Jobs To Be Done

Status: Documented, not shipped · Evidence: P · Family: Strategy and opportunity · Verdict: reject (2026-06-03)

What it is

Jobs To Be Done (JTBD) reframes demand: instead of describing a customer by who they are (demographics, persona) or what product they bought, you describe the underlying “job” they were trying to get done, and you treat the product as something they “hired” to make progress on that job. The classic illustration is the milkshake a commuter buys at 7am - not because they like milkshakes but to do a job (stave off boredom and hunger through a long, one-handed drive), a job for which the milkshake competes with bananas, bagels, and a candy bar rather than with other milkshakes. The durable cognitive move is a reframe of the unit of analysis: shift from “what does the customer want from my product” to “what progress is the customer trying to make, and what are they really competing against to make it.” Done well, that reframe widens the competitive set, surfaces non-obvious substitutes, and anchors design and positioning to a stable target (the job changes slowly) rather than a volatile one (product features, customer attributes).

The complication that the popular write-ups hide is that “Jobs To Be Done” is not one method - it is a brand attached to two materially different schools that disagree about what a “job” even is, and the disagreement is not cosmetic.

• The jobs-as-progress school (Clayton Christensen, Bob Moesta and the Rewired Group, popularized by Alan Klement) defines a job as a process of moving from a current situation to a preferred one under real-world constraints, freighted with functional, emotional, and social dimensions. Its tools are causal interviews (“the switch”) and the progress-making forces (push, pull, anxiety, habit). It is a sensemaking and demand-framing lens, light on metrics.
• The jobs-as-activity school (Tony Ulwick / Strategyn, under the name Outcome-Driven Innovation) defines the job as a functional task the customer is trying to execute, then decomposes it into up to ~100+ measurable “desired outcomes” (each a direction-metric-object statement) and prioritizes by importance-versus-satisfaction. It is a quantitative needs-measurement and opportunity-scoring method.

So “use JTBD” can mean a one-page progress narrative or a multi-week 100-outcome survey study. Both are real, both are useful, and they are not interchangeable - which is the whole reason this entry exists as a documented-with-caveat page rather than a shipped skill.

When it helps / when it misleads

The reframe genuinely helps when:

• Positioning and demand are stuck on the wrong unit. When a team is optimizing the product against its category competitors and sales are flat, asking “what job is this hired for, and what is it really competing against” can break the frame open (the milkshake competing with bagels, not milkshakes). This is the move at its best.
• You need a stable target for discovery. A well-stated job outlives feature churn, so it is a durable anchor for a discovery or roadmap conversation.
• The functional task is decomposable and you want to find under-served needs - the Ulwick mode, where job-mapping the steps and measuring desired outcomes can locate where the current solutions leave the most opportunity.

It misleads or wastes effort when:

• The school is left unspecified. Because two schools wear the same name, “let’s do JTBD” routinely produces two people doing two incompatible things and arguing past each other. The single most common failure is method-ambiguity, not method-failure.
• The “job” is stated so broadly it becomes a tautology. “The customer wants to make progress” or “the customer wants a beautiful home” can be stretched to fit any product after the fact, which makes the framing feel profound while predicting nothing. Critics (see below) flag exactly this elasticity.
• It is treated as a measurement when it is a narrative (or vice versa). The progress school yields a hypothesis, not a validated need count; the outcome school yields metrics whose quality depends entirely on the survey craft. Confusing the two imports false precision or false rigor.
• It is pointed at a problem that is not about demand framing at all. JTBD is a way to frame what the customer is trying to accomplish; it does not decide between options, structure a problem space exhaustively, or stress-test a plan. Reaching for it as a general thinking move flattens it.

What the evidence says

The honest governing grade for the candidate’s stated move - “frame demand as the progress a customer is trying to make” - is P (practitioner). JTBD is a famous, durable, widely-taught practitioner framework with a clear lineage and real adoption, and that is the extent of what the record cleanly supports.

What the record does NOT support, and the laundering trap. I can locate no controlled or comparative study that tests the JTBD reframe itself - “does framing demand as a job produce better product or positioning decisions than not?” - against any alternative. The literature is of three kinds, none of which grades the move upward: (1) practitioner books and HBR articles that assert and illustrate the framing (Christensen et al.; Ulwick; Bettencourt and Ulwick); (2) vendor case material; and (3) the academic critique of Christensen’s broader theory program, which is skeptical rather than confirmatory.

The figure most often quoted to make JTBD look evidence-backed is Strategyn’s claim that Outcome-Driven Innovation has an “86% success rate” versus a 17% industry average. That number must be quarantined, not laundered. By Strategyn’s own description it is an internal tally drawn from its own client base - reported in various places as on the order of 18 of 21 product launches across a few dozen Strategyn engagements rated successful by the sponsoring companies themselves. It is self-reported, self-selected, vendor-measured, with no independent control, no published method for how “success” was scored, and no third-party replication. The contrasting “17% industry average” is stitched from a dozen unrelated secondary sources measuring different things. Under this library’s evidence rule the 86%/17% pairing is excluded as fact: it is a vendor marketing statistic, not a research finding, and it does not influence the grade. (It is named here only so the reader knows precisely which number to distrust.) Similarly, Ulwick’s frequently-cited Cordis case - market share reportedly rising from ~1% to ~20% - is a single uncontrolled client anecdote, not evidence the method caused the outcome.

What there is instead. There is one nameable, rigorous body of work, and it cuts toward caution rather than endorsement: the academic debate over Clayton Christensen’s theory program. Jill Lepore’s 2014 New Yorker critique (“The Disruption Machine”) argued the disruption framework was retrofitted to cases and behaved more like a “secular religion” than a tested theory; Andrew King and Baljir Baatartogtokh’s 2015 MIT Sloan Management Review study examined 77 of Christensen’s own disruption cases and found the theory’s core predictions held in only a minority of them. That work is about disruption theory, not JTBD specifically, but it bounds the credibility of the same author’s adjacent, even less-formalized JTBD claims: a famous theory from the same source has been shown to be loosely specified and over-generalized, which is reason for conservatism, not confidence, about the unformalized cousin.

Transfer caveat (required). All of the supporting material - books, HBR pieces, vendor cases - describes human marketers and product teams in business settings. None of it studies JTBD performed by or with an AI agent. The evidence is transferred from human commercial contexts and is not validated for AI-augmented use. There is no S- or M-tier research on this move to borrow from, so the grade is P on its own merits, with the one rigorous adjacent body of evidence (the disruption-theory critique) pointing down, not up.

Why it is / is not a skill here

Verdict: Flag - documented with caveats, not shipped; and substantially a pm-domain method. The registry reasoning is “multi-school ambiguity; also a pm-domain method. Include only with the ambiguity caveat,” and both halves of that reasoning are load-bearing.

The ambiguity caveat is the first gate. A skill in this library has to teach one well-specified cognitive move with consistent steps and a single honest evidence claim. JTBD cannot satisfy that as a single skill because the name resolves to two genuinely different methods - jobs-as-progress (a qualitative demand-framing narrative) and jobs-as-activity / Outcome-Driven Innovation (a quantitative ~100-outcome measurement study). The split is not a nuance; the principals fought publicly over it (Klement’s “two very different interpretations” essays and the multi-round Klement-Ulwick debate), and Christensen’s own early citation of Ulwick’s work as evidence is part of the tangle the two schools spent years trying to unwind. A skill that says “do JTBD” would either silently pick one school (mislabeling) or blur both (incoherent steps, and an evidence claim that cannot be true of both at once). Documenting the framework with the ambiguity caveat is the honest service; shipping it as one skill would manufacture a false unity.

The domain caveat is the second gate. Even resolved to a single school, the remainder is product-and-marketing-domain work, not a general-purpose thinking move. The family is the catalog’s explicitly weakest cross-domain fit - “many defer to pm-skills” - and JTBD’s neighbors here (Porter’s Five Forces, Blue Ocean tools, the opportunity-solution tree, value-proposition contrast) are tagged pm or flag for the same reason. JTBD’s natural home is product discovery: defining the job, mapping the forces, scoring outcomes, framing positioning. That remainder already has a home in the sibling pm-skills library, which ships a dedicated define-jtbd-canvas skill capturing the functional, emotional, and social dimensions of a job. Re-shipping a JTBD skill here would duplicate the sibling library’s product-domain coverage while importing the multi-school ambiguity this library is built to avoid.

Why flag rather than fold or out-and-out reject. A clean fold would require a single shipped move that already produces JTBD’s contribution, and there is a partial one - the demand-reframe (“what is the customer really trying to accomplish, and what is it competing against”) overlaps the kind of problem-reframing the catalog’s framing skills do, and the pm-skills canvas covers the product version. But folding would erase the precise warning that is this page’s value: that “JTBD” names two methods, and that the most cited evidence for one of them is a quarantined vendor statistic. A bare reject would throw away a genuinely useful reframe. So the honest verdict is to document it, name both schools, attribute it, quarantine the laundered number, and point the product use case to pm-skills - which is exactly what flag encodes. The learning value of the NO: a famous, genuinely useful framework can fail the bar not because the idea is weak but because the name is overloaded and the evidence most often waved at it is a vendor’s own scorecard. A library that ships one well-specified move per skill, with one honest evidence claim, cannot ship a brand that resolves to two methods and leans on a self-reported success rate.

Lineage and who to read

JTBD has two origin lines that later collided under one name.

• The phrase and the progress framing trace to Clayton Christensen (Harvard Business School), who introduced “jobs to be done” loosely in The Innovator’s Solution (with Michael Raynor, 2003) and developed the marketing argument in “Marketing Malpractice: The Cause and the Cure” (Christensen, Cook and Hall, Harvard Business Review, 2005). The milkshake study is usually credited to Bob Moesta and the Rewired Group, who ran the fieldwork. The school’s fullest statement is Competing Against Luck (Christensen, Taddy Hall, Karen Dillon and David Duncan, 2016), summarized in the HBR article “Know Your Customers’ Jobs to Be Done” (Christensen, Hall, Dillon and Duncan, 2016). Alan Klement popularized and sharpened this “jobs-as-progress” reading in When Coffee and Kale Compete (2016) and in his “two interpretations” essays.
• The measurement line is older and independent: Tony Ulwick (founder of Strategyn) began the work in the early 1990s (the Cordis engagement), named his process Outcome-Driven Innovation in 1999, and laid out the “desired outcomes” method in “Turn Customer Input into Innovation” (Harvard Business Review, 2002), in What Customers Want (2005), and in Lance Bettencourt and Anthony Ulwick, “The Customer-Centered Innovation Map” (Harvard Business Review, 2008), which introduced job mapping. Ulwick’s book-length statement is Jobs to Be Done: Theory to Practice (2016).

“Jobs To Be Done” / “JTBD” is in broad generic use across both schools and is treated descriptively here. “Outcome-Driven Innovation” / “ODI” is a registered mark of Strategyn; it is named with attribution and its self-reported success statistic is quarantined above. For the critical context that bounds the broader theory’s credibility, read Lepore (2014) and King and Baatartogtokh (2015) on disruption theory, and pair any JTBD source with the awareness that you must first ask which school it belongs to.

Named sources

• Clayton M. Christensen and Michael E. Raynor, The Innovator’s Solution (Harvard Business School Press, 2003). Where Christensen first uses “jobs to be done” loosely; the looseness is itself part of the later multi-school ambiguity. Practitioner / foundational. (P)
• Clayton M. Christensen, Scott Cook and Taddy Hall, “Marketing Malpractice: The Cause and the Cure,” Harvard Business Review (December 2005). The progress-school marketing argument and the milkshake framing. Practitioner. (P)
• Clayton M. Christensen, Taddy Hall, Karen Dillon and David S. Duncan, Competing Against Luck (HarperBusiness, 2016), and the companion HBR article “Know Your Customers’ Jobs to Be Done” (2016). The fullest progress-school statement; assertion-and-illustration, no controlled test. Practitioner. (P)
• Anthony W. Ulwick, “Turn Customer Input into Innovation,” Harvard Business Review (January 2002), and What Customers Want (McGraw-Hill, 2005). The outcome-driven / desired-outcomes school. Practitioner. (P)
• Lance A. Bettencourt and Anthony W. Ulwick, “The Customer-Centered Innovation Map,” Harvard Business Review 86(5) (May 2008): 109-114. Introduces job mapping (the eight universal job steps). Practitioner. (P)
• Alan Klement, When Coffee and Kale Compete (2016) and “Know the Two - Very - Different Interpretations of Jobs to be Done” (jtbd.info). The clearest statement of the schism this entry turns on. Practitioner / critical. (P)
• Andrew A. King and Baljir Baatartogtokh, “How Useful Is the Theory of Disruptive Innovation?,” MIT Sloan Management Review 57(1) (Fall 2015). Examined 77 of Christensen’s own disruption cases and found the theory’s predictions held in a minority; the rigorous adjacent evidence, and it points down. Academic / critical. (M, for disruption theory - not a test of JTBD)
• Jill Lepore, “The Disruption Machine,” The New Yorker (June 23, 2014). Argues the disruption framework was retrofitted to cases; bounds the credibility of the same author’s adjacent, less-formalized claims. Critical essay. (Critical)

Excluded under the evidence rule: Strategyn’s “86% success rate versus 17% industry average” for Outcome-Driven Innovation is a self-reported, self-selected, vendor-measured figure from Strategyn’s own client base, with no independent control or published scoring method, and is excluded as fact - it does not influence the grade and is named only to mark which number to distrust. The Cordis “1% to 20% market share” case is a single uncontrolled client anecdote and is likewise not counted as evidence of effect.