The panel
The session was hosted by Rob Verheul, CEO of Graphite Digital, and brought together:
- Ksenia Levina, Global Oncology Customer Engagement Head, Pierre Fabre
- Pavel Bogdashov, CX Omnichannel Lead, Boehringer Ingelheim
- Christian Velten, pharma CX strategist and AI advocate, formerly Roche
- Tom Botting, Co-Founder, Forge DC, formerly AstraZeneca
What the conversation covers
The starting point is a pattern most commercial teams will recognise: dashboards full of sends, opens, clicks, and call counts, and very little that tells leadership whether any of it moved a prescriber. The panel works through why that pattern is so persistent before turning to what breaks it.
Why activity crowds out impact. Executional data is free. Every platform produces it automatically, so reporting fills up with what is abundant rather than what is meaningful. Proving impact, by contrast, takes effort and produces an estimate rather than a certainty, and estimates feel risky to present upwards. Add incentives that reward doing more (campaigns launched, calls made, content produced), data that sits in separate functional silos, and the instinct that holding out a control group means losing sales, and the pattern is self-reinforcing.
Data quality and data trust are different problems. Clean data does not automatically earn belief. The panel discusses why teams can fix one without fixing the other, and what it takes for a number to be trusted enough to act on.
Behavioural signals and qualitative insight. Engagement data shows what customers did. It cannot show what they understood or whether the experience created value. The discussion looks at how to close that gap rather than treating the two as competing sources of truth.
Does AI help us decide better, or reach the wrong conclusion faster? With AI now sitting on top of the same executional data, the panel is candid about the risk of accelerating a flawed measurement model, and about where AI has lowered the barriers that used to make better measurement expensive.
Tracing a budget back to a strategic objective. The thread that runs through the hour is traceability: being able to read any interaction back up the chain to the key result and the objective it was meant to serve, and therefore to say what a given pound of spend returned against the plan.
Four changes that turn activity into impact
Tom’s contribution to the session sets out four changes, two that fix the data and two that fix the evidence:
- Turn strategic information into data. Convert brand strategy, key results, and budgets into metadata that travels with every campaign and asset, so executional data arrives already carrying its strategic context.
- Establish the link between engagement and impact. Matched-cohort analysis shows whether engagement patterns predict behaviour change. Validated once, everyday metrics become leading indicators instead of activity counts.
- Triangulate with user research. Two independent signals, behaviour and perception, give leadership something it can trust. High engagement with low perceived value is a trap, and only research reveals it.
- Pressure-test with marketing mix modelling. Decompose outcomes across your activity and everything else in the market. An honest split (owning two and a half points of growth, channel by channel, rather than claiming all five) is what earns next year’s allocation.
The closing argument is that each of these used to be slow or expensive, and AI has changed that: auto-classification replaces manual tagging, uplift analysis is now commodity tooling, research operations scale with AI, and mix models are open source and run at geography level.
Where Forge DC fits
The first of those four changes is the one we spend most of our time on. Without strategy expressed as structured data, the other three have nothing to connect to. If this session resonates, two pieces on this site go deeper: Calculating Strategic Drift (how to measure whether execution is following the plan) and Content Taxonomy (the classification layer that makes the connection possible).
Watch the recording
The recording is hosted by Graphite Digital and embedded with their permission.
Ready to talk about your own measurement model? Get in touch.