Jobs To Be Done (JTBD)

Reject the category's stated job and write the honest one

The category assumed the job was become fluent in Spanish. The behaviour says otherwise: sessions run three to five minutes, they cluster in commutes, queues, lift lobbies and the ten minutes before sleep, and a large majority of users never finish a course tree. If fluency were the job, usage would look like a course with a completion curve; instead it looks like a habit with a daily rhythm. The solution-free job statement is closer to turn a few idle minutes into something that feels like progress, without effort or embarrassment. Notice that this statement contains no mention of language, which is exactly why it explains the product.

Segment by circumstance and discover there are two businesses, not one

Cut usage by the moment rather than by learner age or country and two very different clusters appear. Cluster one is the micro-session habit user with no deadline, no exam and no measurable consequence for quitting. Cluster two is the deadline learner: someone who must demonstrate English proficiency for a university application or a visa by a fixed date, where failure has a real cost. These are different jobs with different willingness to pay, which is why Duolingo runs the free habit app for cluster one and sells the Duolingo English Test, a proctored certification priced around 65 dollars (approximate), to cluster two. One circumstance cut, two products, two price architectures, and no confusion about which customer pays what.

Unpack the functional, emotional and social layers

Functionally the app converts dead time into a task that produces a visible score, which is a very low bar compared with teaching fluency. Emotionally it sells relief from the guilt of wasted time, plus the sensation of improving, and this is the layer Rosetta Stone never touched because a 200 dollar purchase produces guilt rather than relieving it. Socially it sells identity: a 400-day streak, a Diamond League position, and the right to say at a dinner table that you are learning Italian. Strip out the emotional and social layers and the functional core is worth close to zero, which is precisely why the company can afford to give it away free.

Map the real competitive set including doing nothing

The honest competitor list for that three-minute queue is Instagram, TikTok, a mobile game, a podcast, WhatsApp, and staring out of the window. Duolingo's leadership has said publicly and repeatedly that it competes for attention with entertainment apps rather than with Babbel or Rosetta Stone. Two consequences follow immediately and both are decisive. The engagement benchmark becomes a social app, so daily active over monthly active ratio, notification timing and streak mechanics become the core metrics rather than lessons completed. The price benchmark becomes zero, because scrolling is free, which makes any upfront-paid model structurally unable to win that occasion.

Run the four forces on why the old category never converted

Push was weak: almost nobody has an urgent, dated need to learn Italian, and a weak push is why the paid category stayed small no matter how good the pedagogy got. Pull in the old model was a distant promise, fluency in some number of months, which is worth very little against a three-minute attention window. Anxiety was the binding force: a few hundred dollars upfront plus the private fear that you are simply bad at languages and will waste the money, a fear reinforced by every previous abandoned attempt. Habit was the phone already being in your hand with an app that never asks you to feel stupid. Duolingo attacked anxiety and habit rather than pull, which is the whole strategy: free removes financial risk, three-minute lessons remove time risk, hearts and gentle correction remove embarrassment risk, and notifications plus streak loss aversion hijack the existing phone habit.

Find the underserved outcome and build only there

Lay the job out as steps: decide to start, find a moment, complete a session, see progress, come back tomorrow. Score each on importance and satisfaction in the pre-Duolingo world. Complete a session was already well served, the old software taught perfectly well. Come back tomorrow was high importance and near-zero satisfaction, since abandoned language software was the norm. So the underserved outcome is minimise the likelihood that I quit before I feel progress, and every famous product investment maps to it: streaks, streak freezes, weekly leagues, home-screen widgets, notification timing tuned to each user's habitual slot. Monetisation then rides on top of the habit rather than in front of it, with the subscription removing hearts and ads for people already hooked, and the AI conversation tier serving the minority whose job has genuinely grown into speak to a human.

Close the loop on economics before recommending anything

A free product against a free competitor only works if the arithmetic survives. On approximate public shape, roughly 100 million monthly actives converting at eight to ten percent into paid, with blended revenue per monthly active of only a few dollars a year, means the model cannot tolerate meaningful paid acquisition. That is the constraint that explains the rest of the strategy: organic app-store discovery, the owl mascot's deliberately absurd social presence, and word of mouth off streak-sharing, all of which push customer acquisition cost close to zero. If you had run the JTBD analysis and then recommended a premium repositioning, you would have produced a lovely insight attached to a business that cannot pay for its own distribution.

Refuse the demographic cut and ask for the trigger cut

The client's segmentation is age band, income band and city tier, and it is useless here because buyers and non-buyers look nearly identical on all three: a 34-year-old salaried man in Indore is equally likely to buy and not to buy. The high-signal question to the interviewer is what happened in the 30 days before purchase, which reframes the data from who to when. Four circumstance clusters fall out: the tax-saving window between January and March under Section 80D, a hospitalisation of a parent or close relative in the last 60 days, a life-stage event such as marriage, a first child, a home loan or parents crossing 60, and loss of employer group cover on a job change or a move to freelancing. In this case roughly 45 percent of purchases sit in the January to March window and about 20 percent follow a hospitalisation, illustratively. Four circumstances, four jobs, and crucially four different competitor sets.

Write a solution-free job statement for each trigger

Tax window: reduce this year's tax outgo before the 31 March deadline with a payment I can defend to myself as useful rather than wasted. Post-hospitalisation: make sure a four lakh rupee bill never lands on my savings again. Life stage: prove to myself and to my family that I have taken responsibility for people who now depend on me. Employer-cover loss: replace the safety net my company used to provide before anything happens. Read the first one again and notice that health barely features in it, which tells you the tax buyer's true rivals are ELSS, PPF and term insurance, and also tells you exactly why that cohort lapses in year two: once the deduction is claimed, the job is complete and the policy has no further purpose.

Unpack the three layers on the highest-intent job

Take the post-hospitalisation trigger, which converts at several times the base rate. Functionally the buyer is transferring a low-probability, high-severity liability of three to ten lakh rupees off a household balance sheet that cannot absorb it. Emotionally they are buying an end to the 3am arithmetic of what happens if papa is admitted again, which is a fear that has just been made vivid and specific by an actual discharge bill. Socially they are buying the standing of being the child who handled it, said in front of relatives and siblings, and this layer is why buyers routinely pay three to four times more premium for a parents' policy than for their own even though it is functionally the worse purchase per rupee of expected claim. Any pitch that speaks only to the functional layer is competing on price in a category where price is not the decision.

Map the real competitive set, and find that non-consumption is the market leader

The client benchmarks premiums against other insurers, which is the wrong list. The genuine competitor set is the five lakh rupee fixed deposit the family calls the emergency fund, gold that can be pledged at a jeweller or a gold-loan branch within hours, borrowing from siblings and in-laws, the employer group cover that is already there and costs nothing, a crowdfunding page on a platform like Ketto or Milaap, the government scheme for eligible households, and by far the biggest of all, doing nothing. Retail health penetration in India is low, so the largest single share holder in this market is non-consumption. That single reframe repositions the entire pricing problem: an 18,000 rupee premium is not fighting a 16,000 rupee rival policy, it is fighting 18,000 rupees that would otherwise sit in an FD visibly earning interest and remaining the family's own money. Against that rival, a 10 percent price cut moves nothing, because the objection is not that the policy is expensive, it is that the money disappears.

Run the four forces and name the binding one

Push is weak for ten months of the year, because the pain is entirely hypothetical, and it spikes only in the discharge lobby or against the 31 March deadline. Pull is moderate and already maxed out: the client has the best network and the best settlement ratio and it has not helped. Anxiety is the binding force and it is very specific in this category: they will reject the claim on a technicality, the pre-existing disease waiting period of three to four years means I am paying for years of nothing, and room-rent sub-limits mean I will end up paying a large share of the bill anyway. Habit is the second constraint, because the employer cover and the FD both already exist and require no action at all. Push plus pull is losing to anxiety plus habit, which immediately disqualifies both board options: a price cut attacks pull and simultaneously removes the funding for the fixes that would attack anxiety, and more advertising attacks awareness in a category where people already know what health insurance is.

Convert the binding forces into specific interventions

To kill anxiety, move every dispute from claim time to sale time: a 20-minute video medical that pre-underwrites and puts existing conditions in writing on the policy schedule, a published per-hospital ceiling stating the maximum the customer will pay out of pocket for a cashless admission, removal of room-rent sub-limits on the flagship product, and a written claim-decision SLA with an automatic penalty paid to the customer if it is missed. To kill habit, stop asking people to abandon the free alternative and sell alongside it: a top-up product priced explicitly against employer cover, your company gives three lakh, buy ten lakh on top of it for about 6,000 rupees, plus a monthly premium option, because 1,500 rupees a month is read by the household as an expense while 18,000 rupees a year is read as a loss of capital that could have stayed in the FD. To show up at the trigger, place a desk and a QR code at the discharge counter of the top network hospitals and run the 80D calculator campaign only in January to March, because the correct message is tax in January and fear in the discharge lobby, and using one message for both is why the funnel leaks.

Size the trigger and pressure-test the economics

Size the discharge trigger alone. If the network has roughly 4,000 hospitals, and each sees about 120 discharges a year of uninsured or under-insured adults in the target income band who paid cash, that is about 4.8 lakh struggling moments a year with the push force at its absolute maximum. Capture 4 percent at the discharge desk, which is conservative for a warm trigger against a cold funnel converting at 6 percent, and that is about 19,000 policies at 18,000 rupees, roughly 34 crore rupees of new written premium. At a combined ratio near 95 percent the first-year underwriting margin is thin, so the case does not pay on year one, it pays on persistency: a buyer whose job was triggered by a real bill renews far better than a tax-window buyer whose job ends on 31 March, so model 85 percent renewal against the portfolio's 70 percent and the value sits in renewal years two to five. Finally check whether the fix pays for itself: if the pre-underwriting video medical costs about 700 rupees a policy against a current acquisition cost of roughly 2,500 to 3,000 rupees, it is affordable, and it should reduce claim disputes and therefore lapse.