Operational Excellence (2026): Frameworks, Operating Cadence, Metrics and How AI Changes It

TL;DR

• Operational excellence is the discipline of running a company so it executes its strategy consistently, with problems found and fixed at their source and improvement built into the weekly routine rather than run as a project.
• The main operational excellence frameworks are Lean, Six Sigma, the Toyota Production System and the Shingo Model. Each gives you tools. None of them works without an operating cadence that finds problems and tracks who fixes them.
• A practical system has four parts: a cadence of daily, weekly, monthly and quarterly reviews, a short set of metrics with owners, a standard way to find root causes, and a log that follows every action to closure.
• Sentra is the layer that makes the cadence proactive. It reads the company's systems and conversations, finds what is going wrong before the review, tells the owner, and tracks whether the fix happened.

What is operational excellence?

Operational excellence is a management approach in which a company executes its strategy more consistently and efficiently than its competitors, by standardizing how work is done, measuring it, and improving it continuously at every level. In practice, operational excellence means problems surface early, reach the person who can fix them, and stay fixed.

The term entered strategy language through Michael Treacy and Fred Wiersema's 1993 Harvard Business Review article, "Customer Intimacy and Other Value Disciplines" (hbr.org). They described operational excellence as one of three value disciplines, alongside customer intimacy and product leadership, and argued that market leaders pick one to excel at while meeting the industry standard on the other two (saylordotorg.github.io).

Today the term is used more broadly. A COO running an operational excellence program is usually not choosing a value discipline. They are building the routines that keep a company's execution reliable as it grows: clear metrics, named owners, fast root-cause work and follow-through.

Operational excellence frameworks

Four frameworks account for most operational excellence programs. They overlap, and most companies borrow from several.

Lean

The Lean Enterprise Institute defines lean as "a way of thinking about creating needed value with fewer resources and less waste" (lean.org). The widely taught five principles, from James Womack and Daniel Jones's 1996 book Lean Thinking, are to identify value from the customer's view, map the value stream, create flow, establish pull and pursue perfection (umbrex.com). Lean is strongest where work moves through a sequence of steps and waiting time is the main cost.

Six Sigma

Six Sigma reduces variation in a process. A Six Sigma process is one where the closest specification limit sits six standard deviations from the mean, which equates to 3.4 defects per million opportunities (isixsigma.com). Its improvement method is DMAIC: Define, Measure, Analyze, Improve, Control. Six Sigma is strongest where defects are measurable and costly, such as manufacturing, claims processing or billing.

Toyota Production System

Toyota describes the Toyota Production System as resting on two pillars (global.toyota). Jidoka, "automation with a human touch", lets a machine or worker stop production the moment a problem appears so defects do not move downstream. Just-in-Time means "making only what is needed, when it is needed, and in the amount needed."

The Shingo Model

The Shingo Model, maintained by the Shingo Institute at Utah State University's Jon M. Huntsman School of Business, organizes ten guiding principles into three dimensions (shingo.org).

• Cultural enablers. Respect every individual. Lead with humility.
• Continuous improvement. Assure quality at the source. Improve flow and pull. Seek perfection. Embrace scientific thinking. Focus on process.
• Enterprise alignment. Create value for the customer. Create consistency of purpose. Think systemically.

The Shingo Model is less a toolkit than a test of culture. It asks whether behaviors across the company reflect those principles, not only whether tools are in use.

Operational excellence framework comparison

The operating cadence

Frameworks describe how to improve a process. The operating cadence decides whether improvement happens every week. A typical cadence for a mid-sized company has four layers.

Each layer escalates what it cannot solve to the next and passes decisions back down. The weekly layer is where most companies are weakest, because it depends on owners arriving with causes rather than guesses. Our guide to the weekly business review covers how Amazon runs it and gives a template.

Operational excellence metrics

Pick a short set of metrics that show whether the operating system is working, not only whether output is on target. Each needs one owner and a threshold.

OEE combines availability, performance and quality losses, and reaches 100% only when all three are perfect (oee.com). For how OTIF can look healthy while shelves are empty, see our OTIF guide.

The last four rows matter most for leadership. They measure the operating system itself. A company can hit its output targets for a quarter on effort alone. It cannot keep doing it if problems are found late, explained slowly and fixed twice.

Operational excellence examples

Toyota: stop at the first defect

Jidoka is the oldest example of operational excellence built into the work itself. Any worker can stop the line when something is wrong, so a problem is handled where it starts rather than found at final inspection (global.toyota).

Amazon: inputs over outputs

Amazon's weekly business review reviews controllable input metrics ahead of outputs. In one case the team learned through the WBR that counting new product pages was the wrong input for selection, and replaced it with the share of page views where the item was in stock and ready to ship in two days (holistics.io). The metric changed because the review kept testing whether inputs predicted outputs.

Lenskart: the report said green

At Lenskart, the eyewear retailer with 3,200 stores, reports showed a top product line about 95% in stock while Sentra measured 12.8% availability on the shelves. About $860K of sales were lost in the two weeks before the fix, and time to root cause fell from 3 to 4 days to about 30 minutes. The metric was correct by its own definition. The operation underneath it had changed.

Continuous improvement and why programs stall

Continuous improvement is the habit of finding a problem, confirming its cause, fixing it and checking the fix held. DMAIC is one version of that loop. Amazon's WBR was originally built on it (goodreads.com).

Programs rarely stall because the method is wrong. They stall at three handoffs.

• Detection. The problem is visible somewhere, in a supplier email, a support thread or a site manager's chat, long before it shows on a dashboard. Nobody joins those signals to the metric.
• Ownership. The problem is raised in a meeting, but nobody is sure who owns it. It waits for the next meeting.
• Follow-through. An action is agreed and written in someone's notes. Three weeks later nobody can say whether it happened.

Each handoff depends on someone reading across systems and conversations and remembering what was agreed. That is the work that scales worst as a company grows. For the step that limits throughput in a process, our bottleneck analysis guide shows how to find it, including when the bottleneck is a person rather than a machine.

Operational intelligence and decision intelligence

Two adjacent terms come up in operational excellence work.

Operational intelligence is analytics applied to live operations: monitoring current conditions, alerting on exceptions and helping teams respond while there is still time. Its sources are usually event streams and systems of record.

Decision intelligence is broader. TechTarget describes it as a field that combines data science, AI and decision theory, and notes that Gartner narrowly defines it as "the process of documenting and engineering how decisions are made" (techtarget.com).

Both usually start from structured data. The gap in most operations is the unstructured half: the reasons, promises and warnings that live in email, chat and meetings. Without them, operational intelligence can tell you a number moved but not why, and decision intelligence can model a decision but not see that it was already made in last Tuesday's call.

How AI changes operational excellence

AI changes the three handoffs that stall improvement programs.

• Problems found before the review. A system that reads both the metrics and the conversations can see a supplier delay in an email before it appears as an OTIF miss, and raise it while it is still cheap to fix.
• Owners notified. Instead of a problem waiting for the next meeting, the system works out who owns the area and tells that person directly, with the evidence.
• Follow-through tracked. Every commitment made in a meeting or a thread is recorded with an owner and a date, and checked against what actually happened. Missed commitments come back automatically.

The review meeting does not go away. It changes from a hunt for problems into a check on decisions already in motion.

Where Sentra fits

Sentra understands how a company really runs and acts on it before anyone asks. For an operations leader, that means the three handoffs above stop depending on someone remembering.

Sentra connects email, chat, meetings and documents alongside ERP, CRM, HRIS and BI, through 200+ integrations. It reasons over them to work out what is true now, why it changed, and who is working on it. It knows when a fact became true and when it stopped being true, so an old status is not reported as current. When something starts to go wrong, such as a stalled dependency, a supplier slip or a commitment past its date, Sentra works out who should act and tells them, with every claim cited to its source.

A named person approves anything Sentra raises before it reaches anyone else. Answers respect the permissions of each source, so people see only what they could already open. Sentra replaces none of the company's systems and is not a Lean or Six Sigma toolkit. It supplies the evidence and the follow-through your existing method depends on, and any agent or model, including ChatGPT, Claude, Copilot and Gemini, can use the same company memory over REST and MCP. Sentra holds SOC 2 Type II and ISO 27001 certifications and deploys in the cloud, in an isolated VPC or air-gapped on premises.

For more on how this works across an operations team, see company brain for operations. To see it on your own operation, book a demo at /demo.

How to start an operational excellence program

1. Pick one value stream. Order to cash, procure to pay or hire to productive. Map it end to end with the people who run it.

2. Choose six to ten metrics with owners. Include at least two that measure the operating system itself, such as time to root cause and commitments closed on time.

3. Install the weekly review. Same metrics, same format, every week. Owners explain abnormal moves before the meeting.

4. Standardize root cause work. Use one template across teams. Our root cause analysis template is a starting point.

5. Keep one action log. One owner and one due date per action, read out at the start of every review.

6. Measure the handoffs after 90 days. How long from first signal to detection, from detection to owner, from owner to fix. Automate whichever handoff is slowest.

FAQ