The Compass.

Why this case exists

Most product cases start from a solution and work backwards to justify it. I wanted to write one that could fail. So the whole document hangs off a single hypothesis, and section one states the evidence that would force me to abandon it.

The hypothesis

The category sells breadth. Aptitude tests, career libraries, counsellor directories, all built to show a family how many options exist. My argument is that affluent families are not short of options and never were. They are short of conviction about one.

Writing the falsification test first changes what you build. It rules out the feature everyone asks for early, which is a bigger catalogue, and it makes the first deliverable a test rather than a launch.

Problem, and why now

The category imposes a real cost on the families it serves, and that cost is not the fee. It is the months of low grade doubt that follow a decision made on thin reasoning, and the cycle of re-litigating it every time a cousin gets into a better college.

The case sets out four forces that make this a now problem rather than a someday problem, covering how many more career paths now exist in India, how the parent buying this has changed, and what has become possible to build.

Three insights from discovery

The discovery programme produced three findings that shaped every later decision.

1. Affluent families want depth, not breadth

The category has been giving them breadth. Every additional option added to a shortlist lowers confidence rather than raising it, because nobody in the room has the time to reason about twenty paths properly.

2. The headline pain is not choosing the career, it is living with the doubt

Families describe the decision as the problem. What they actually describe when you push is the eighteen months after it, when there is no way to tell whether the decision was good and no mechanism for checking. That reframes the product from a recommendation engine to something that has to keep showing its reasoning over time.

3. The kid's voice arrives third in the room

Parent, spouse, then student. Pretending otherwise produces a product that the buyer will not pay for and the user will not open. The design has to be honest about that order while still giving the student a real say, because they are the eventual co-decider.

Users, as jobs to be done

Four parties sit in the decision, and only one of them pays.

• The primary parent. Primary economic buyer, and the person whose doubt the product exists to resolve.
• The co-buying spouse. Joint economic buyer, usually the one who raises the objection that stalls the purchase.
• The student. Observed user now, co-decider later.
• The family network influencer. Pays nothing, affects the decision anyway.

Segmenting this way rather than by demographics forces a priority order, and that order settles several product arguments that would otherwise stay open.

The product

V1 is four quarterly artifacts plus a dashboard that persists between them. The artifacts are the product; the dashboard is what keeps the reasoning visible in the gap.

• A calibration report that establishes the starting position
• A career shortlist, deliberately short
• A reality test plan, so the family can check the shortlist against the world
• A recommendation, delivered last and only once the tests have run

Engagement is designed around a time commitment constraint, because these are busy households and a product that needs weekly attention will not survive contact with them. The case makes five trade offs explicit rather than leaving them implied, and describes wireframe logic in words instead of pretending a mockup is a decision.

The taxonomy and the recommendation engine

Underneath sits a three layer career taxonomy and a documented mapping from a family's inputs to the paths it surfaces. Methodology transparency is treated as a product feature rather than an appendix. If a family cannot see why a path appeared, they will not act on it.

Business model

Pricing runs on two tiers, a founding family rate for the pilot cohort and a standard rate after it, with the cohort itself acting as the access mechanism. The case models unit economics per standard engagement, projects Y1 and Y2 directionally, and runs a sensitivity analysis to isolate the two variables that actually move the model. Everything else is noise.

Go to market is phased rather than launched, which matches a product whose first job is to test a hypothesis.

Metrics and prioritisation

Three metric layers, one for each primary job to be done, so nothing gets measured by proxy. Y1 funnel assumptions are stated as assumptions and labelled as such. Prioritisation uses RICE with units adapted to a services shaped business, where reach is a cohort of families rather than monthly actives.

Roadmap, by learning milestone

Each quarter is named for the thing it teaches, not the feature it ships.

• Q1, Y1Founding family recruitment and V1 build
• Q2, Y1Pilot launch and calibration report delivery
• Q3, Y1Expansion to 40 founding families, career shortlist for wave one
• Q4, Y1Reality test plan delivery and the move to standard pricing
• Q1 to Q2, Y2Methodology publication and school counsellor partnerships
• Q3 to Q4, Y2Hyderabad and Pune, first wave one recommendations delivered

Risks, regulation and ethics

The case closes on the five risks most likely to kill it, and on a privacy and regulatory posture that has to hold up given the product handles data about minors. That constraint is not a compliance checkbox, it changes what the product is allowed to store and for how long.