A few months ago, I was asked to support the recruitment of System Architects for our programme. What started as standard hiring quickly became unsustainable.
The reality we faced:
- Reviewing 20+ CVs every week, with 80% rejected at first screening
- Only 25% of interviewed candidates were successful
- Spending 3–4 hours in interviews each week at peak
- Even successful hires were struggling because of role misalignment
After two months of this grind, we paused. Something wasn't working.
What we discovered
Our job description was too vague, attracting a flood of mismatched applications. We had no consistent assessment framework. Worst of all, candidates who passed our process were finding the actual role different from what they'd interviewed for.
The problem wasn't the candidates. It was us.
The fix
We went back to basics — four changes, each one simple on its own, but transformative together.
1. Rewrote the job description — not based on what sounded impressive, but on what people actually did day-to-day. We sat with leads and rebuilt it from scratch.
2. Built an interview matrix — every candidate evaluated on the same dimensions, consistently. This turned subjective gut-feel into comparable data points.
3. Aligned with the agency — we stopped accepting a firehose of CVs and gave our recruitment partners precise, specific guidance on what "good" looked like.
4. Created a Capability Guild — so that when new hires joined, they had a community, a support structure, and a clear picture of what success looked like.
The results
- CV volume dropped from 20+ to ~4 per week — but nearly every CV progressed
- Interview success rate jumped from 25% to over 50%
- Before: ~100 CVs and ~20 interviews to hire fewer than 5 people
- After: 5 interviews to successfully hire 3 candidates
The lesson
Sometimes the answer isn't working harder — it's working smarter. Clear role definitions, consistent evaluation, and proper onboarding aren't just nice-to-haves. They're the foundation.
The same principle applies to architecture decisions, by the way. If requirements aren't fit for purpose, don't accept them and hope for the best. Fix the problem at source, not where it surfaces.