Twenty years bootstrapping a help desk SaaS from on-prem roots

Executive overview

Most SaaS founders chase hyper-growth or an exit. Ian Landsman has run Help Spot profitably for nearly 20 years without outside funding, reaching ~$2M ARR with five employees.

He launched on-prem in 2005 — before AWS, Laravel, or the SaaS playbook existed — then added a cloud version in 2015 only when the economics made sense. Staying profitable gave him the freedom to explore side projects without betting the company.

The core insight: a profitable, stable business buys optionality that neither raising money nor chasing growth can replace.

Building and launching in 2005

• Quit his job and coded six days a week, 12 hours a day for six months — necessary then, not advisable now
• No Ruby on Rails, no Laravel, no frameworks; wrote every line in PHP while learning JavaScript from a physical book
• Launched to an 84-person email list drawn from Joel on Software forums and his blog
• First month: ~$4,000 in one-time license sales (~27 licenses at ~$150 each), not MRR
• SEO became the primary growth channel within months; ranked first or second for "help desk software" for years
• Month two dropped to ~$1,000, then climbed steadily as SEO compounded

On-prem vs SaaS economics

• On-prem customers pay a one-time license plus ~30% annual support/maintenance fee — effectively recurring, but lumpy
• One-time revenue is front-loaded: great for cash flow, but catastrophic if SEO rankings drop (revenue fell ~80% overnight during the 2008 financial crisis)
• On-prem customers in regulated industries (finance, healthcare) often stay 10–14 years and pay $10–20K/year
• They pay by check or transfer, avoiding the ~5% credit card processing cost that erodes SaaS margins
• On-prem provides a SOC 2 workaround: the customer installs in their own compliant data centre

Why the SaaS launch waited until 2015

• At founding, no money to hire server ops and no AWS — on-prem was the only viable path
• An interim hosting partnership let a third-party run the hosted version, delaying the need for a full SaaS rebuild
• Help Spot's data structures were not designed for multi-tenancy; a clean SaaS rewrite would have been expensive
• AWS nano instances (~$2/month) eventually made a per-customer server model affordable without multi-tenant refactoring
• Each customer now runs on their own small AWS instance sharing a central RDS cluster — the same codebase serves both on-prem and cloud

Migrating on-prem customers to the cloud

• Switched all new customers (on-prem and cloud) to annual subscription pricing
• Existing on-prem customers were not forced to migrate — they can stay indefinitely
• Database migrations take weeks of back-and-forth; some customers have 70 GB databases
• Help Spot still onboards roughly one on-prem-to-cloud migration per month
• On-prem remains ~10% of new business but represents some of the highest-value accounts

Avoiding boredom over two decades

• Profitability funded side projects every three to four years: Laravel conferences (Laracon), a Laravel job board, and two products he later sold
• Side excursions provided re-energisation; returning to Help Spot felt fresh rather than stale
• Revenue was never quite large enough to justify a compelling exit, so the calculus always favoured staying
• Building a strong second product proved harder than expected — the first product's stability kept winning the comparison

The Laravel connection

• Taylor Otwell (Laravel's creator) worked at Ian's company, Userscape, for three years around 2010
• Ian hired him to add enterprise features to Laravel while also working on Help Spot
• Laravel raised $57M in 2024 to fund Laravel Cloud (a Heroku-like platform for PHP) — a move that required VC scale
• Ian runs the official Laravel job board; the community relationship has been a long-term asset

AI risk and the help desk market

• Customer support is a cost centre — companies are more willing to cut headcount there if AI performs adequately
• Klarna's near-full AI support rollout got press, but worked partly because their query types are narrow and uniform
• Help Spot's near-term AI roadmap: custom agent writing prompts (shipped), auto-triage and ticket routing (in progress), assisted auto-response (R&D)
• The open question: will AI consolidate support into a few large platforms, or will access to the same APIs keep smaller players competitive?
• On-prem and self-hosted deployments may gain relevance as enterprises want AI running inside their own infrastructure