McKinsey’s latest on MENA grocery retail contains a striking contradiction. Consumer confidence is up. New store openings are up. Revenue growth has slowed.
More activity, less result.
This is the growth paradox — and it is not a data problem.
The Pattern Behind the Numbers
What the report describes is an organization that has the infrastructure and the market access but is not translating either into faster, more confident decisions. More stores, more inventory, more footfall — and the same deliberation cycle as two years ago.
The companies in this report are not data-poor. They are decision-poor. They have the data. They have the teams. What they do not have is a system that produces a decision by Thursday so the action happens on Friday.
The symptom is everywhere once you know how to look for it. A retailer opens three new locations and the weekly leadership review still can’t clear a marketing campaign decision without a follow-up meeting. A procurement team has real-time supplier pricing data and still runs RFQs the same way they did before the platform was installed. More activity, amplified by the same decision bottlenecks.
Why More Infrastructure Makes It Worse
Here is what most analytics transformations get wrong: they invest in the data layer — the dashboards, the warehouse, the AI models — without ever redesigning the decision layer underneath it.
The decision layer is the set of questions nobody writes down but everybody lives by: Who owns this decision? By when does it need to be made? What information does it actually require? What does a wrong decision cost versus a delayed one?
When those questions are never asked deliberately, the operating defaults take over. The Thursday meeting that keeps slipping. The CFO who needs three data sources to feel confident. The approval chain that grows with company size instead of shrinking around urgency.
More data infrastructure amplifies the problem rather than solving it. You end up with faster access to more information flowing into the same unchanged decision process — which means either the same slow decisions or more contested ones.
This is the operating debt nobody talks about. It hides inside analytics transformation budgets because it is not a tooling line item.
What the Companies Getting This Right Are Doing Differently
The organizations pulling ahead are not necessarily the ones with the most sophisticated dashboards or the most AI deployment. They are the ones who redesigned their decision layer so the business moves faster than the technology feeding it.
Some patterns we see in practice:
Decision mapping before dashboard building. Before a single metric gets defined, the recurring decisions get named — the ones that happen on a calendar, with a named owner, with real consequences for getting wrong. The reporting gets built around those decisions, not around the data sources.
Velocity measurement, not just output measurement. Measuring how many dashboards were shipped tells you about the analytics team’s workload. Measuring how long it takes from a decision being surfaced to it being made and acted on tells you whether the organization is actually moving faster.
Designed handoffs. Every decision has a last mile. The meeting that produces the decision, the person who communicates it, the process that turns it into action. That last mile is where most decision velocity gets lost, and it is almost never audited.
The AI Infrastructure Complication
There is a layer on top of this that is making the problem more urgent. GCC organizations are investing heavily in AI compute infrastructure — data centers, GPU clusters, enterprise AI platforms. BCG’s latest confirms it: the organizations that invested early in AI maturity are already generating measurably higher revenue.
That is the good news. The complication is this: AI infrastructure compounds its value when decisions happen fast enough to act on it. If your AI model produces a better demand signal on Monday and your decision process doesn’t route that signal to action until the following Monday’s review meeting, you have paid for intelligence and not used it.
The companies struggling to show ROI on AI investments are not lacking compute. They are lacking decision speed.
The Operating Model Fix
This is solvable. It is not a technology problem — it is an operating model problem, which is a different thing entirely.
The teams that get this right treat their decision infrastructure the same way they treat their financial infrastructure: with deliberate design, named ownership, and regular review of what is working and what is not.
That is the work we built AUXO’s CODE model around. Center the decision before you organize the data. Deliver what enables the decision, not what impresses in the review meeting. Embed the capability so the team runs it without us.
The MENA growth story is real. The question is whether your organization is positioned to capture its share — and that depends less on which AI platform you chose and more on whether your decision processes can move at the speed the market is creating.
More activity without fixing the decision layer does not produce more growth. It produces more expensive bottlenecks.
The teams that win the next three years will not be the ones with the best AI infrastructure. They will be the ones whose operating model can keep pace with it.
Ready to work backward from the decisions your organization actually needs to make?