Original source: Paul Gibbons - The Great Collisions Keynotes
This video from Paul Gibbons - The Great Collisions Keynotes covered a lot of ground. 3 segments stood out as worth your time. Everything below links directly to the timestamp in the original video.
Most companies spend heavily on data infrastructure while remaining blind to the economics that would make that investment pay off. Schmarzo's framework names the gap precisely.
98% of Companies Stuck in Retrospective Data Use, Schmarzo's Index Finds
The central finding of Bill Schmarzo's Big Data Business Model Maturity Index is that roughly 98% of organisations remain trapped in what he calls the 'business monitoring' phase — using data only to look backward at what has already happened, rather than to predict or prescribe future action. The obstacle, he argues, is not a shortage of technology but a failure to understand the economics of data at a granular level: what a specific customer is likely to buy, which individual product is likely to fail, which single operation can be optimised. Throwing more database infrastructure or dashboard tooling at the problem, as traditional vendors have tried, cannot bridge what Schmarzo calls the 'analytics chasm'.
What this exposes is a structural misalignment between how organisations invest in data and where the actual value resides. Volume, as Schmarzo frames it, is not the monetisable asset — granularity is, a distinction that underpins Amazon's product recommendations and the ad-targeting systems he built at Yahoo.
"You don't monetize the volume of big data — you monetize the granularity."
Schmarzo's Five-Phase Model Puts Cultural Transformation — Not Technology — at the Top
Schmarzo's maturity framework moves through five phases: monitoring, predicting individual-level insights, optimising operations using prescriptive analytics, monetising those insights to identify market 'white spaces', and finally transformation — a state in which data's economic value is institutionally embedded and compensation plans actively reward employees for sharing insights across the organisation. That last condition is, he notes, where most transformation initiatives quietly die: incentive structures that reward individual performance over collective knowledge-sharing systematically undermine the cultural shift Phase Five requires.
The structural issue here is that AI's greatest returns, in Schmarzo's reading, will not flow to organisations that feed executive knowledge into models, but to those that unlock the analytical capacity of frontline workers closest to the customer — a claim that inverts the conventional top-down model of enterprise AI adoption.
"The real knowledge in an organisation is at the front lines, at the point of customer — that is where AI is going to have the big impact."
Chipotle's 7% Sales Target Becomes a Teaching Model for Pre-Scientific Data Thinking
Using Chipotle's publicly stated goal of increasing same-store sales by 7% as a classroom case, Schmarzo walks students through an eight-step process he calls 'thinking like a data scientist' — identifying stakeholders, decisions, success metrics, and candidate data sources before a single data scientist is engaged. The exercise routinely surfaces non-obvious drivers: students frequently converge on local event calendars, such as Little League schedules for stores near sports fields, as stronger predictors of foot traffic than internal sales history. All ideas are formalised into a 'hypothesis development canvas', a design-thinking document that takes three to ten days to complete in client engagements and specifies the cost of false positives and false negatives alongside every other project parameter.
The real argument here is not about Chipotle but about sequencing: the human, analytical work of framing a problem correctly is what Schmarzo believes makes the subsequent data science disproportionately effective — a process advantage, he admits, that amounts to deliberate cheating.
"I think we cheat — we cheat because we do all this work before we ever put science to the data."
Summarised from Paul Gibbons - The Great Collisions Keynotes · 40:39. All credit belongs to the original creators. Streamed.News summarises publicly available video content.
