← Back to Field Notes
BCG published a piece late last year on how GenAI is reshaping supply chain management — How GenAI Reimagines Supply Chain Management — and reading it felt less like learning something new and more like seeing a name put to something we’ve been building toward.
The core argument: AI in supply chains has underdelivered for years because organisations keep force-fitting new technology into old processes. The real unlock isn’t better analytics bolted onto broken workflows. It’s rethinking the workflow entirely — from human-operated to human-designed systems where agents handle orchestration, routing, and real-time rebalancing.
That framing hit home because we’d just finished a discovery engagement that was, underneath all the marketing language, a supply chain problem.
A national operator in the medical industry. 400+ clinics across 20+ sub-brands. Each brand running its own Google Ads, its own SEO, its own social campaigns. Completely independently.
The result: their own brands were bidding against each other. In some suburbs, three or four of their own brands appeared in the same search results, competing for the same patient. Every dollar spent outbidding themselves was a dollar that could have gone toward acquiring someone genuinely new.
When we mapped it, we found 125 clinics — 40% of the entire network — sitting in cannibalisation zones. 48 postcodes where multiple brands from the same parent company overlapped. The estimated waste: $2.49 million a year in self-bidding alone.
This is a routing problem. Not a marketing problem. And the moment you see it that way, the solution changes completely.
BCG identifies four levels of AI maturity in supply chains: point solutions, process enhancements, deep process transformation, and cross-functional automation. What’s striking is how precisely this maps to what we see in marketing operations at scale.
Level 1 — Point solutions. Most multi-brand operators are here. Each brand has its own dashboards, its own agency, its own campaign manager. AI shows up as a chatbot or a bidding algorithm inside a single platform. It optimises locally but has no visibility into the broader system. Each brand’s Google Ads “AI” is happily optimising bids — including bids against sister brands it doesn’t know exist.
Level 2 — Process enhancements. You start connecting data across brands. You can see the overlap. You can generate alerts when cannibalisation is happening. But the response is still manual — someone has to decide what to do and coordinate across brand teams who may not even report to the same person.
Level 3 — Deep transformation. This is where the cannibalisation project landed. Instead of optimising brand-level campaigns, you rethink the unit of operation entirely. We shifted from brand-level marketing to catchment-level marketing — one consolidated presence per suburb that intelligently routes patients to the right clinic based on availability, proximity, and quality ratings. The brand becomes a routing variable, not the organising principle.
Level 4 — Autonomous orchestration. BCG describes a vision where specialised agents — demand agents, inventory agents, supply agents — collaborate continuously to rebalance a supply chain in real time. The marketing equivalent: an agent that monitors search demand by postcode, allocates budget across the network based on capacity and conversion probability, adjusts creative by catchment demographics, and routes leads to whichever clinic can actually serve them. No human scheduling campaigns. Humans designing the system and reviewing the outcomes.
We built a prototype of this — the “Agent Autopilot” landing page — where a scoring algorithm ranks clinics by vacancy (40%), quality rating (30%), historical conversion (20%), and proximity (10%), with a full decision log showing why. It’s not theoretical. It works. The question is organisational readiness, not technical feasibility.
BCG flags the same execution challenges we hit on every engagement: disconnected data, poor data quality, processes designed around legacy systems, and people who don’t trust the AI because they can’t see how it reaches its conclusions.
Their prescription sounds familiar too: don’t try to integrate AI into every aspect of operations at once. Map where the most important decisions are made. Prioritise the highest-return areas. Rethink the full workflow. Start building with the right ecosystem.
That’s almost word-for-word how we structured the cannibalisation engagement:
The deeper insight from BCG that applies well beyond supply chains: the talent shortage is real, and it changes the calculus. They note that without wide adoption of AI-assisted workflows, companies will struggle to run fundamental business processes as skilled professionals become scarcer.
We see this in marketing operations constantly. The multi-brand medical operator doesn’t have 20 skilled digital marketers to run 20 brands intelligently. They have maybe three or four good people spread across the whole network. The rest is being done by junior staff following playbooks, or agencies with no cross-brand visibility, or nobody at all.
AI doesn’t solve this by replacing the three good people. It solves it by making the system intelligent enough that you don’t need a skilled human making every routing decision in every suburb. The good people design the system. The system handles the volume.
That’s BCG’s “human-operated to human-designed” shift. And it’s exactly what we mean when we say “don’t replace your team — make them unreasonably effective.”
The pattern we keep seeing: the problems that look like marketing problems, data problems, or operations problems are often logistics problems in disguise. Routing. Allocation. Balancing supply and demand across a network. Minimising waste in a system with competing internal signals.
If you run a multi-brand, multi-location business and your brands are competing with each other in digital channels, you don’t have a marketing problem. You have a supply chain problem. And the tools to solve it aren’t in your ad platform’s AI features. They’re in rethinking what unit you operate at.
Catchments, not brands. Systems, not campaigns.