Rob Kleiman

• Written by Rob Kleiman

Quality Over Volume: Rebuilding the Data Behind Developer Marketing

Two problems surfaced in my first months at Capital One, and they had the same root: we were working from a picture of developers that wasn’t accurate.

The challenge

Our annual developer survey, meant to track brand awareness, returned results like zero awareness of Snowflake. That’s a sign the panel wasn’t reaching working developers at all. At the same time, an inherited agency was producing persona documents and strategy materials that were generic and off-audience. Both were on track to hit their targets, and both would have led us to the wrong decisions.

What I did

The survey

  • Identified the gap between the existing panel and our real audience: developers with hands-on experience in the tools we cared about.
  • Picked a sharper competitive set and brought in an agency to curate a more precise panel, deliberately trading volume for relevance.
  • Redesigned the survey around what leadership needed for media decisions, with geographic analysis and benchmarks to test assumptions.

The agency

  • Diagnosed the root problem: the agency didn’t understand the difference between an AI engineer, a data engineer, a software engineer and an analyst, or why it mattered.
  • Stepped in as the subject-matter expert, tightened review cycles and rebuilt the personas from the ground up.
  • Used AI-assisted research (Perplexity, with a source tracked for every claim) and a knowledge base built from Gartner consultations and call notes to move faster without losing rigor.
  • Reset the timeline with stakeholders, framed as an investment in quality rather than a miss, and extended the contract to finish properly.

What changed

The rebuilt tracker produced consistent, credible themes for the first time, with relevancy improving quarter over quarter. Attribution became more accurate, and better targeting cut wasted ad spend. The new personas went to 6 partner teams and 40+ colleagues and informed 30 new SEO-optimized blog posts, and the agency became a long-term partner.

What I took away

A smaller, truer data set beats a big wrong one. It’s worth missing a vanity target to hit the real one.