Applied: Redesigning a Process to Double Throughput
Doubling throughput rarely starts with buying another machine or hiring twice the people. In a busy lab, kitchen, warehouse or loan-processing team, the real constraint is often one slow approval, one setup delay, one rework loop, or one queue that nobody measures.
- Throughput is the rate at which good output is completed - units per hour, orders per day, claims per shift.
- To double throughput, do not improve every step equally. Find and redesign the bottleneck, because the bottleneck sets system capacity.
- The practical sequence is: map the process, measure cycle times, identify the constraint, redesign the constraint, rebalance, then stabilise with controls.
- Use Little's Law: WIP = Throughput x Lead Time. If WIP explodes, your throughput gain may just be hidden waiting.
- The fastest levers are usually waste removal, parallel capacity at the bottleneck, work transfer, setup reduction, and quality-at-source.
- A good redesign doubles sustainable good throughput, not just one-day output with overtime, defects, burnout or excess inventory.
- In interviews, always quantify before recommending: current throughput, bottleneck capacity, target takt, investment needed, risks and control metrics.
Big Picture: Throughput Doubles When the Constraint Moves
Think of a process as a chain of workstations. The whole chain can move only as fast as its slowest effective step. The job of process redesign is not to “make everyone faster”; it is to increase the capacity of the constraint, then check whether a new constraint appears.
Core Explanation: The Bottleneck-First Way to Double Throughput
Start with the unit of flow: a patient, order, application, parcel, table, ticket or SKU. Then define throughput as completed good units per time period. “Good” matters - defective output, rework and partial completions do not count as real throughput.
If you need a clean prerequisite, revise process mapping and value stream mapping before attempting redesign. Without a map, candidates usually optimise a visible step instead of the true constraint.
The Five-Step Redesign Process
This is where finding the bottleneck and the Theory of Constraints becomes practical. The bottleneck is not always the step with the longest task time; it is the step whose effective capacity limits the whole system.
Worked Example: Doubling a Back-Office Approval Process
Suppose a fintech operations team processes customer applications. Demand is rising, and leadership wants output to double from 25 completed applications per hour to 50 per hour.
The current system cannot exceed 25 applications per hour because document verification is the constraint. To reach 50 per hour, the redesign must lift document verification from 25 to at least 50 per hour. There are two clean options:
- Parallel capacity: add a second verification lane, each capable of 25 per hour, with a common queue and standard checklist.
- Cycle-time reduction: reduce verification time from 2.4 minutes to 1.2 minutes using pre-validation, better forms and exception-based checking.
But the redesign is not complete there. Once verification reaches 50 per hour, risk approval also sits at 50 per hour. Any variability, absenteeism or rework at risk approval will immediately create queues. This is why process redesign must include line balancing and workstation design, not just one bottleneck fix.
Metrics to Track: Prove the Throughput Gain Is Real
A serious answer uses metrics. The best candidates do not say “we improved productivity”; they show whether the process can sustain the new rate without creating hidden queues, defects or burnout.
Use Little's Law and reading a process mathematically when the interviewer gives WIP and lead time. If throughput appears to double but WIP triples, the process may simply be pushing more work into queues.
Definitions: Say These in One Breath
Throughput: the rate at which a process completes good output in a defined time period.
Bottleneck: the resource or step whose effective capacity limits total process output.
Cycle time: the time required to complete one unit at a process step.
Takt time: available production time divided by customer demand in the same period.
WIP: work that has entered the process but has not yet been completed.
The interview-safe logic is simple: if takt time is lower than bottleneck cycle time, the process cannot meet demand. If you need a refresher, revise cycle time, takt time and lead time.
Case Study: Aravind Eye Care's High-Throughput Cataract Process
Aravind Eye Care shows how a service process can achieve high throughput by designing flow, roles, standard work and quality together.
Aravind Eye Care is a powerful Indian example because the “product” is not a manufactured part - it is a medical service where safety, trust and speed must coexist. The operating model is often studied because it treats cataract surgery as a carefully designed flow system rather than a random sequence of appointments, waiting and surgery.
Situation: Eye care demand in India is large, and many patients need affordable, reliable cataract treatment. A conventional hospital process can leave expensive surgeons waiting for patients, instruments, rooms or paperwork.
The move: Aravind redesigned the process around the scarce constraint - skilled surgeon time. The primary driver is role specialisation around the surgeon: support teams prepare patients, instruments and rooms so the surgeon's time is focused on the procedure. Supporting drivers include standardised protocols, high-volume scheduling, trained paramedical staff, repeatable room preparation and disciplined patient flow.
Outcome and lesson: The lesson is not “work faster.” It is “protect the constraint.” In a high-throughput service process, the scarce expert should not be waiting, searching, switching context or doing tasks that trained support staff can safely do. That is exactly the thinking you need when asked to double throughput in a bank branch, diagnostic lab, restaurant kitchen or fulfilment operation.
So what: A throughput redesign wins when the primary bottleneck is protected and the supporting system is redesigned around it. One-factor explanations fail here - volume comes from constraint focus, standard work, trained support roles, scheduling discipline and quality control working together.
How AI Changes Redesigning a Process to Double Throughput
AI does not replace process thinking. It makes the measurement, testing and control loop faster.
- AI-assisted time study: teams can use computer vision or timestamp logs to estimate wait time, service time, handoff delay and rework frequency more continuously than a one-day manual observation.
- Simulation before implementation: managers can test whether adding one verifier, changing batch size or altering shift patterns actually increases throughput before disturbing live operations. For the deeper method, revise using AI and simulation to test a process design.
- Exception prediction: ML models can predict which orders, claims or applications are likely to fail checks, allowing normal cases to flow faster and exceptions to be routed to specialists.
Use ChatGPT or Claude to practise: paste a process map with cycle times, yields and WIP, then ask, “Find the bottleneck, propose two redesigns to double throughput, list risks, and create interview-style follow-up questions.” Then verify the maths yourself.
Interview Relevance
“A quick-commerce dark store currently ships 600 orders per shift. The CEO wants 1,200 without doubling space. How would you redesign the process?”
Use the phrase “effective capacity,” not just “capacity.” Effective capacity accounts for downtime, defects, changeovers, absenteeism and variability - exactly where real operations lose throughput.
Common Mistake
The biggest mistake is recommending “add more people or machines” before proving the bottleneck. It costs candidates because it sounds expensive, generic and operationally naive. The one-line fix: measure step capacity first, redesign the constraint second, and only then justify added capacity.