Season one of It’s Data Habibi covered six conversations that, on paper, have almost nothing to do with each other. A loyalty consultant and a hospitality CDO trading war stories about birthday emails. A data scientist and a founder debating whether agentic AI is ready for the enterprise. A customer data platform co-founder and a retail marketing lead arguing about identity graphs. A Collibra co-founder and a governance lead comparing notes on who actually owns a spreadsheet.
Listened to in order, though, it stops sounding like six topics and starts sounding like one argument, made six different ways. Every episode this season landed on the same conclusion: the thing separating organisations that get value from their data and AI investments from the ones that don’t isn’t the tool, the model, or the campaign. It’s whether anyone can say, with confidence, who owns the data, whether it can be trusted, and what happens when something goes wrong with it.
The season didn’t start there. It started somewhere much more relatable.
The symptom: a birthday email nobody wanted
Episode one opened with the most mundane artefact in loyalty and marketing, the birthday discount email, and used it to make an uncomfortable point. Sending the same 10% off message to ten thousand people isn’t personalisation, our guests argued, it’s a broadcast dressed up as a relationship.
What followed in episode two, on points programs, pushed the same idea further: loyalty schemes fail not because the mechanics are wrong, but because they’re built on redemption data instead of an understanding of why a customer is actually loyal in the first place. Both conversations were, underneath the marketing language, about data quality and structure. A duplicated customer record, an unclean single customer view, a program built before anyone asked what problem it was solving, these were the actual causes of the “lazy loyalty” our guests kept describing.
The same problem, wearing an AI costume
By episode three, the conversation moved to agentic AI, and the vocabulary changed completely, reasoning, planning, autonomous execution, but the underlying diagnosis didn’t. An agent is only as reliable as the documentation and data foundation it’s operating on top of. Our guests were explicit that agentic systems need the same clean, structured, well-governed data that a new human hire would need on day one; skipping that step doesn’t get you to value faster, it just gets you to failure with more confidence.
Episode four made the cost of skipping that step explicit. Multiple guests referenced the same uncomfortable statistic circulating in the market, something in the range of 90-plus percent of AI projects failing to show measurable ROI, and traced it back to the same root cause discussed in the first two episodes: organisations investing in infrastructure and models before they’ve resolved data quality, ownership, or a genuine business case. The FOMO-driven AI spend of the past two years is now being questioned by CFOs asking a very simple question, “what’s the return, and over what period”, and most organisations still can’t answer it convincingly.
Naming the mechanism
Episode five gave the pattern its technical name: identity resolution. Fragmented customer data across web, app, CRM, and in-store touchpoints isn’t a minor inconvenience — it’s the literal, structural form that “we don’t really know our customer” takes inside a business. Our guests were candid that most organisations, including large, sophisticated ones, are working from a baseline of a customer record and a list of orders, and calling it a 360-degree view.
Episode six named the mechanism behind the mechanism: ownership. Every fragmented identity graph, every duplicated loyalty record, every AI pilot that stalled on data quality traces back to the same unresolved question, who is accountable for this data asset, and what happens when something breaks. Our guests made the point plainly: data ownership isn’t a technical problem waiting for an IT team to solve. It’s an organisational and accountability problem, and until it’s assigned formally rather than absorbed informally by whoever happens to be nearest the mess, everything built on top of it, loyalty programs, AI pilots, personalisation engines, inherits the same instability.
The arc, looked at as a whole
Read end to end, season one moves from symptom to mechanism to cause. It opens with a customer-facing frustration anyone can recognise, and by the finale it’s naming the organisational root cause in board-level language: ownership, accountability, and governance aren’t bureaucratic overhead sitting in the way of AI and personalisation ambitions. They’re the precondition for them. Every guest this season, whether they came from loyalty consulting, product leadership, AI startups, customer data platforms, or enterprise governance, arrived at some version of the same sentence: you cannot build reliable value on top of data nobody owns.
That’s also, not coincidentally, the argument this season set up for what comes next. If season one spent six episodes diagnosing why loyalty programs, AI pilots, and personalisation efforts keep stalling on the same structural problem, season two is where the conversation turns to what building the fix actually looks like in practice, inside real organisations, with real constraints, not in theory.
If any of season one’s conversations sounded familiar, a birthday email that missed the mark, an AI pilot that never scaled, a data asset nobody wants to be accountable for, it’s worth sitting with why. The pattern was the same in every episode. The fix starts in the same place too.
Catch all six episodes of It’s Data Habibi on Spotify, YouTube, and LinkedIn, and stay tuned for season two.
