1. Verdict
Why a ladder instead of one target
This idea is assessed at six MRR levels instead of one fixed target, like the rest of the portfolio. For a small data API the useful question isn't "will it hit $20k?". It's how likely each level is, and what each level costs in hours.
The anchor: what base rate?
There is no verified base rate for bootstrapped B2B data APIs reaching MRR levels from launch.
- MicroConf's State of Independent SaaS surveys founders who already run businesses. It measures survivors, not launches, and its secondary figures conflict.
- RevenueCat 2026 is verified, but covers mobile subscription apps. Across all categories, 17.3% reach $1k a month within 2 years and 4.6% reach $10k. The Business category is lower, at 14.7% and 1.6%. A B2B API needs far fewer customers for each level, so those figures understate the odds at $1k. The Indian buyer pool for this product is small, so they probably overstate them at $20k and above.
The odds below are judgment estimates, anchored to these comparables:
| Comparable | Dataset | Path | What it teaches |
|---|---|---|---|
| IPinfo | IP geolocation | Weekend project 2014 → paid tiers $50/$100/$200 in 2015 → "a few hundred thousand a year" end of 2016 → ~$1M ARR 2017 → multi-million, +50% a year, 15 staff, 100K+ customers (founder interview) | Data APIs compound for years, and churn is low once embedded. It grew through Stack Overflow answers, and it launched when the space was thin. |
| Open Exchange Rates | FX rates | Founded 2011, 1 employee (Dec 2024), still selling at $12 / $47 / $97 a month | A commodity feed can last for 15 years with one person, but price is pinned near the free sources. Revenue isn't disclosed. |
| ipdata | IP geolocation | Founded 2017, unfunded, 25M calls a month on ~$150 a month of AWS (company blog, per the catalog) | Infrastructure cost is trivial. The cost is data quality. |
| OpenCorporates | Company registries | £598,370 profit in FY2024 with 45 staff (UK filing, per the catalog) | Company data works at scale, with an open-data mission, many years and a team. It's not a solo shape. |
| apilayer | fixer, ipstack and others | Bought by Idera in Jan 2021, price undisclosed (per the catalog) | There is an exit route: portfolios of small data APIs get rolled up. |
All of these are survivors. Many data APIs never reach $1k. None of them started in 2026 against a free, MIT-licensed library covering 250 countries.
Step 1: which dataset? Six tests decide
The brief asked for data that is hard to assemble, changes often, and is paid for by businesses. Two more tests were added: a solo founder must be able to keep it accurate, and it must be legally safe to publish. Each is scored 1–5.
| Candidate | Hard for the customer to assemble | Changes at short notice | Businesses pay, and wrong data costs them | Gap in free and cheap sources | Solo founder can keep it accurate | No licence or legal risk | Total /30 |
|---|---|---|---|---|---|---|---|
| IP geolocation | 5 | 4 | 4 | 1 (MaxMind GeoLite, free tiers) | 1 (needs probes, BGP, scale) | 4 | 19 |
| Global financial market calendars | 4 | 3 | 5 | 2 (Copp Clark, TradingHours.com $50–150/mo) | 1 (1,200+ markets) | 4 | 19 |
| Indian company (MCA) data | 3 | 4 | 4 | 2 (Apify wrappers from $2.50/1,000; Tofler; data.gov.in) | 2 | 2 (portal scraping, terms) | 17 |
| Exchange rates | 1 | 1 (daily, predictable) | 3 | 1 (ECB, RBI free; $10–97/mo APIs) | 5 | 4 | 15 |
| Indian GSTIN verification | 2 | 3 | 4 | 1 (dozens of APIs at rupees per call) | 3 | 2 (GSP licence for scale) | 15 |
| Global public holidays | 2 | 2 | 2 | 1 (python-holidays, Nager.Date free; Calendarific $12) | 2 | 5 | 14 |
| India bank, payment-rail and exchange business days | 4 (34 RBI centres, HTML only; exchange circulars; state gazettes) | 4 (declared days ahead: 22 Jan 2024; 31 Mar 2025) | 3 (lenders, fintechs, brokers; NACH auto-shift softens it) | 4 (python-holidays has no RBI holidays; RBI has no API) | 4 (bounded: 34 centres, 3 exchanges, 36 states) | 5 (government facts, primary sources) | 24 |
Scores are judgment. The sources are in research/10.
Why India business days wins:
- The official source is not machine-readable. RBI's matrix is an HTML table by office centre and month. Every buyer re-keys it, and re-keys it again when it changes.
- The free library leaves exactly this gap. python-holidays covers Indian states and the NSE, but its India module has no RBI or Negotiable Instruments Act category (checked 30 Sept 2026).
- It's bounded. 34 centres, 3 exchanges and 36 states is a dataset one careful person can keep right. 250 countries isn't.
- Freshness is the product. A list you can copy once a year. A change that lands two days before a holiday you can't, unless someone is watching.
Its weak point is score 3 on "businesses pay." NPCI already moves NACH debits off holidays automatically, so the most expensive failure is handled for lenders. That's the question the day-45 kill test answers.
Odds at each MRR level (reached by month 24)
| MRR at month 24 | Base rate | As described: generic data feed | India company data | Global financial calendars | Recommended, ~15 hrs/week | Recommended, full-time |
|---|---|---|---|---|---|---|
| $1k | No verified base rate | 20% | 18% | 15% | 35% | 50% |
| $3k | No verified base rate | 9% | 8% | 8% | 20% | 32% |
| $5k | No verified base rate | 5% | 4% | 5% | 12% | 21% |
| $10k | No verified base rate | 2% | 2% | 3% | 5% | 10% |
| $20k | No verified base rate | <1% | <1% | 1% | 2% | 4% |
| $50k | No verified base rate | <0.5% | <0.5% | <0.5% | <0.5% | 1% |
The "as described" and other candidate columns assume full-time work. The as-described column matches the "niche data feed" column in ../../docpress/analysis/01-verdict.md.
Recommended means the India business-day API set out in WHAT-TO-DO-AND-AVOID.md, with the 3 conditions met:
- seven-day data cover, with changes published within 4 hours;
- direct access to 30 buyers in month 1;
- 5 paying design partners by day 45.
If any one condition is missing, multiply the recommended odds by about 0.6.
These are judgment estimates, not computed statistics.
How to read the ladder
- $1k–3k is realistic. It takes 17–52 firms at ~$58 a month, and the pitch is concrete: "your calendar was wrong on 22 Jan 2024; ours wasn't".
- $5k needs a second segment: brokers and wealth apps, or global firms with Indian operations on the USD plan.
- $10k needs a second product line (bank cut-off times, settlement windows) or a second country. Both add upkeep.
- $20k isn't a plan. It's ~345 Indian firms paying for something most of them can copy for free.
- Hours matter less than coverage. The product is small. What hours buy is a faster response to changes and more sales calls.
Portfolio definitions (for comparison with other ideas)
| Tier | Definition | Recommended, ~15 hrs/week | Recommended, full-time | As described |
|---|---|---|---|---|
| Tier A (venture) | $1M ARR within 36 months, ≥70% margin, ≥50% month-12 logo retention | <0.5% | <0.5% | <0.5% |
| Tier B (profitable business) | $20k+ MRR within 24 months, profitable, ≥80% gross margin, ≥70% month-3 retention | ~2% | ~4% | <1% |
Compared with the other ideas in this portfolio:
- It's the lowest in the portfolio at $10k and $20k. DocPress, the other developer API, is at 12% / 22% for $10k because a legal deadline forces buyers to act. FeedWise has no deadline.
- At $1k it's close to BhashaCity (35% / 50%), and cheaper to run. The build is small and the costs are near zero.
- Its best feature is durability. Once embedded, a business-day API gets left alone for years. Its worst is the ceiling.
Fundable in current form?
No. It's a small dataset for one country's developers, the free substitutes are good, and there's no team or traction. It doesn't need funding either. See 06-fundability.md.
One-paragraph verdict
The catalog's instinct is sound: data APIs that get embedded churn little and can compound for a decade, as IPinfo shows. But the specific datasets listed are mostly free now, or owned by incumbents whose moat is breadth. Scored on the brief's own tests, only one candidate clears the bar: Indian bank, payment-rail and exchange business days, where RBI publishes a 34-centre HTML table, the free library skips bank holidays, and dates move with a few days' notice. Sell freshness and proof, not the list: change alerts within hours, a source on every date, and a history an auditor can query. Then test hard, early, whether anyone pays: 5 design partners by day 45, or stop. $1k–3k MRR is a sound expectation, $5k is a good result, and $10k is a stretch. Anyone who wants $20k should pick a different idea from this portfolio.