey questions covers wording, and post-purchase surveys has the tests that came from them.
Surveys give you shorter answers and no timeline, but they scale.
From job to page: what the tests show
JTBD research produces two kinds of copy change: lead with the progress (pull), and answer the worries (anxiety). Both show up repeatedly in our database.
Lead with the job, not the product
Kasper Dolk reports that changing a product page hero headline to the customer's desired outcome, such as "Wake Up Feeling Like You Actually Slept" for a sleep supplement, was worth +$823,115.54 revenue per year for one client.
In another post, Kasper Dolk reports that an outcome headline plus a tagline handling the #1 objection lifted conversion around 10% across multiple clients. Both are self-reported results from social posts.
Vasta's test for Infinite Age is better documented: replacing broad wellness slogans with headlines under 12 words that led with the supplement's main outcome produced +35% completed purchases in an even 10-day split at 95% confidence.
Vasta tested two variants and does not say which one won.
The clearest JTBD-style test is older. Animoto replaced industry-specific sample videos (real estate, photography, online retail) with use-case videos that fit any industry (company overview, product description, testimonial) and saw a double-digit lift in conversions.
That is the milkshake lesson on a landing page: segmenting by what people are trying to do beat segmenting by who they are. The exact percentage was not published.
Conversion Fanatics found something similar on a catalog page.
Intro text that changed to match the use case a shopper selected produced +12.2% completed orders and +37.5% revenue per visitor, though no sample size was given.
Answer the anxiety in the customer's words
Kasper Dolk's team found three objections in customer surveys for a phone accessory brand: "is it durable?", "is it the official X?", "does it protect my phone camera?"
Adding benefit bullets answering them in the hero was reported as +$763,679.14 profit per year.
For a supplement brand, survey comments like "what if it doesn't work" led to a "how you'll feel" timeline section with risk reversal underneath, reported as CVR +14% while still being confirmed.
Oliver Kenyon's team saw the habit force at work on a wild bird food store, where first-time visitors had no reason to buy from that shop rather than the supermarket.
Three lines taken from post-purchase surveys, placed above the collection grid, were worth $44,890 per month.
Rebuild a flow around the job
Applica rebuilt the Five Minute Journal app's onboarding on Jobs-to-be-Done research, with personalized content, testimonials early, and clearer core benefits.
The result was +20% ARPU, about +10% trial starts and +14% trial conversions, with ARPU at p = 0.04. Several things changed at once, so the test shows the redesign worked, not which piece did it.
@oliverkenyon
Vasta
@kasper_dolk@kasper_dolkA caution on all of these: our database has many more winning tests than published losers, because companies publish wins.
JTBD-informed copy still loses tests. Treat each finding as a hypothesis to test on your own traffic, and don't forecast a lift from someone else's result.
Where JTBD goes wrong
- Writing the job as a feature. "Track my habits" is a feature. "Stop feeling like I start every day behind" is a job. If a competitor's product could fit the sentence, you are closer.
- Asking "why" too early. People invent tidy reasons. Ask what happened, and the reasons appear in the events.
- Interviewing the wrong people. Loyal customers from three years ago can't recall the decision. Recent switchers and near-buyers can.
- Treating JTBD as a replacement for personas. It answers a different question. See personas for when a profile of who buys is still useful.
- Skipping the test. Five interviews give you language and hypotheses. They don't tell you the size of the effect. That is what an A/B test is for.
For the rest of the research toolkit, start at customer research for CRO.