Marketing & commercial analytics
I make marketing measurable.
Most companies cannot tell which marketing actually worked, because the trail breaks somewhere between the ad platform, the campaign tools, and the CRM. I build the connective tissue that keeps it intact - so you can follow a person from first click to outcome and know what earned the result.
The taxonomy is the attribution metadata carried with every touch - tagged on the UTM, preserved through the session, form, and CRM. It captures the strategy behind each campaign, not just the platform, and structures reporting.
How it breaks, what changes when it holds, and how the signal survives the stack - below.
Why it breaks in the handoffs
It breaks when no one owns the data end to end. A click, a form fill, and a patient record each live in a different system, and the attribution metadata that ties them together, the campaign, the source, the identifier, has to be captured, recorded, and preserved at every handoff between them. When nobody is responsible for keeping that signal intact across the whole ecosystem, it degrades one handoff at a time until the trail no longer connects. A better attribution model cannot recover a signal that was never preserved. Someone owning it end to end can.
What changes when the thread holds
- One person, one thread, from first click through lead flow to patient.
- Cost per acquisition you can measure. In one global rare-disease pharma organization, this is what made it computable end to end - not a better model on the same broken data, the thread finally held all the way through.
- Campaign results that reconcile with what the CRM and the data platform say, so nobody has to pick a version.
- Spend decisions made on evidence rather than on whoever reports fastest.
- Reporting that brand teams in different regions read the same way.
Where it usually goes wrong
- The taxonomy is designed by one team and adopted by none, so it survives in the ad platform and dies in the CRM.
- An identifier is dropped at a handoff nobody owns, usually a form or a redirect, and the join fails silently for months before anyone notices the numbers stopped adding up.
- Attribution gets bolted on at the end, when the only place it can work is at the point the signal is defined.
None of those are modeling problems. They are design and coordination problems, which is why this work is cross-functional before it is technical.
How the signal survives the stack
- Partner with business and marketing to understand the strategy. Not the report they asked for, the outcome they are trying to reach.
- Translate that strategy into an execution plan on the platforms, and define exactly which signals have to be surfaced to answer it.
- Create and execute a taxonomy strategy, so those signals mean the same thing in every system that touches them.
- Work with platform owners and developers so the signals and the taxonomy are not just surfaced but preserved on a user through the entire stack. This is the step that usually gets skipped, and it is the one that decides whether attribution works at all.
- The result is proper attribution through lead flow to patient.