TL;DR: Global CPG IT fails when it designs for fairness instead of complexity. It should prioritize and build for the largest and hardest market first, then scale outward. Should, but historically hasn't.
The Fairness Delusion
Global CPG IT runs on equity. Grotesquely too much. It builds an operating model around giving too many markets a fair shake and ends up building platforms optimized for the average, which in this industry means optimized for the void between markets. A sort of digital Switzerland that takes no position, offends no one, and moves little that actually matters.
One market, almost always the US, generates the largest revenue share, carries retail complexity that would make a logistics professor weep, and disproportionately funds the global infrastructure that everyone else benefits from without having to justify. The P&L knows this. Leadership knows this. Somehow IT rarely gets the memo.
Centralization was supposed to fix this. One platform, one roadmap, one team to rule them all. What it produced was a global committee that treated the mechanics of a Walmart trade model with the same grave deliberation it applied to a vending route in Eastern Europe. This is the organizational equivalent of a hospital triaging a paper cut and a compound fracture with identical forms and identical wait times in the name of process fairness. The forms get filled. The patient bleeds out.
The federated experiment swung the other direction and handed every market autonomy, which sounds sensible until you discover that dozens of local solutions means nobody's data agrees with anybody else's and answering a basic business question requires a three-week reconciliation exercise that produces a number nobody fully trusts and everyone quietly adjusts in a spreadsheet before putting it in front of senior leadership.
Both approaches had the same disease. They were organized around structure rather than results, around the kumbaya politics of all voices getting heard rather than the reality of what the business actually needs.
What Authority Actually Looks Like
The fix isn't complicated, which is probably why so many organizations manage to avoid it. Call it the U.S.-anchored Federated Product Structure model where the mechanics of it matter as much as the label.
Here’s how it works. You anchor the model to Global IT who owns platforms and standards outright, non-negotiable and not subject to local reinvention by a regional IT director who wants to prove their market is special. Dedicated product owners with real budget accountability drive execution and roadmap priority. They sit embedded in the US market with an eye toward satisfying the largest internal client where the pressure is highest. They have actual decision rights: vendor selection, scoping calls, global resource allocation, and the standing to say no to other markets asking for a custom integration that breaks the enterprise. The most complex, highest revenue market sets the tempo. Every other market adopts, adapts, and collects capabilities it never could've justified building on its own. The sequence gets locked: prove it and build it where the friction is highest, then extend it outward. That's it.
What that structure kills is the polite fiction that all stakeholder's requirements deserve equal weight. They don't, and pretending they do is how you end up building systems that are mediocre everywhere and excellent nowhere. A single market doing 35 percent of global revenue while simultaneously navigating Walmart, Costco, Target, Amazon Vendor Central, D2C, and a rotating cast of regional grocery chains is not solving the same problem as multiple disparate markets doing 5-10 percent each through a handful of national wholesalers. Treating them as equivalent inputs isn't fairness. It's a failure of analysis dressed up in the language of inclusion.
Ghosts In The Spreadsheet
Nobody publishes case studies about where this regularly goes wrong, which is itself worth sitting with. The silence is the tell. When a global CPG's IT program fails its largest market, it doesn't show up in a press release or a Gartner report. It shows up in a North America commercial team running parallel spreadsheet workarounds alongside the enterprise system because the enterprise system chokes on promotional pricing and deductions at Walmart's scale. It shows up in US analytics teams quietly funding their own data environments and solutions because the global measurement architecture was designed around markets where media complexity and budgets are a fraction of what the US requires. The post-mortems do exist though. They just end up living in internal slide decks and consultant debriefs that no one outside the building ever sees. The company funds the workaround, absorbs the inefficiency, tells itself it's a local configuration issue, and the rot continues under a fresh coat of "we're aligned."
Two Companies Who Did The Math
PepsiCo looked at this honestly a few years ago. They shifted from project-based builds to a platform-first thinking, created a unified digital backbone built to serve everywhere by prioritizing top market needs. No attempting to reconcile competing wish lists of every region into something coherent. They built once at the level of sophistication the most demanding market required and then extended it outward. The result was Sales+, which replaced more than 40 standalone applications with a single commercial ecosystem now running over 500,000 weekly store visits across North America and a formidable plan to scale to other markets. That's a commitment to results over consensus.
Mondelez is doing the same thing, less glamorously but with equal conviction. Their new IT leadership had been clear-eyed about the problem: different technical configurations per region, different operating systems, different databases, different data centers, different everything. The response is a five-year, $1.2 billion program to consolidate three regional tech ecosystems into one shared operating model and stack. Staged by region to limit disruption and building it first at the most complex market by people actually embedded in that market rather than specialists rotating in from a shared services hub several time zones away. Build once and then scale it everywhere else.
Consensus Is Just Slow Surrender
The question worth asking is why so many CPG organizations still can't bring themselves to do this.
A big part of the answer is political. Telling Germany or Brazil that their requirements are downstream of North America's feels like an insult, and nobody wants to be the person who writes that email to peers and friends. So instead they run endless requirements-gathering exercises designed to surface everyone's needs with maximum diplomatic sensitivity, produce a requirements document that is a masterpiece of cross-functional harmony, and build a platform from it that makes everyone slightly unhappy in different ways. The stakeholders feel heard. The system doesn't work well. The workarounds proliferate. And everyone gets to tell themselves they ran a good process.
The other part is structural. The governance machinery of most large CPGs was specifically engineered to prevent any single market or function from having disproportionate influence. Which sounds responsible in theory. The problem though is that not allowing the most complex, highest-revenue market to have disproportionate influence over the IT roadmap defies logic. Allowing that market to take the lead and call the shots is the only arrangement that produces systems sophisticated enough to matter.
Stop Asking If It's Fair
Let’s be clear. There is no clean and comfortable political path to this. The CIO, with genuine CEO backing, who wants to run a US-anchored model rather than the ceremonial kind that evaporates at the first regional pushback, has to be willing to say plainly that not all markets are equal inputs to the roadmap. That some requirements will be deferred indefinitely. That the job of the enterprise platform is to serve the top business that drives revenue and complexity first, and then extend to the business that benefits from it second.
The companies doing it well stopped asking whether that was fair.
They asked whether it worked.
*** Views expressed on Push Pull Pivot are personal and shared in an individual capacity. They do not represent those of any current or former employer ***