The invisible problem of modern retail: decision fragmentation

Ask ten retail leadership teams what their main operational problem is, you'll get ten different answers. "Our overstock." "Our mis-calibrated markdowns." "Our stockouts on best-sellers." "The slowness of our replenishments." "The lack of coordination between our stores." "Our predictive models that aren't applied on the floor."

Each answer is accurate. None is complete. Because above each of these symptoms sits a deeper, more structural, and almost always invisible problem: decision fragmentation. That structural characteristic of modern retail where every operational decision is taken in an isolated system, by a specialized team, with partial data, in its own cycle, without coordination with the rest of the network's decisions.

This fragmentation has no identified direct cost. No reporting surfaces it. No KPI measures it. And yet it's what explains why, despite massive investments in data, modern ERPs, BI, planning and predictive models, the sector's operational performance progresses so slowly. Not because the tools are bad. Because the decisions they produce never converge into coherent steering.

This article looks straight at this fragmentation: what it exactly is, why it settled in, what it really costs, and what it takes to overcome it.

Anatomy of a modern retail decision

To understand fragmentation, you first have to look at what an operational decision has become in a contemporary chain. Take a concrete case. A product, in a store, under-performs.

What happens? First step: sales data is recorded in the POS, consolidated in a data warehouse, and appears several days later in a BI dashboard consulted by the merchandising team. Second step: the merch team identifies the under-performance, often at a category aggregate level, and launches an analysis — usually in Excel, with reprocessed extracts. Third step: a meeting is held with pricing to discuss a possible markdown. Fourth step: the decision, if taken, is entered into the ERP for price update. Fifth step: the information propagates to stores, who update their displays. Sixth step: two to three weeks later, the effect on sales is observed.

Meanwhile, in parallel, on the same product, other decisions are taken by other teams, in other systems, on other cycles. Supply chain looks at stock cover and decides whether or not to replenish. E-commerce adjusts its digital merchandising based on its own KPIs. Stores manage their local exposure based on intuition. Nobody, at any point, consolidates these decisions at the product level to make sure they converge.

The result is almost always sub-optimal. While merch considers a markdown, supply chain is replenishing the product in the same store. While e-commerce moves the product up the homepage to revive sales, stores tuck it at the back of the shelf. Pricing accepts a discount to restart rotation, but the data they're based on is already stale because it predates the previous day's digital merchandising change.

Each team does its job correctly. And the global result is disordered — because nobody steers the whole.

The six fractures that compose fragmentation

This fragmentation isn't a monolithic phenomenon. It's composed of six distinct fractures that compound and reinforce each other.

Fracture #1: data fragmentation

The most known, and historically the first addressed. Data lives in different systems — POS, ERP, e-commerce, WMS, HR tools, BI — that don't speak to each other spontaneously. Retailers have massively invested for fifteen years in data warehouse, data lake, data integration projects. The promise was to unify data for coherent steering.

The promise has been partly delivered. Historical data is better unified today than ten years ago. But live data — what's needed for near real-time operational steering — stays largely fragmented, desynchronized, of uneven quality across sources. The result: operations teams work with snapshots of reality not taken at the same moment.

Fracture #2: tool fragmentation

Each decision type has its dedicated tool. Markdown in a pricing tool. Replenishment in a planning tool. Transfer in the WMS. Promotion in yet another tool. Each tool is competent for its specific decision — but none sees the other decisions in progress on the same product.

This tool specialization made sense in a world where decisions were infrequent and weakly interdependent. It becomes a massive handicap in modern retail where the same SKU can, in the same week, be subject to a considered markdown, a scheduled replenishment, a proposed transfer, and an activated promotion — without any of these signals being consolidated.

Fracture #3: organizational fragmentation

Each team has its perimeter, its objectives, its KPIs. Merch optimizes rotation and gross margin. Supply chain optimizes service rate. Pricing optimizes elasticity. Stores optimize their local P&L. Each does their craft correctly.

The problem is that these objectives are partly contradictory. Raising service rate implies accepting overstock. Protecting gross margin implies delaying markdowns at the risk of future stockouts. Optimizing the local store P&L prevents inter-store transfers beneficial to the network. Without explicit arbitration above these objectives, each team optimizes locally — and the sum of these local optimizations produces a global sub-optimization.

Fracture #4: temporal fragmentation

Decision cycles are desynchronized. Merch runs weekly. Supply chain daily. Pricing by campaign. Stores in real-time. Planning by season.

This desynchronization creates permanent side effects. The data available Monday for the merch meeting is no longer up to date Tuesday when supply chain takes its decisions. The markdown validated in week 12 is executed in week 14, but the replenishment decision that renders it obsolete was taken in week 13. The result is a flow of disordered decisions, where each corrects or contradicts the previous without ever converging to an optimum.

Fracture #5: decisional fragmentation itself

Here's the subtlest fracture, and the most structural. Even if all other fractures were resolved — unified data, integrated tools, aligned teams, synchronized cycles — one problem would remain: the decision itself isn't conceived as a unified object.

In most chains, you don't talk about "the decision on product X in store Y". You talk about "the markdown", "the replenishment", "the transfer" — as if they were separate objects. In reality, they're all modalities of the same underlying decision: what to do with this stock, in this context, to maximize net margin at a defined horizon?

As long as the decision isn't conceptualized as a single object with several possible modalities (transfer, mark down, replenish, return, do nothing), it will stay fragmented by essence — not just by accident.

Fracture #6: executive fragmentation

The last fracture: execution. Even a perfectly coordinated decision must, ultimately, cross several systems to become operational reality. Price update in the ERP. Transfer order generation in the WMS. Information to stores. Sync with e-commerce.

Each of these steps adds delay, friction points, desynchronization risks. A decision taken on a Monday can take five to ten days to become effective on the floor. During this delay, context has changed, and the decision is already partly obsolete by the time it executes.

The hidden cost of fragmentation: a multiplicative effect

These six fractures don't add — they multiply. That's what makes their global cost so hard to measure, and so massive in reality.

Take a concrete example. On a given product, the probability that the optimal decision is taken depends on the convergence of all the fractures. If data is up to date with 80% reliability, if the tool used is fit with 80% relevance, if objectives are aligned 80%, if cycles are synchronized 80%, if the decision is conceptualized correctly 80%, and if execution happens in time 80%... then the probability that the final decision is actually optimal is 0.8 to the 6th, about 26%.

Three quarters of decisions are, by construction, sub-optimal — not because one step failed, but because the complete chain has too many fragmentation points to systematically produce coherence.

At a retailer generating €500M with 3% net margin, this systemic sub-optimization regularly amounts to 3 to 8 EBIT points unrecovered. Not a marginal cost. Probably the sector's largest performance reservoir — and the one that remains, almost everywhere, structurally left on the table.

Why fragmentation settled in (and why it persists)

You have to understand that fragmentation isn't a governance failure. It's the result of a perfectly rational historical accumulation that only becomes a problem with time.

For thirty years, retailers equipped their functions one after the other. First POS, because cash had to be digitized. Then ERPs, to structure management. Then planning tools, because assortment complexity exploded. Then BI, to regain control of data. Then pricing, e-commerce, WMS, promotion tools, ML models.

Each investment, taken in isolation, was justified and produced value. But the sum of these investments, without an integrative architecture, produced a composite system where each tool solves a local problem and nobody carries global coherence. It's the equivalent of a city built neighborhood by neighborhood, without an urban plan: each neighborhood works, but traveling between them is a nightmare.

This history also explains why fragmentation persists. No team has an interest in solving it, because each invested in its tool, trained its people, structured its processes. Fragmentation is, paradoxically, a stable equilibrium — costly for the organization, but comfortable for each silo.

The wrong fix: stacking a governance layer

Faced with this observation, many chains have tried the same response: adding a governance layer on top of existing tools. Cross-functional committees, consolidated dashboards, cross-steering rituals.

These approaches are useful, but they never solve the underlying problem. For three reasons.

First, human governance doesn't scale to the complexity of modern retail decisions. When there are ten thousand decisions to coordinate each week on a network, no committee is enough. Committees handle the salient cases, miss the overwhelming majority of the flow.

Next, governance through dashboards doesn't change the mechanics of decisions. Visualizing fragmentation isn't solving it. You get better awareness of the problem, not better steering. And awareness without action quickly becomes frustration.

Finally, governance adds steps rather than removing them. The time spent in cross-functional meetings, arbitration committees, cross-validations becomes an extra cost — added to an already high initial fragmentation cost. At some point, retailers spend more time managing fragmentation than making operational decisions themselves.

The real answer isn't in added governance. It's in decision reunification at the level where they form.

What to build: a unified decision layer

Exiting fragmentation requires a structurally different approach. Not replace existing tools — they remain necessary for what they do well. But add, on top, a unified decision layer that consolidates signals, formulates arbitrations, and propagates executed actions.

Concretely, this layer must carry five capabilities.

Capability 1: unify live data. Not a historical data warehouse, but a continuous, fresh, usable view at the SKU/store level, that aggregates in near real-time the signals from all operational systems.

Capability 2: think the decision as a single object. Instead of reasoning by silo (markdown, replenishment, transfer, promotion), reason by integrated decision: "on this SKU in this store, what action — across all modalities — maximizes net margin at a given horizon, under constraints?"

Capability 3: integrate business rules as first-class citizens. Margin floors, supplier constraints, commercial calendars, brand- or cluster-specific handling — all the rules that structure the real decision must be formalized, contextualized, prioritized, executable.

Capability 4: execute without breakage. A validated decision must propagate into execution systems (ERP, WMS, pricing, e-commerce) without human re-entry, with a delay compatible with real commercial dynamics.

Capability 5: learn in loop. The observed effect of a decision must feed the system to adjust future decisions. Without this learning loop, the unified layer itself degrades over time.

A platform carrying these five capabilities radically changes operational dynamics. Teams no longer spend their time coordinating fragmented decisions — they arbitrate the truly complex cases, based on coherent recommendations integrating all action modalities and the chain's constraints.

The Solya approach: the missing layer between your tools and your decisions

That's exactly the mission Solya carries at retailers who want to overcome fragmentation. Not replace your ERP, your BI, your ML models, your planning tools — they stay in place, in their role. But add, on top, the unified decision and execution layer they're missing to produce performance together.

Concretely, Solya connects to your data sources — POS, ERP, e-commerce, supply chain, internal tools — and rebuilds a live view of the network at the SKU/store level. The decision engine reasons on this unified view to continuously formulate integrated recommendations combining all possible action modalities: markdown, transfer, replenishment, supplier return, targeted promotion, or argued status quo. Your business rules — margin floors, supplier constraints, commercial calendars, specific handling — are embedded in the engine, accessible to business teams, continuously parameterizable. Validated decisions propagate to your execution systems without re-entry, with a delay compatible with commercial cadence.

The paradigm shift is simple to state: you move from a retail where each decision is taken in a silo, with partial data, on its own cycle, to a retail where decisions are thought, arbitrated and executed as a unified flow. Teams keep the hand on every structural trade-off — they define rules, validate sensitive cases, adjust strategy. Solya takes care of operational coherence: making sure thousands of weekly decisions converge rather than contradict.

Chains that crossed this threshold report gains that aren't marginal. Between 1 and 3 net margin points recovered, not by selling more products, but by stopping to pay the invisible bill of fragmentation. And beyond the numbers, a deeper change: the recovered ability to steer the network as a system, not as a collection of stores, categories, and isolated decisions.

The real question to ask

How many operational decisions does your organization take, each week, in real cross-team coherence? Not how many are taken in total — how many are taken with consolidation of all relevant signals, respecting the full set of business rules, and converging with the rest of the network's decisions.

If the answer is "a fraction", you live fragmentation daily — without necessarily naming it as such. And you probably let slip, every season, most of modern retail's most accessible performance reservoir.

Fixing this fragmentation isn't yet another IT project. It's not yet another tool to stack. It's a paradigm shift: moving from a silo-steered retail to a unified-decision-steered retail. And it is, more than any other transformation, what separates today's retailers who endure their network from those who actually steer it.

---

Map your decisional fragmentation

At Solya, we offer retail leadership teams a personalized 30-minute diagnostic to identify, on your own context, where the main decisional fractures of your organization sit — and quantify the margin potential recoverable through a unified decision layer.

👉 [Book your Solya diagnostic] — 30 minutes, by video, with one of our retail experts.

You'll walk away with:

• A map of the six decisional fractures on your perimeter
• A quantified estimate of margin potential recoverable through their resolution
• The first high-ROI use cases to move from silo-steered to unified-decision steered