The venture capital world optimizes for market size. How big is the TAM? Is it a billion? Is it ten billion? Most investors won't seriously engage with a company that isn't, in theory, a very large market. That filter eliminates a lot of bad companies. It also eliminates a lot of excellent ones.
The best-retained, highest-margin software companies in most industries are the ones nobody's written a TechCrunch article about. They serve specific workflows in specific verticals. They have 90%+ gross retention because replacing them is painful. Their customer acquisition cost is low because the founder came from the industry and can pick up the phone and close a deal. This is niche SaaS. It's not a consolation prize.
Why "too niche" is usually the wrong read
A $200M TAM sounds small until you consider what it takes to dominate it. A horizontal SaaS company fighting for 0.5% of a $10B market has to beat 50 competitors on product, pricing, and brand. A niche SaaS company targeting $200M has to be meaningfully better than the alternative for one specific workflow, in one specific type of company.
Those are different businesses. One requires a massive sales team, constant feature competition, and pricing pressure. The other requires deep knowledge of the domain, a founder who can close deals by credibility rather than by demo, and a product that fits a workflow so specifically that switching it out is a serious operational decision.
The unit economics of going narrow
Customer acquisition cost is lower. The founder or early sales team has direct access to the buyer network through an industry they spent years in. Cold outreach gets replaced by warm intros within an industry where everyone knows each other.
Churn is lower. When software is embedded in a specific, non-standard workflow, ripping it out isn't just inconvenient — it requires re-architecting how the team does a core part of their job. Customers stay not because they love the software, but because leaving is too expensive.
Expansion revenue is higher. Vertical domains have adjacent workflows. A company that owns job costing for a specific type of contractor can expand into dispatch, compliance, invoicing — each expansion a sale to an existing customer who already trusts you.
What "niche" actually means
Not niche as in a market too small to matter. Niche as in: one specific job function, in one specific type of company, at one specific scale.
"Construction project managers at commercial general contractors with 50–200 employees" is a niche. There are tens of thousands of companies in that category. Many of them use the same broken workflow. A vertical software company that solves it well could charge $800 a month and own 20% of that market. That number is real. The market is real. It just doesn't make a compelling story for a certain kind of pitch meeting.
The actual failure mode
Most niche SaaS failures don't happen because the market was too small. They happen because the founder tried to go horizontal before owning the niche.
The pressure to expand — more verticals, more use cases, more customer profiles — often comes from investors who want a bigger TAM story, not from customers asking for it. The result is a product that's mediocre for everyone instead of essential for one specific buyer type.
Own the niche first. Build to 60–80% market penetration. Then expand — into an adjacent vertical, or up-market, or to a new buyer type in the same organization. The moat you build in the niche is what makes the expansion credible.
Where the gaps are right now
Industries that still run on spreadsheets, legacy ERPs, and email chains. Workflows too specific for the major platforms to prioritize, too painful for customers to ignore. Places where a founder with 10 years in the industry has a permanent knowledge advantage over any outsider who tries to build the same thing.
That's most industries, if you look for them. The ones we're most interested in right now share a common profile: large workforce, high-frequency operational decisions, and software that hasn't been updated meaningfully in a decade.