You have the data. Do you have the insight?

The panel's opening exchange made something clear quickly: the data gap most pharma organisations think they have isn't really a data gap. 

Ksenia framed it well when she described confident decision-making failing at three distinct levels.

"The first one is the data itself — is it sufficiently connected, reliable and complete? The second one is the interpretation, so we may have marketing, medical, analytics and sales looking at the same signal and coming to different conclusions. And the third one, which I think is the most underutilised, is the decision process. We may generate an interesting insight, but without clarity on what decision should it impact, who owns that decision and when it needs to be made, we may fail to put evidence into action. Because data informs, but people decide in the end."

Christian took a similar view, pushing back on the tendency to blame silos and culture. "I am a little bit hesitant personally of claiming silos or culture, because it's a little bit too easy sometimes — I've seen it being misused as a bad excuse for laziness." 

His more pressing concern was simpler: knowing what question you want the data to answer in the first place. Too often, teams don't. The result is data pushed onto people's desks that overwhelms rather than informs.

Tom's presentation mid-session put some structure around what the panel had been circling. The core argument: there's a difference between measuring whether your strategy worked and measuring whether you actually delivered it. For most organisations, the connection between the two is missing entirely.

"Very few organisations actually look at whether their activity was what they set out in the brand plan. Rather, they just look at frequency and coverage."

The practical implication is significant. If you can't trace a pound of budget through your brand strategy to a specific customer, channel and message, then the data you're generating is activity tracking, not measurement. And if AI or next-best-action models are then making decisions on top of that, the problem compounds.

Pavel pushed this further when the conversation turned to what organisations are actually measuring in omnichannel. His point: it's easy to gravitate toward what's measurable rather than what matters.

"It's not just about delivering three messages to this GP and job done, that's it. Creating behavioural change involves building trust, understanding the customer and building the relationship to allow them to be open to a change — not just in behaviour, but the underlying fundamental attitudes and beliefs."

One thread that ran through the whole session was the gap between quantitative signals and genuine customer understanding. Engagement data tells you what people did. It rarely tells you why, and without that, it's hard to know whether high engagement reflects real value or just novelty, noise, or the wrong audience clicking.

Christian referenced a Jeff Bezos principle that captures it cleanly: "If your quantitative data and a customer anecdote disagree, most likely your quantitative data are wrong."

Tom framed user research not as a bolt-on, but as an essential cross-check. When behavioural data and research tell the same story, you have something you can actually trust. When they diverge, you have a question worth asking.

The final question — whether AI will help organisations make better decisions or simply reach the wrong conclusions faster — produced the most honest answers of the session.

Ksenia: "I think it can do both. AI can help us to make better decisions, but it can also create false confidence. It should not at any cost remove human accountability for the decision-making process."

Pavel described AI as "a really powerful accelerator[...] it can accelerate you down the right track or off the cliff. Where you point it is the difference."

Christian kept it simple: "Crap in, crap out. If you ask a stupid question, you will get a stupid answer. That's not the fault of the AI, that's your fault."

The consensus was consistent: AI is only as useful as the quality and structure of the data feeding it, and the clarity of the question being asked. Both are still human jobs.

The session closed with a question to each panellist: what's one practical change a digital or omnichannel team could make in the next three months to become more evidence-led?

The answers pointed in the same direction. Start with the business question, not the data. Get clear on what behavioural change you're trying to drive and for whom. Build cross-functional alignment around that objective before you worry about which platform or dashboard to use.

The recording, Tom's slides, and ForgeDC's latest white paper on calculating strategic drift are all available below.

• Webinar recording: Watch here
• Tom's slides and practical frameworks: Download here
• ForgeDC's latest white paper (published this week): Read here