How a million-dollar agency founder quit consulting to build a SaaS

Executive overview

Running a successful agency creates a cash addiction: every hour billed returns immediate revenue, making it nearly impossible to prioritise a side SaaS earning a fraction of that. Keith Parahak built SegMetrics — a marketing attribution tool that unifies lead data across ad platforms, ESPs, and payment gateways — while running a seven-figure conversion optimisation agency.

After three years of neglect, he wound down the agency, let go of his team, and went all-in. Growth only followed when focus did.

The agency's time-for-money model is incompatible with building a product — and that conflict extends to your entire team.

What SegMetrics does

• Pulls data from Google Ads, Facebook Ads, ESPs, CRMs, and payment gateways into a single customer journey view
• Assigns individual lead values rather than aggregated channel-level metrics
• Lets marketers slice by demographic or behavioural segment to find who is most valuable
• Solves the "garbage in, garbage out" problem of tools like Mixpanel that require manual data wiring
• Built to replace hours of weekly spreadsheet exports and pivot tables

The build: from CSV to product in two weeks

• First version was a raw data dump from client files into a PHP script — no UI, no signup
• Validated with two or three agency clients before building anything visual
• Set a two-week sprint during a quiet agency period; shipped a working product with UI, graphs, and signup
• Launched on Product Hunt, then discovered a space in the Stripe public key meant no one could purchase
• Hit $1K MRR within a month of fixing the launch

Why growth stalled for over a year

• Agency contracts started at $10K/month; the SaaS was making $1K total
• Impossible to justify pulling the team off billable work for product development
• Remained stuck near $1K MRR for 15 months — until mid-2018

Transitioning from agency to SaaS

• Decided in summer 2018 to wind down client work over six months and focus on SegMetrics
• Attempted to bring the agency team with him, but the mental model — time equals money — didn't transfer
• Team couldn't prioritise SaaS tasks with zero immediate financial return
• Roles also didn't map: an account manager for under 100 customers had no clear function
• In January 2019, let the whole team go; helped each person find new work with former clients
• Within two years, brought them back as SegMetrics grew — into roles suited to their strengths

Going all-in: the psychology of solo focus

• Once alone, all external excuses for low productivity disappeared
• Had to confront personal patterns: defaulting to coding new features instead of marketing or launching
• Hired a project manager at the start of 2020 to enforce quarterly and monthly goals
• She flagged drift from priorities and provided end-of-period evidence of what actually got done
• Completing 80% of stated key tasks per month felt invisible without someone tracking it

The database crisis that cost six weeks

• A MySQL 5.7 bug caused the database to fill completely within an hour
• Manual migration to a new database: one week of work, then the same failure recurred two weeks later
• Moved to a managed DigitalOcean database; a settings mismatch caused a third round of failures
• Total productivity loss: four to six weeks
• Root cause: chose the self-hosted option to save roughly $100–$300/month
• Lesson: for infrastructure that is core to the product, the cost of failure far exceeds the cost of managed hosting

Pricing changes that accelerated growth

• Original model: large contact-based tiers with sudden step-up increases
• Customers tried to stay under thresholds; upgrades were manual and friction-heavy
• New model: price increases in $5 increments as contact count grows — no sudden doublings
• Added a feature tier for enterprise integrations (e.g. HubSpot), reflecting higher support cost
• Result: one of two inflection points on the revenue graph — the other being joining Tiny Seed