Line Balancing: Size the Gain Before You Commit to the Change

Line Balancing does one job. It balances process step cycle times to takt time to minimise idle time and bottlenecks. You give it the steps in your process, the time each one takes, and the rate your customer pulls work at. It tells you where the work piles up and how many stations the work actually needs.

The tool sits in the Improve phase of RDMAICS. That placement matters. By the time you open it, you should already know what is slow and why. Measure gave you the cycle times. Analyse told you which step drives the delay. Improve is where you decide what to do about it, and Line Balancing is where you size that decision before you spend anything.

Use it when your process runs as a sequence of steps with a handover between them. Use it when one step keeps the others waiting. Use it when someone proposes adding headcount and you want to know whether the work content justifies it. Do not use it to find the problem. Use it to shape the response.

Everything that follows uses the sample data built into the tool. The case is fictional. It is there so you can see the mechanics without needing your own data loaded first.

The process is an invoice handling flow with six steps. Work arrives, it gets captured, checked, approved, paid and filed. The six steps are Auto-ingest, Auto-populate, 3-way match, Tiered approval, Payment and Archive.

The diagnosed cause, carried in from Analyse, is Tiered approval. It is the step people complain about. It is the step where work sits. The question for Improve is not whether approval is slow. The question is what a change to approval would be worth, and whether the rest of the process could even absorb the gain.

The worksheet is deliberately small. Each row is one process step. Each row carries a cycle time in minutes and a takt time in minutes. The tool then calculates stations needed and a balance status for that row.

In the sample, takt time is set to 5 minutes per unit across every row. That is the reference rate. It represents how often a finished unit needs to leave the process to meet demand. Keep it constant across rows unless you have a genuine reason to vary it.

The cycle times entered are Auto-ingest at 2 minutes, Auto-populate at 1 minute, 3-way match at 1 minute, Tiered approval at 5 minutes, Payment at 5 minutes and Archive at 1 minute. Each step shows 1 station needed.

The balance status column is where the worksheet starts talking back to you. Auto-ingest reads Balanced. Auto-populate, 3-way match and Archive all read Under-loaded. Payment reads Balanced. Tiered approval reads Balanced (bottleneck). That one label is the bridge between what Analyse told you and what Improve has to decide.

The result panel converts the six rows into four numbers and a chart. Total work content is 15 minutes. That is the sum of every cycle time in the process. Reference takt is 5 minutes per unit. Minimum stations is 3. Steps above takt is 0.

Minimum stations of 3 comes from the relationship between total work content and takt. Fifteen minutes of work, divided across a 5 minute takt, needs three stations' worth of capacity to keep pace. The process currently shows six steps each with one station. That gap between six steps and three stations of required capacity is the whole opportunity.

Steps above takt reading 0 is the second finding, and it is the one that changes the conversation. No single step exceeds takt. Tiered approval and Payment both sit at 5 minutes, exactly on takt. They are tight, not overloaded. That distinction matters when you decide what to pilot.

The chart gives you the Yamazumi-style view. Each step appears as a bar against the takt line. You can see at a glance that Auto-populate, 3-way match and Archive sit well below the line at 1 minute each, while Tiered approval and Payment touch it. The tool also offers an AI review that explains the result in plain language if you want a second read.

The first decision is whether to attack the bottleneck or the balance. Tiered approval is the diagnosed cause and the tool confirms the label. But steps above takt is 0. Nothing is breaching the rate. So the gain from speeding up approval alone is capped. Approval at 5 minutes is already running at takt.

The second decision is consolidation. Three steps sit at 1 minute each against a 5 minute takt. Auto-populate, 3-way match and Archive together total 3 minutes. The minimum stations figure of 3 is the tool telling you the work content can be carried by fewer stations than the current six steps imply. Combining under-loaded steps is a change you can pilot on one team without touching the approval policy.

The third decision is sequencing. You now have two candidate changes and a way to rank them. Redistributing under-loaded work is low risk and internally controlled. Changing tiered approval usually involves policy, delegation limits and people outside the process. Pilot the low risk one first and measure whether the 15 minutes of total work content moves.

That is how a diagnosed cause becomes a change worth piloting. You do not jump from a named bottleneck to a business case. You pass it through a balance check, see what the numbers permit, and pick the change that the data actually supports.

Most errors happen in the worksheet, not the result. The tool calculates faithfully from what you give it. If the inputs are rough, the minimum stations figure will be rough too.

Watch for these patterns when you load your own process.

Each of these quietly distorts total work content, and total work content drives the minimum stations number you will quote to a sponsor.

The reason to run this tool before a pilot is simple. It gives you a defensible number to put against a proposal. In the sample case that number is the distance between six steps with one station each and a minimum of 3 stations for 15 minutes of work content at a 5 minute takt.

That is enough to scope a pilot. It is not enough to promise a saving. The tool shows required capacity. It does not show what you will choose to do with the freed capacity, and it does not show whether the change will hold once people return to habit.

Run the worksheet again after the pilot with fresh measured cycle times. If total work content falls and minimum stations falls with it, the change is real. If the figures hold steady, the change did not land, and you have learnt that cheaply.

Line Balancing works on time and sequence. It does not work on quality, risk or cost. It will tell you that three steps are under-loaded. It will not tell you whether the 3-way match exists for a control reason that makes combining it unwise. That judgement stays with you and your risk owners.

It does not account for variation. Every cycle time in the worksheet is a single figure. Real steps vary by case, by day and by person. A step that averages 1 minute may occasionally take 10. The tool treats the number you enter as the number, so a balanced-looking process can still queue in practice.

It does not model demand change. The reference takt of 5 minutes per unit is a fixed input. If volume rises, takt shortens and the whole picture shifts. Rerun the tool when demand moves rather than trusting an old result.

Finally, it does not make the change happen. Minimum stations of 3 is a calculation, not a plan. You still need the pilot design, the people conversation and the Control phase measures that stop the process drifting back. The tool sizes the gain. You still have to go and get it.