Founders ask me almost weekly: "How do you find the niche?" After running 20+ ventures across 14 years — some won, some lost, some sit in profitable boring middle ground — I've found the pattern is repeatable. Niche discovery isn't luck. It's three techniques that compound, and avoiding five seductive dead-ends. Here's what I'd tell a founder version of myself starting today.
The 20-venture map (one line each)
Before the techniques, here's the inventory — every venture and the niche thesis behind it.
- NextBangla Ltd (2012) — mother co, multi-disciplinary digital services for BD market
- NB IT Firm — niche: BD enterprise SaaS implementation
- Promo Content Studio — niche: visual content for BD-based product teams
- Promo Video Studio — niche: high-end corporate video at BD pricing
- NB IT Institute — niche: applied IT training Nilphamari → world
- KaruBazaar — niche: BD artisan → global buyer marketplace
- HelloSeba — niche: hyperlocal service marketplace, Dhaka-first
- Nextech — niche: tech retail with BD-specific warranty + support
- Pawnaakhata — niche: SMB invoicing for BD micro-businesses
- ORBIX — niche: AI-first Business OS for solo founders running 2–10 ventures
- Voxly — niche: AI message coach with cultural-tone translation
- NOBBYO — niche: hyperlocal everything-app for BD shops + providers
- Pannakhata — niche: AI bookkeeping for BD SMBs with VAT compliance
- Kothalipi (now Voxly) — niche: Bangla voice-to-text foundation
- VoiceBridge — niche: AI voice translation for cross-border meetings
- GiftNao (coming) — niche: gifting platform for BD diaspora to local recipients
- ShadGhor (coming) — niche: housing micro-vertical
- Pawport + Mewmela — niche: pet wellness + cat-marketplace (BD)
14 in market, 2 coming. Three are clearly winning. Four are stable. The rest are in the build, pivot, or kill columns at any given time. That's normal.
Technique 1: Adjacency to existing operations
The single highest hit rate for niche discovery is building next to what's already running. You already have customers, vendors, and operational learnings — pick a niche where 80% of the playbook ports over.
Example: Promo Content Studio came out of NextBangla. We already did content for our own ventures — turning that into a productized service was a 60-day move, not a 6-month one. Same again for Promo Video Studio.
Counter-example: Don't pick adjacencies that look adjacent but aren't. "We run a marketplace, so let's build a fintech" — those share zero ops. Test: can you ship V1 with the same team in < 90 days? If not, it's not actually adjacent.
Technique 2: Customer wedge from a prior venture
The second-highest hit rate: your existing customers asking for the next thing. Pannakhata came from Pawnaakhata customers asking "do you also do VAT?" Voxly came from ORBIX users asking for better voice notes. NOBBYO came from HelloSeba's provider side wanting their own marketplace.
The signal is when 3+ unrelated customers ask the same question. That's market pull. Build it.
Counter-example: "We surveyed 100 people and 78 said they'd use it." Surveys lie. Three paying customers volunteering the same wish is 100× more reliable than 1000 survey respondents.
Technique 3: "I'd pay for this" — personal pain
Some of my best niches came from problems I had myself. ORBIX exists because I was running 10 ventures and using 18 different tools to track them. Voxly exists because I dictate 30+ messages a day and "tone right" is half the work.
Rule of thumb: if you'd pay $50/mo today for the solution you're building, you've passed the personal-pain test. Then check: do at least 100 other people fit your same profile?
Five niches I tried and abandoned (and why)
- BD-market crypto wallet (2018). Killed by regulation, not market. Lesson: even great markets fail if the regulatory wind isn't behind you.
- Influencer marketing SaaS for BD brands (2021). Market was real but too few brands willing to pay SaaS pricing. Lesson: BD market wants services, not subscriptions, in many categories.
- Bangla typing software (2014). Solved a real problem but Google Indic + Avro made it free. Lesson: don't build where infrastructure players will absorb you.
- Co-working space brand (2019). Capex-heavy, low margin, geographic constraint. Lesson: avoid niches where capex > revenue for 2+ years.
- White-label CMS for BD media houses (2017). Sales cycle was 9 months per customer. Lesson: enterprise BD sales cycles will kill your runway unless you have $1M+ before starting.
The "founder gravity" test
The most reliable predictor of which niche I should pursue: founder gravity. Can I credibly imagine spending 36 months on this — including the boring middle 18 months — without becoming bitter?
If yes, the niche has founder gravity for me. If no, it's a hobby project pretending to be a startup. The 5 niches I abandoned above all failed the gravity test by month 6 even though the markets were real.
Test prompts:
- Would I still pursue this if I had $5M in the bank and could pick anything?
- Will I still be reading about this market 5 years from now?
- Are the people in this market people I'd grab coffee with?
Three yeses = strong gravity. Less than two = walk away.
When to kill a venture (the hardest skill)
Most founders kill too late. The signals are quantitative — not vibes:
- Revenue not 2× in 12 months. Pre-PMF is OK at flat. Post-PMF and stagnant = kill or sell.
- The team has lost belief. You'll feel it in standup energy 60 days before you admit it on the spreadsheet.
- You stop dogfooding. If the founder doesn't use the product weekly, no one outside will pay for it.
- Costs growing faster than revenue for 2+ quarters. Burn-multiple > 3 is the silent killer.
Killing a venture is the highest-leverage decision a multi-venture founder makes. The freed-up time and team funds the next one.