Owning the Seams: Integration as a Feature, Not a Fix

Every enterprise of any size is organized in columns. Sales reports to a chief revenue officer. Operations to a chief operating officer. Finance to a chief financial officer. Technology to a chief information officer. Each column has its own head, its own budget, its own tools, its own scorecard, and its own year-end review. This is not accidental. It is how the modern corporation was designed — specialization by discipline, professional depth inside each column, clean accountability at the top of each one.

The work does not respect the columns. A quote-to-cash cycle crosses at least five of them. A capital project crosses seven. A new-product launch crosses almost every function the company has. A new hire on Monday morning has traveled through recruiting, legal, IT provisioning, facilities, payroll, and the manager’s calendar before she can do the first useful thing. Each of those flows is a chain of handoffs, and each handoff is a seam — a place where the tools do not talk, the metrics do not align, and the local optimum wins over the flow.

Operating-model literature has been writing about the shift from function-first to process-first for at least three decades. Companies put slides up in strategy offsites about end-to-end value streams and process ownership. Then they go back to work, and the work is still organized by function, because nothing about the reporting lines, the budgets, or the incentives actually changed. The talk is horizontal. The org is vertical. Predictably, the vertical wins.

Hershey’s three-system go-live is the case study people teach because the failure was public and expensive, but the mechanics are the same wherever a customer order crosses more than one system of record. Orders were captured in one place, credit was cleared in a second, inventory was allocated in a third, and shipments were promised out of a fourth. Each system was live. Between them, the data did not reconcile, the exceptions had nowhere to go, and the swivel-chair operator who used to hold the flow together no longer existed. More than a hundred million dollars in orders sat unfilled with product on the shelf.

Lidl’s program failed more slowly. Seven years of work, a new SAP Retail platform on Hana, a plan to reach 10,000 stores and 140 logistics hubs, and in the end a memo saying the strategic goals could not be reached without spending more than the retailer wanted to spend. The integration gaps were named directly in the retailer’s own explanation. Master data did not agree across the estate. Process breaks between merchandising, logistics, and store operations were treated as somebody else’s problem in each of the three organizations.

The cost of the seam is not always as visible as a shortened Halloween or a headline write-off. Most of the time it shows up quietly. Cycle times swell by a few days at each handoff and nobody notices because no function is missing its own milestone. The month-end close takes an extra week because five systems have to be reconciled by hand. Customers can tell they are being passed between departments and that nobody they can name owns their situation. Projects stall between functions, and the stall is invisible until someone escalates. None of this is a failure of any individual function. It is a failure of the space between them, which is what nobody owns.

The structural reason is incentive design. The head of sales is measured on bookings. The head of operations is measured on cost and delivery. The head of finance is measured on close cycle and controls. None of them is measured on how long a customer order takes to move from signature to shipment, or on how many hires are productive by day thirty, or on the cycle time of a new-product launch from concept to first revenue. The metric that would capture the flow either does not exist inside the company or exists as a report that nobody uses to make a decision.

Where cross-functional process owners exist, they often exist on paper. A vice president of order-to-cash. A director of new-product introduction. Their titles imply authority across function lines. Their reality is that they can convene meetings and escalate. They cannot reassign resources, cannot change how another function is measured, and cannot break a tie with the head of sales or the head of operations. Responsibility without authority is a well-known way to make sure the job does not get done, and horizontal roles are the most common place it happens.

The tooling matches the org. Systems arrived function by function, over years, chosen by leaders solving their own function’s problem. Sales landed on a CRM. Finance on an ERP. Operations on an MES. Service on a ticketing platform. Engineering on a PLM. Each was a defensible choice for the function that chose it, and none of them speak to each other without middleware, batch jobs, and someone copying data from one screen to the next. The horizontal flow crosses systems that were designed to be islands.

And the data reflects it. Five owners each have their own version of what a customer is, what an order is, what an active project is, what a completed hire is. The reports do not reconcile because the definitions do not agree. The month-end conversation turns into an argument about which system is right, rather than what the business is doing. Every enterprise architect in the room knows this. The reason it does not get fixed is not that the problem is invisible. It is that fixing it forces functions to give up things they thought they owned.

A workflow needs an owner in the same sense a system needs one. One name, on the hook for the flow working end to end — not a coordinator, not a co-chair of a steering committee, not a title with dotted lines to five functions. Someone whose scorecard is the cycle time and quality of the flow, and who has enough authority over budget, metrics, and escalation to make the trade-offs the flow requires.

Budget authority means the process owner controls the shared work — the integration layer, the master-data cleanup, the exception queues, the workflow platform if there is one. Not the whole budget of every function the flow crosses. Enough of the shared budget that the answer to which function funds the interface between two systems is not a six-week negotiation. Metric authority means the process has a measurable cycle time and quality that show up on someone’s scorecard at the executive level, alongside the functional numbers rather than in a separate deck that gets skipped for time. Escalation authority means a named tie-break at the top, and a standing understanding that the tie-break gets used.

The role fails predictably when any one of the three is missing. Budget without metrics produces activity nobody can grade. Metrics without budget produces reports nobody acts on. Both without an escalation path produces a queue of unresolved trade-offs and, eventually, a resignation letter. The role is a design, not a title, and the design has three parts.

The path forward is not another matrix. Matrix reporting was the last generation’s answer to this problem, and it mostly produced longer meetings and softer accountability. The path is to build the horizontal deliberately, in a specific order, with the same care the functions were built with.

Start with the operating model. Name the four or five flows that actually matter to the business — quote to cash, idea to launch, hire to productive, plan to deliver, incident to resolution, whatever the shape of the business calls for — and describe each of them as a real process. Handoffs identified. Ownership of each handoff explicit. Exception paths written down. This is uncomfortable work because it forces functions to give up things they thought they owned and to accept accountability for things they thought were somebody else’s. Skipping the discomfort turns the operating-model exercise into a slide deck rather than a change.

Give the process owners real authority next, not last. Budget, metric, escalation. A named executive tie-break. A scorecard that lives in the same review as the functional scorecards. If the process owner cannot make a decision that a functional leader has to live with, the role is decorative, and every downstream investment in tools and data will land on top of the same old ownership gap.

Data governance follows the flow, not the function. If a customer record moves through five systems, one person is accountable for what a customer is, across all five. Same for orders, projects, employees, and whatever else the business runs on. The definitions are set once and enforced at the boundaries, and the systems reconcile because they are working from the same shared meaning rather than five local ones. Master-data cleanup done inside a single function produces a cleaner column and does not fix the flow.

Enterprise architecture stops being a fiefdom exercise and becomes a workflow exercise. Systems are pieces of a flow, not endpoints of a function. When a new system is under evaluation, the question is not what does this let the function do, but where this fits in the flow that crosses the function, and what integration work has to come with it. The Hershey lesson, rediscovered often, is that coupling three systems on the same weekend without an owner for the coupling is a decision, not an accident.

Then, and only then, put the workflow platform behind it. Modern workflow tools have gotten better, and used correctly can make the horizontal flow visible in a way that a stack of siloed systems cannot. But the platform is downstream. Installing it before the operating model is designed and the ownership is named produces an expensive way to keep doing the same handoffs, now with a nicer dashboard. The Lidl program is that story at scale.

Governance reviews cross-functional metrics, not just functional ones. If the executive team only looks at function-level scorecards, the flows do not exist inside the meeting that runs the company. Adding two or three horizontal metrics to the standing review changes what gets talked about, and what gets talked about gets fixed.

Integration diagrams tend to show boxes and arrows. The boxes are named, funded, and staffed. The arrows are not. That is the design flaw, drawn in the same picture the leadership team approved. Every one of those arrows is a piece of work that someone has to own, with authority to change how it runs when it stops running well.

The standard worth holding is that every flow that matters to a customer, or to a strategic outcome, has a named owner with the authority to hold the flow together across functions. Not a coordinator. An owner. Someone whose scorecard is the cycle time and the quality of the flow, and who has enough budget, metric, and escalation authority to make the trade-offs the flow requires. That standard is recognizable to any senior operator, and most senior operators would agree it is not what they have today.

The uncomfortable part is that closing the gap forces a conversation about who gives up what. Function heads have to accept a metric on their scorecard that they do not fully control. Process owners have to accept accountability for outcomes that live inside other people’s teams. The chief executive has to be willing to break ties in favor of the flow when the flow and the function disagree. None of that is a technology decision. It is an operating decision that has to be made before the next platform arrives, or the next platform will land in the same organization that the last one did, and produce the same shape of failure.

Hershey shipped Halloween the year after. Lidl went back to a version of its legacy system and kept running. Businesses survive; they usually do. What did not survive was the assumption that the flow would take care of itself once the systems were in. Value moves sideways. Accountability, on most org charts, still moves only up and down. The organizations that close the gap treat the horizontal not as a coordination problem to be smoothed over, but as a design problem to be solved — with owners, with authority, and with governance that looks at the flow instead of at the columns.