How Earn Path ranks income options
Last updated 24 October 2026. This page describes exactly what the code in src/lib/engine/score.ts does.
On this page (10 sections)
Earn Path does not try to discover your "entrepreneur personality". It asks a narrower, more useful question: given your real skills, money, time, equipment, connectivity, distribution and willingness to sell, which income options are you most likely to make work in South Africa, and which should you avoid?
The ranking has four stages.
1. Hard exclusions
Some options are simply not available to you, however much you might like them. An option is removed from your ranking (and may appear in "What to avoid", with the reason) if any of these is true:
| Rule | Example |
|---|---|
| Age | Most freelance and task platforms require users to be 18+. |
| Time | You have fewer weekly hours than the option's minimum. |
| Equipment | It needs a laptop and you do not have one. |
| Connectivity | It needs reliable internet and yours is limited or absent. |
| Transport | It needs deliveries or meetings and you have no reliable transport. |
| Visibility | It needs you on camera (or on voice) and you ruled that out. |
| Money | Its realistic minimum startup cost is above the money you can genuinely risk. We use the lower bound of your band, halved if it is only partly spare, and zero if it is needed for essentials. |
| Audience | It depends on an existing audience or list that you do not have. |
| Selling channels | Every way customers are usually found is one you ruled out, or it needs direct outreach and you ruled out all direct outreach. |
| International payments | It pays through PayPal/Payoneer and you said you cannot use them. |
| Urgency | Even the best case for first income is slower than your deadline. |
| Location | It is hard to do from a rural area without fixed-line internet or nearby customers. |
| Skills | It needs a skill you rated two or more levels below what is required. |
2. Personal fit (0–100)
Each remaining option gets ten sub-scores (eleven if you took the full profile), each with a plain-English explanation in your results.
| Dimension | What it measures | Default weight |
|---|---|---|
| Skills fit | Your best relevant skill against the level needed; work samples add points. Options needing no skill score a neutral 50, because "anyone can do it" also means no advantage. | 26% |
| Sales fit | Your comfort with the channels customers actually come through. Options that require approaching people are capped by your comfort with direct outreach. | 16% |
| Time to first revenue | Typical days to first income against your deadline. | 14% (24% if you need money within a month, 6% if you can wait 6+ months) |
| Time fit | Your weekly hours against the option's minimum. | 10% |
| Distribution fit | Whether your audience, website, list, network or existing customers give you a head start. | 10% (6% if urgent) |
| Capital fit | Startup cost against money you can risk. | 8% |
| Income-target fit | Your monthly target against the realistic part-time ceiling for that option (minimal ≈ R800, low ≈ R3,000, medium ≈ R15,000, high ≈ R50,000). These ceilings are classes used for ranking. They are not earnings estimates. | 12% (16% if you want a full-time income) |
| Growth potential | How far it can grow beyond your own hours and whether income recurs. | 2% (8% if patient or full-time) |
| Work preferences | Client work vs assets, creative vs analytical, appetite for uncertainty. | 5% |
| Location fit | Local vs national vs global customers given where you live. | 3% |
| Value-profile fit (full profile only) | How strongly you use the value modes the option draws on (see section 5), minus small penalties for tasks you said drain you. Not used at all in the quick match. | 10% |
Weights are normalised to add to 100%. Preferences are deliberately a small weight: liking something does not make it viable.
3. Commercial viability (0–100)
Scored from the opportunity profile, independently of you: demand evidence (30%), competition (20%, inverted), risk (20%, inverted), complexity (10%, inverted), scalability (10%) and recurring revenue (10%). If an option benefits from an audience and you have one, viability gets +10, because your customer-acquisition cost is genuinely lower.
Personal fit and viability are shown separately so that "you'd enjoy it" never hides "the market is weak". If personal fit is high (70+) but viability is low (under 45), you see an explicit warning.
4. Overall score and evidence confidence
Overall = (65% × personal fit + 35% × viability) × evidence factor, where the evidence factor is 1.00 for high-confidence profiles, 0.97 for medium and 0.90 for low. Weakly evidenced options are not hidden, but they cannot outrank equally good, well-evidenced ones.
You see your top three (free) or up to five (paid report). Ties break on viability, then alphabetically, so results are deterministic.
5. Your value profile (full assessment)
The full assessment adds a personal profile built on Earn Path's own value-mode framework: eight ways people tend to create value (Creator, Analyst, Builder, Communicator, Persuader, Organiser, Teacher, Connector). Each mode gets two separate scores:
- Tendency comes from scenario questions (what you would do in six everyday situations), the work you did best and worst, trade-off questions (alone vs team, starting vs finishing, and so on) and what people ask you for help with.
- Proven comes from evidence: skills you rated competent or paid, things you have actually done (been paid outside a job, sold to a stranger, trained someone, built something people use…), your role and field of work, your network and existing customers.
The combined score is 45% tendency and 55% proven, because evidence predicts what people will pay for better than preference does. The profile flags untested strengths (high tendency, little proof: test cheaply before building on them) and learned capabilities (proof but little enjoyment: they can pay the bills, but watch for burnout). Every conclusion shows the answers it came from and a confidence level. Confidence is lower when tendency and proof disagree, when the top modes are close together, or when you said you were partly guessing.
Your motivators and the tasks that drain you are compared with each option (for example, chasing payment in client work, or negotiation in sales-led options) and listed as burnout risks in the report.
The framework is original. It is informed by well-established public ideas about vocational interests and work styles, such as John Holland's theory of six interest types (the general idea behind many interest inventories, including the O*NET® Interest Profiler). It does not reuse questions, scoring or descriptions from any commercial assessment, and it has not been validated as a psychometric instrument. That is why it carries only 10% of personal fit and can never override hard limits. Changing only your profile answers never changes which options pass your hard limits (this is tested). Full description: the eight value profiles.
Rand scenarios
Where we have credible South African price evidence, the paid report shows conservative, base and optimistic monthly figures: 20%, 40% and 60% of your weekly hours actually paid, at the low, middle and high end of the published price range, minus the relevant platform or payment fees and R300 a month for data and tools. These are arithmetic, not forecasts, and assume you have already found paying customers. Where price evidence is insufficient we say so and show no numbers.
The evidence behind each profile
Every opportunity profile records its sources (primary sources preferred: platform fee pages, SARS, Stats SA, regulators), the date each was accessed, a confidence level and a review status (draft, verified, needs review, retired). Draft and retired profiles are never shown or scored. Profiles are due for re-review every 90 days; the admin dashboard lists overdue ones.
What we never do: invent search volumes, earnings, testimonials or user counts; accept payment for placing an option higher; or recommend trading signals, pyramid schemes, task-group "jobs" or gambling systems.
Testing
The engine is tested against twelve personas (student, unemployed person with only a phone, professional, developer, teacher, creative, no-outreach introvert, website owner with low traffic, audience owner, rural resident, someone needing money urgently, and a patient builder). Test results are published in docs/research/persona-test-results.md in the project repository and the automated tests fail if a persona receives an option its constraints should exclude.
Limitations
- Answers are self-reported; over-rating your skills will produce over-optimistic results.
- Demand, competition and risk scores are researched judgements on a 1–5 scale, not measured market data.
- Platform rules and fees change. Each profile shows when it was last verified.
How we match your personality
Added 25 October 2026. Code: src/lib/engine/traits.ts, style.ts and score.ts.
The quiz now asks a short "How you like to work" section in both modes: what you most enjoy working with (people, things, numbers and details, or ideas), how you feel after a day with people, routine or freedom (full profile), what you are into (up to three), online or in person, and how physical the work can be. It also asks when you are free, how you get around, which spaces you could use, and which documents you have for getting paid.
Every path has matching traits (who and what you work with, how much contact, how physical, routine or freedom, online or in person, when the work happens, and whether it pays through a platform or needs a licence, a vehicle or a space). These are derived from each path's family and data with a few explicit overrides.
- New sub-scores, used only when you answered: personality fit (14%), interest fit (10%) and when-you-are-free fit (6%), normalised with the others.
- New exclusions: heavy physical work if you said light or seated work; platform and employer paths without an SA ID or valid permit; paths that need a driver's licence or a vehicle you do not have (unless you already do that work).
- Soft adjustments: standing work for people who need to sit (x0.85), online versus in person against your preference (x0.85), platform paths without a bank account (x0.9, with a "watch out" line about opening one), and work that is unlike how you like to work and about none of your interests (x0.9).
- Variety in the top three: if all three are the same kind of work, the best close match (within 8 points) of another kind moves to #3. If you said online or in person are both fine and all three are one type, the best close match of the other type (within 6 points) moves to #3.
- Location stays a small input: township-friendly paths get a 2% lift for people in a township, and nothing is lowered for where you live.
Each result shows a named work style (for example "The Practical Fixer" or "The Storyteller"), a five-part fit (personality, skills, money, time, location), reasons that quote your answers first, and one honest "watch out for" line.