Managing Change at Merger Scale: Why Financial Integrations Fail Long After Systems Go Live

Large financial mergers are often judged by visible milestones: a clean legal close, a stable cutover weekend, a new org chart that looks settled. When those moments pass without incident, integration is treated as a solved problem.
But the most expensive failures tend to appear later and quieter. They show up as decision paralysis, duplicated controls and teams taking different paths through the same policy because the organisation never fully reconciled how authority and risk now work. In a regulated institution, those problems rarely create a single dramatic outage. They accumulate. By the time they become visible, they have already reshaped how the business operates.

Shalini Mani leads work inside that transition layer. As Senior Director at Discover and a recipient of Discover’s President’s Award, she coordinates cross-functional merger integration efforts following the Capital One–Discover combination. Her lens is not limited to systems and schedules. It is centered on how organisations preserve decision quality while rules, ownership and accountability are still in motion.

We spoke with her about why change management has become the decisive capability in large financial integrations.

When large financial mergers struggle, where do the first cracks actually appear?
They rarely appear as outages. The earliest signs are behavioural. Decisions that used to take hours start taking days. Teams begin escalating routine calls because no one is fully sure which standard governs the work now.

That uncertainty begins the moment a merger becomes real. People start revising how they interpret risk, approvals and accountability. Even if platforms have not changed, the context has. If leaders do not manage that shift deliberately, the organisation slows itself down. Workarounds appear. Controls are applied inconsistently. By the time systems converge, the operating rhythm has already been damaged.

A merger that is technically sound can still fail if the institution teaches its teams to hesitate.

What changes when integration happens at the scale of a national card issuer?
Scale removes the margin for informal alignment. In a small integration, a mismatch in definitions or a policy interpretation gap can be corrected locally. At the merger scale, it propagates.

That is why the combined issuer profile matters. Public reporting around the Capital One–Discover deal has described the merged card book as roughly $250 billion in card balances, which signals the level of exposure created by even small decision inconsistencies.

At that magnitude, integration is not primarily a technology exercise. It is a coherence exercise. Teams need explicit decision mechanics while the organisation is still in transition: which standard wins when legacy rules conflict, how exceptions are handled, and who is accountable in the interim. Without that, people either default to old habits or stop making decisions altogether. Both outcomes create risk before customers see anything change.

That is also why our integration work was recognized with an Integration Team Award for the Capital One–Discover merger. The award reflected the unglamorous part of the job: keeping definitions, approvals, and interim ownership crisp enough that thousands of decisions can still move without constant escalation.

What failure patterns repeat across large change programs, even in well-run organisations?
Two patterns show up repeatedly.

The first is treating transformation as separate from operations. Many programs plan as if the business can continue unchanged while integration runs in parallel. In reality, the front line absorbs the impact immediately. When escalation paths, ownership and metrics are not updated for transition conditions, operations invent their own stability. That is where drift begins.
The second is over-building the future state while under-defining the transition state. The in-between period can last longer than leaders expect. If the organisation does not define how success is measured during transition, teams will use whichever metrics feel safest. If it does not define how accountability works during transition, teams will push decisions upward until the system clogs.

Most integration decay is not sabotage. It is ambiguity left unattended.

Why does governance become the real integration problem in regulated financial services?
Because regulators evaluate decision integrity, not just outcomes. In a merger, the question is not only whether processes work. It is whether decisions can be traced, explained and defended while systems and controls are still converging.
That forces a different approach to change design. Evidence cannot be optional. Data lineage cannot be treated as a later cleanup step. Controls must be mapped to interim states, not only to the end state.

That perspective comes from spending a lot of time looking at work that fails under scrutiny. In my role as Judge for the ESP’s International Research Journal of Engineering, Management & Science, I have seen that outcomes alone are rarely the problem. What breaks is traceability. The work that holds up is the work where assumptions are explicit, decision paths can be followed end-to-end, and governance is designed into the delivery rather than added afterwards.

In integrations, the same rule applies. If the institution cannot explain why a decision happened, it will eventually be forced to rework the decision, usually at the worst time.

Cost synergies create urgency. How do you avoid letting that urgency destabilise execution?
By refusing to treat efficiency as the governing signal.
When teams believe the primary goal is cost takeout, they optimize defensively. They postpone decisions, over-escalate risk calls and preserve local processes that feel safe. Integration slows precisely when leadership believes it should accelerate.

In the Capital One–Discover case, public merger coverage has cited expected $2.7 billion in pre-tax synergies in 2027, which creates pressure to move quickly. That pressure needs sequencing discipline. Stabilise decision paths and controls first. Consolidate redundancy only after reliability is established.
Efficiency achieved without clarity does not compound. It collapses under the first real stress event.

What is the durable trend here, and how do you know when change management is actually working?
Consolidation is not a short-term event. It is a structural pattern driven by scale economics, rising compliance demands and the cost of maintaining parallel platforms. Global M&A value was projected to reach about $3.5 trillion in 2024, which signals that dealmaking is not disappearing, even under macro uncertainty.

In that environment, institutions will not be differentiated by how quickly they announce integrations. They will be differentiated by how coherently they operate during prolonged transition.

I always prioritise consistency over speed. Are similar issues resolved the same way across teams? Are decisions being made at the right level without constant escalation? Do people understand why a process changed, not only that it did? When those signals stabilise, performance metrics follow. When they do not, even a technically successful integration becomes fragile.

The core lesson is uncomfortable but practical: most merger risk lives in the period where nothing is final, yet everything still has to work.