← Back to work

Jun 2023 — Jun 2025 · Zürich

Goldbach Media — DMP product ownership and cost reduction

Product Owner, Farcaster DMP · Goldbach Media

Owned a Data Management Platform at Goldbach Media, lowered operational costs, and reduced time-to-audience from days to near real-time.

Context

Farcaster was Goldbach’s Data Management Platform: it gathered behavioral data from Swiss publisher websites, segmented users, and fed audience segments to ad servers for targeted advertising. I joined as Product Owner in June 2023 with a distributed team of eight developers, a data scientist, and a Scrum Master — all based in Belgrade, with me and the stakeholders in Zürich.

Problem

Three things needed fixing in parallel.

  • Cost. AWS operational costs were high; the Head of Data wanted them reduced.
  • Product clarity. Farcaster was well-known internally by name but poorly understood — unclear stakeholders, unclear priorities, weak visibility.
  • Team alignment. No structured roadmap coming down, no clear channel for team input going up, no feedback loop tying work to business outcomes.

What I did

  • Ran structured architecture-review workshops with the development team to identify cost optimization opportunities — a mix of infrastructure changes and functional reductions.
  • Designed a probabilistic segmentation model to complement the overnight batch pipeline: traded some precision for near real-time placement of unknown users into likely segments.
  • Conducted stakeholder discovery interviews across the organization, ran company-wide demos, mapped priorities, and rebuilt product visibility from scratch.
  • Implemented a proper roadmap, well-defined user stories, and feedback loops that connected the team’s work to measurable business impact.
  • When company strategy shifted from build to buy (after a major customer departure), led an objective vendor evaluation that resulted in migrating to Permutive — knowing the migration would eliminate my team’s role.

Outcome

  • Operational costs lowered through a mix of architectural changes and functional scope reductions.
  • Time-to-audience reduced from 2+ days to near real-time, expanding addressable inventory for ad products.
  • Product visibility rebuilt; Farcaster repositioned from “known but misunderstood” to “strategically aligned.”
  • Permutive migration delivered cleanly.

Reflection

The hardest single thing I did at Goldbach wasn’t the cost reduction or the segmentation redesign — it was maintaining objectivity during the Permutive evaluation when the outcome was going to end my team. The technical decisions were easy compared to that.

Also: media and adtech were a useful but finite experience for me. I work better in product domains where the customer is the end user, not the advertiser.