The Control Phase
The Improve phase delivered results. Performance is up, defects are down, and the team is energized. But here is a hard truth from banking operations: without deliberate controls, processes drift back to their old ways within months. Staff turnover, shifting priorities, system changes, and sheer habit all conspire to erode improvements over time.
The Control phase ensures that the gains you achieved in Improve are locked in permanently. It is the difference between a temporary spike in performance and a lasting transformation in how the process operates.
The Control phase answers five questions: What will we monitor? How will we monitor it? Who is responsible? How often do we check? And what do we do when something goes wrong?
Control Plans
A control plan is the single most important deliverable of the Control phase. It is a living document that specifies exactly how the improved process will be monitored and maintained going forward.
Every control plan should include the following elements:
Metric to monitor — The specific KPI that reflects process health. For a reconciliation process, this might be the daily auto-match rate or the number of aged breaks over five days.
Measurement method — How the metric is captured. Is it pulled from a system report, calculated from a data extract, or recorded manually? Define the source and calculation precisely.
Responsible owner — The individual accountable for monitoring the metric and responding to deviations. This should be a named person, not a team or function.
Monitoring frequency — How often the metric is reviewed. Daily dashboards for operational metrics, weekly for trend analysis, monthly for management reporting. The frequency should match the speed at which the process can deteriorate.
Control limits and targets — The acceptable range of performance. Define the target, the upper control limit (UCL), and the lower control limit (LCL). Anything outside these boundaries triggers action.
Response plan — What happens when a metric breaches its control limits? Specify the escalation path, the investigation steps, and the corrective actions. A control plan without a response plan is just a reporting exercise.
For a reconciliation process, a control plan row might read: "Monitor daily auto-match rate via the Reconciliation Dashboard. Owner: Reconciliation Manager. Frequency: daily review, weekly trend analysis. Target: 95%. LCL: 90%. If match rate falls below 90% for two consecutive days, escalate to Operations Head and initiate root cause investigation within 24 hours."
Standard Operating Procedures
If the control plan tells you what to monitor, the Standard Operating Procedure (SOP) tells everyone how to do the work. Updated SOPs codify the improved process so that every person — including new joiners six months from now — follows the same standardized approach.
An effective SOP in banking includes these key elements:
- Purpose — Why this procedure exists and what process it governs.
- Scope — Which products, teams, or entities the SOP applies to.
- Step-by-step instructions — Clear, numbered steps for executing the process. Include screenshots or system references where relevant.
- Exception handling — What to do when the standard path does not apply. In banking, exceptions are inevitable — the SOP should address the most common ones explicitly.
- Escalation procedures — Who to contact when an issue cannot be resolved at the operational level, and the expected response time.
- Version control — Date of last update, author, and approval. SOPs that are never updated quickly become shelfware.
The most common failure with SOPs is writing them once and forgetting them. Build a review cycle — quarterly at minimum — into the control plan itself. When the process changes, the SOP must change with it.
Control Charts: An Introduction
Control charts are the visual heartbeat of the Control phase. They plot a metric over time and overlay statistical limits that distinguish normal variation (common cause) from abnormal variation (special cause). This distinction matters enormously in banking operations, where reacting to every minor fluctuation wastes resources, but ignoring a genuine shift risks operational failure.
The most common control chart for continuous data is the X-bar chart. It has three key lines:
Center line (CL) — The mean of the process metric. This represents the expected average performance after improvement.
Upper Control Limit (UCL) — Typically set at the mean plus three standard deviations (mean + 3 sigma). This is the upper boundary of expected normal variation.
Lower Control Limit (LCL) — The mean minus three standard deviations (mean - 3 sigma). The lower boundary of expected normal variation.
When all data points fall within the UCL and LCL with no patterns, the process is in control — variation is random and expected. But several signals indicate special cause variation that demands investigation:
- A single point outside the UCL or LCL — Something unusual happened on that day. Investigate immediately.
- A run — Seven or more consecutive points on the same side of the center line. The process has shifted.
- A trend — Six or more consecutive points moving consistently upward or downward. The process is drifting.
Consider a banking example: after automating a cash reconciliation process, the team tracks the daily auto-match rate on a control chart. The center line is 95.2%, the UCL is 98.1%, and the LCL is 92.3%. For two weeks, all points fall within limits. Then on day 15, the match rate drops to 88%. The control chart immediately flags this as special cause variation. Investigation reveals that a counterparty changed their payment reference format, causing the matching algorithm to fail. The team adjusts the matching rules and performance returns to normal. Without the control chart, this issue might have gone unnoticed for days, generating a backlog of unmatched items.
Visual Management
Control charts are one element of a broader visual management approach that sustains improvement in banking operations. Effective visual management tools include:
Real-time dashboards — Display key metrics on screens visible to the entire team. When the auto-match rate, break ageing, or SLA performance is visible to everyone, accountability is built into the environment.
Daily stand-ups — Brief (10-15 minute) team meetings each morning to review yesterday's metrics, flag today's risks, and allocate resources. Many banking operations teams have adopted this practice from agile methodology with strong results.
Exception boards — Physical or digital boards that track open issues, aged items, and escalations. Making problems visible ensures they get attention rather than languishing in email threads.
The principle behind visual management is simple: what gets measured and displayed gets managed. When performance is transparent, teams self-correct faster and problems are caught earlier.
Handoff and Sustainability
A DMAIC project is, by design, temporary. The project team will eventually move on to other work. The handoff to the day-to-day process owner is the final critical step — and the moment where many improvements silently begin to erode.
A successful handoff includes:
Formal transfer of the control plan — Walk the process owner through every metric, limit, and response action. Do not simply email a document.
Updated SOPs and training — Ensure the process owner and their team have been trained on the new procedures. Provide job aids for quick reference.
Periodic reviews — Schedule monthly reviews for the first quarter after handoff, then quarterly thereafter. These reviews check that the control plan is being followed and metrics remain within limits.
Continuous improvement mindset — The end of one DMAIC cycle is not the end of improvement. Encourage the process owner to identify further opportunities and feed them into future projects.
Banking Example: Maintaining Automated Reconciliation
A bank's securities operations team completed a DMAIC project that automated their daily reconciliation process, increasing the auto-match rate from 68% to 95%. Here is how they set up the Control phase:
The control chart tracked the daily auto-match rate with a center line at 95%, UCL at 98.5%, and LCL at 91.5%. The SOP documented the new automated workflow, including how to handle the remaining 5% of items that required manual investigation, with clear escalation paths for breaks over 48 hours. The control plan assigned the Reconciliation Manager as owner, with daily dashboard reviews and weekly trend analysis reported to the Operations Head.
A weekly review meeting every Monday morning brought together the reconciliation team lead, technology support, and the operations manager to review the previous week's control chart, discuss any breaches, and agree on actions. After three months of stable performance, the project team formally handed off to the Reconciliation Manager with full documentation and trained a backup owner to ensure continuity.
In the next module, we will bring the entire DMAIC cycle together with a comprehensive banking case study — corporate actions processing — where you will see how every phase connects from Define through Control.