DeltaMV

Insight Bites · Season 8

AI in Omnichannel – Problem or Solution?

30 July 2025

Pharma has more automation, more data, and more AI-driven personalisation tools than ever to run omnichannel engagement with doctors. So why does actually reaching them often feel harder, not easier? On a recent episode of DeltaMV’s Insight Bites, Neil Doyle argues that AI and automation haven’t solved the marketing funnel – if anything, they’ve made it messier – and that the real fix isn’t more automation. It’s a genuinely human understanding of how doctors and patients actually move through a decision. 

Key takeaways 

  • AI and automation layered onto a linear funnel just adds complexity and fragmentation on top of an outdated map – the real gap is human insight, not another channel or algorithm. 
  • Research from Google and The Behavioural Architects shows people loop chaotically through exploration and evaluation rather than moving in a straight line – the ‘messy middle’1. 
  • Mapping a real DeltaMV vaccination case study into three audience clusters (31%, 12%, 34%) showed exactly where and why the journey was breaking down. 
  • Used to reveal where touchpoints connect rather than to automate more of them in isolation, AI can finally answer the questions a linear model never could. 

AI in Omnichannel – Problem or Solution?

Episode 1 · 30 Jul 2025 · 16:36

Listen onSpotifyApple PodcastsRSS

WATCH ON YOUTUBE

The Problem: More Automation, Same Old Funnel 

The traditional marketing funnel – top, middle, and bottom – assumes people move in a straight line from awareness to consideration to purchase. That assumption was always a simplification, but AI and automation have made it actively misleading. Layering AI-driven personalisation and more digital and non-digital touchpoints onto a linear model doesn’t make the model more accurate; it just adds complexity and fragmentation on top of an outdated map. What’s missing isn’t another channel or another algorithm. It’s what DeltaMV calls Human Insight: a real understanding of what doctors and patients are actually doing, thinking, and feeling as they make decisions in this more fragmented environment. 

Enter the Messy Middle 

Landmark research from Google and The Behavioural Architects gives that messiness a name: the messy middle1. Between a trigger and a final decision, people don’t move in a straight line – they enter a chaotic loop of exploration and evaluation. They research, get distracted, ask someone they trust, compare, drop off, and come back days or weeks later to start again. A doctor deciding whether to adopt a new treatment, or a patient deciding whether to get vaccinated, both move through the same kind of loop, not a tidy funnel. 

Mapping the Loop: A Six-Node Framework 

If the messy middle is the reality, a linear TOF/MOF/BOF model can’t capture it, let alone measure it. DeltaMV’s Neil Doyle proposes an alternative: a six-node framework, sometimes called the Hankins Hexagon, that maps the stages people actually loop between – passive assimilation (background exposure before any active intent), a trigger that sparks interest, active evaluation, outlet selection, brand comparison, and purchase and beyond. Crucially, people move between these nodes forward, backward, and sideways, not in sequence, and mapping those loops is what makes the mess visible and measurable instead of just frustrating. 

Where AI Actually Helps: From Isolated Metrics to Systems Thinking 

This is where AI stops being part of the problem and starts being part of the solution – but only if it’s used differently. Most teams still measure isolated metrics at each funnel stage – awareness scores, click-through rates, conversion rates – each in its own silo. That tells you how many people saw something, not how different touchpoints work together to influence behaviour. The industry’s own analysis of omnichannel performance makes the same point: in a fragmented media landscape, results come from orchestrating a cohesive, responsive experience across channels, not from buying the most impressions in any single one3. Used this way – to reveal how touchpoints connect rather than to automate more of them in isolation – AI and data analysis can answer the questions a linear model never could: where exactly are people dropping off, what happened just before they did, where are they stuck looping, and is every touchpoint pulling its weight? 

Case Study: Mapping the Messy Middle in Adult Vaccination 

DeltaMV applied this approach to a real, patient-driven journey: adult pneumococcal vaccination. 

  • Context: using an AI-driven method that combines qualitative exploration with quantitative coding and multivariate analysis, the team set out to answer a simple question. Roughly 80% of adults have heard of pneumococcal disease, yet only a small fraction get vaccinated – so where, exactly, is that journey breaking down? 
  • Finding: the answer was three distinct clusters, each representing a different marketing problem. Around 31% of the audience is stuck at passive assimilation – aware the vaccine exists but never receiving a trigger strong enough to move them into active consideration. Another 12% is stuck at the trigger stage: broad exposure from social media, TV, family and friends, but almost no contact with an HCP, so the trigger never converts into evaluation. The largest group, 34%, loops between trigger and evaluation – actively researching, but never resolving their uncertainty enough to decide. 
  • Barriers: looking closely at the 12% stuck at trigger showed why a channel-agnostic view of ‘reach’ is misleading. Friends and family have the greatest reach at that stage but are not particularly effective at moving people forward; treating physicians and preventative health doctors are far more effective, but this group has almost no contact with them. Three specific barriers kept triggers from converting: anxiety about side effects, financial and value concerns, and a lack of social proof. 
  • Result: that level of detail made a targeted, emotionally resonant response possible. Evidence suggests well-designed behavioural nudges can move purchase decisions as much as a meaningful discount1. One lever DeltaMV tested for the value-concern barrier was loss aversion – the well-established finding that a loss looms psychologically larger than an equivalent gain2. Rather than pitching the cost of vaccination, the message reframed inaction as the expensive choice: anchoring the cost of a single hospital stay against the cost of vaccinating an entire family turns waiting into the risk, not the safe option. 

What This Means for Pharma Marketers 

So is AI in omnichannel a problem or a solution? Both, depending on how it’s used. Pointed at automating more touchpoints on top of a linear model, AI simply adds noise, complexity, and fragmentation faster than before. Pointed at revealing where people actually get stuck in the messy middle, and why, it becomes the tool that makes a systems view of the journey possible in the first place. The brands that win won’t be the ones with the most automation. They’ll be the ones with the deepest understanding of where real customers drop off, what happened along the way, and how to address the true barrier at each stage in a way that’s emotionally, not just informationally, compelling. 

Frequently Asked Questions 

What is the ‘messy middle’? 

A concept from Google and The Behavioural Architects describing how people actually move between a trigger and a final decision: not in a straight line, but in a chaotic loop of exploration and evaluation1. 

What is the Hankins Hexagon? 

A six-node framework – passive assimilation, trigger, active evaluation, outlet selection, brand comparison, and purchase and beyond – that maps the stages people actually loop between, instead of assuming they move through a tidy funnel. 

Watch the Full Episode 

This explainer only covers the highlights. For the full conversation with DeltaMV’s Neil Doyle – including the complete walkthrough of the Hankins Hexagon and the adult vaccination case study – watch the full episode of DeltaMV’s Insight Bites. 

Ready to see how the messy middle applies to your own funnel? Watch the full episode on YouTube. 

Sources 

  1. Google, & The Behavioural Architects. (2020). Decoding decisions: Making sense of the messy middle. Think with Google. https://www.thinkwithgoogle.com/_qs/documents/9998/Decoding_Decisions_The_Messy_Middle_of_Purchase_Behavior.pdf 
  1. Kahneman, D., & Tversky, A. (1979). Prospect theory: an analysis of decision under risk. Econometrica, 47(2), 263–291. https://doi.org/10.2307/1914185 
  1. Picazo, S. (2024). The alchemy of omnichannel: how connecting channels can boost advertising outcomes. WARC. https://www.warc.com/content/paywall/article/Warc-Research/The_alchemy_of_omnichannel_How_connecting_channels_can_boost_advertising_outcomes/en-GB/158140 

Other Podcasts

What are you trying to change?

Let's explore the behavioural insight behind it.

Connect with us