Growth engineering
SEO / lead gen
11 tools shipped
Claude Code
A free tool suite that earns the lead instead of buying it
Lesser does flat-fee US tax filing for Indians living in America. Paid acquisition was expensive and nearly impossible to attribute. So I built eleven free calculators that answer the questions this audience is already searching for — each one ending with a route to a CPA.
the problem
Ads to this audience were expensive and unattributable
A Meta campaign to 29,000 contacts produced no identifiable registrations — when I checked the first 25 sign-ups against the owned lists, 23 were already on them. We were paying to reach people we already had.
Meanwhile the actual buyer was doing something far more useful: searching "substantial presence test calculator" at 11pm, trying to work out which US tax return they owe. Not browsing. Stuck, mid-decision, and about to hand money to whoever answered first.
The tools do not interrupt anyone. They answer the exact question someone is already asking, for free, and then show them what the hard version of that question looks like.
THE PREMISE
the positioning call
Everyone else builds India-side. I built the mirror.
I researched thirteen competitor tool sites. Belong, SBNRI, iNRI, Savetaxs and the rest all build India-side calculators — GIFT City funds, FD rates, Indian residential status, rupee-dollar. That space is saturated and the traffic is poorly qualified for a US filing service.
The gap was the US side: what the IRS expects from someone who moved from India. Same person, opposite end of the problem, almost nobody serving it.
Built — US side
- Residency. Which return you owe, and the exempt years every other calculator skips
- Equity. RSU and ESPP, including the wrong cost basis brokers report
- Foreign accounts. FBAR, FATCA, PFIC, gifts from parents
- Moving back. The RNOR window, which is worth real money if planned
Deliberately skipped
- HRA, 80C, 80D, indexation. Commodity India calculators a hundred sites already rank for
- Rental depreciation. 1,730 searches a month — the single highest on the list
- Mostly US-born landlords with no India connection. Great for a traffic chart, worthless for pipeline
- Volume was never the metric. Qualified intent was
The call I had to defend: turning down the highest-volume keyword on the research list. A rental depreciation calculator would have roughly doubled the suite's traffic and contributed nothing to revenue. Picking the 210/mo tool over the 1,730/mo one only makes sense if you are measuring the right thing.
the eleven
Every one works. Click any of them.
Self-contained HTML, no backend, nothing leaves the visitor's browser — which matters when you are asking someone to type their bank balances in.
how it was built
Research first, then keyword data, then code
01
Mine the demand
6,576 autocomplete queries scraped, then re-ranked against Google Keyword Planner. Found people search form names, not concepts — "form 8621" at 4,400 vs "pfic indian mutual funds" at 20.
02
Map the field
13 competitor sites researched. Five rendered only in JavaScript and had to be driven in a headless browser to read at all.
03
Verify every figure
Brackets and thresholds pulled from IRS Rev. Proc. 2025-32, the SSA and Treasury in-session. Nothing quoted from memory.
04
Test, don't assume
Every tool driven through Playwright case by case. The $150k H-1B case checked by hand to the dollar: $36,209.
built to grow
A project, not a pile of pages
Nine more tools are mapped. Adding one is an entry in tools.json plus a build command — the hub regenerates itself with the new card, category, search terms and status.
| Decision | Why |
|---|
| One registry file drives the hub | Tool twelve costs minutes, not a redesign. Categories, search and roadmap states all derive from it. |
| Self-contained HTML, no backend | Hosts anywhere, opens from disk, and a visitor's bank balances never leave their machine — which is the trust argument, not a technical one. |
| Tax constants in one block per tool | Named sources, refreshed each autumn when the IRS revenue procedure lands. No hunting through code. |
Per-tool ?src= attribution | The original failure was not knowing which spend produced clients. Every CTA carries its tool's tag. |
| Plain HTML, not a JS framework | Five competitor calculators could not be read by an automated fetch. If a scraper can't read them, neither can an AI crawler. |
Status: the suite is built, tested and running. It is pending sign-off from the client's CPA before it goes live on their own domain — tools that tell someone their tax residency need a licensed professional to confirm the rules, not just verified arithmetic. Shown here as a build, not as tax advice.