How Clarion Technologies scaled from near-bankruptcy to 400 people

Executive overview

Most offshore software firms treat clients as projects. Clarion built dedicated, long-term teams — and that model, combined with disciplined hiring and process, drove 19 straight years of growth.

The company nearly didn't exist: founder Ankur Agarwal went bust in 2002 with no money for food. Survival work — small coding gigs — became the business. Two decisions made it scale: adopting top grading to hire consistently high performers, and using the E-Myth framework to codify every process before resuming growth.

Retention is a growth strategy: great people don't lose customers, so every new customer you add is net growth.

From bust to business: the origin story

• First venture launched 2000, failed within two years; funding came from friends, family, and the founders themselves
• By 2002, no jobs, no money, near homeless — survival coding gigs were the only option
• Gig pipeline grew organically; by 2007 the company hit $5M revenue
• Growth ran at 50–100% year-on-year until the 2008 financial crisis dropped it to ~10% annual growth for nearly a decade
• Consciously kept no single customer above 20% of revenue to avoid dependency risk

The V employee model

• Clarion provides dedicated, full-time software engineering teams based in India
• Teams include engineers, project managers, scrum masters, architects, database experts, and QA — all under Clarion management
• Clients get continuity and a long-term relationship, not a one-off project handoff
• Most competitors operate on the project model; Clarion estimates 80% of the industry does
• Some clients have been with the firm for over 10 years

Hiring: from gut feel to top grading

• Early approach — tests and interviews — showed no correlation with on-the-job performance
• First fix: Lou Adler's Power Hiring — past performance predicts future performance; method worked but wasn't scalable across hiring managers
• Switch to top grading after a Harnish workshop: same principle, simpler process, easier to train
• All hiring managers now trained in top grading; enables consistent recruitment of top performers
• Result: customer retention stays high; revenue grows because you add customers without losing existing ones

Process: the E-Myth breakthrough

• Fast early growth created a pattern of winning and losing customers based on who happened to be assigned to each account
• Reading The E-Myth (Michael Gerber) on a New York–Mumbai flight in ~2005 reframed the problem
• Paused all lead generation; spent a year writing SOPs, guidelines, and instructions for every part of the business
• Goal: "multiply myself" — let the playbook substitute for founder judgment
• Once the base was documented, the system could evolve; CMMI Level 3 certification followed
• Ongoing process: regular start/stop/continue reviews keep it current

Co-founder dynamics and complementary strengths

• Ankur (ideas, creation, optimism) and wife Swati (analytical, critical, pessimistic) cover opposite ends of the decision-making spectrum
• Different cognitive styles produce a 360-degree view on problems and reduce errors
• Ankur won't invest in startups without a co-founder, and won't invest if co-founders are too similar
• Founder superpower: the ability to envision and manifest — but the risk is never stopping to notice you've arrived

Employee engagement post-COVID

• Work from home created comfort but removed social bonding — especially acute in India where home space and privacy are limited
• Internal initiative "Clarion 2.0": convert offices into social spaces employees want to visit, not work mandates
• Goal: choice-based hybrid — work from home when productive, come to office for connection and fun
• ~20% of workforce already showing signs of depression linked to COVID isolation (industry data cited)

Looking ahead: AI-augmented development

• Three-to-five year goal: 80% of current manual programming effort handled by machines
• Building an AI engine for augmented programming to cut software development costs
• Already applying elements inside the V employee model; significant expansion planned
• Near-term internal problem to solve: automate candidate screening to reduce the number of full top grading interviews required