Register Says 85%, the Month Says 60%: Closing the Efficiency Gap

When the shift register says 85% and the month's dispatches say 60%, both numbers can be honest: the register divides what the line, machine or packer says it made by the minutes someone chose to count, while the month divides what left the gate by every minute you paid for. In garment units, machine shops and packing halls alike, the gap comes from the same seven places, from shift edges and excluded downtime to rework counted twice and loose standard times. Sorted into real loss, timing and costing error, the gap tells you how much of it is cash you can recover and how much is a quoting mistake waiting to happen.

By the Mama Editorial Team. We scope camera and floor-visibility projects for Indian factories. Last verified: 13 October 2026. All ₹ figures below are illustrative worked examples built on stated assumptions. They are not benchmarks.

Key takeaways

• The two numbers use different numerators and denominators. The register counts pieces offered at line end against the minutes it booked. The month counts good pieces dispatched against all paid minutes.
• A gap between the good days and the month is common. Writing in July 2012, IE consultant Prasanta Sarkar reported that across 10 NCR garment factories he had observed the previous year, lines worked "up to 65% efficiency" on a given day, but monthly line efficiency went "below 40%". That is one consultant's old, garment-only sample, not a benchmark.
• Sort each gap into one of three types: real loss (recover it), timing (it will show up next month), and costing error (fix the standard minutes before your next quote).
• Reconcile weekly from attendance, dispatch and stock data you already keep.
• Measure the line, not the person. Every gap here is a flow or system problem.

Two numbers, two definitions

Register (or MIS) efficiency is usually earned standard minutes ÷ minutes on the line. "Earned" means pieces counted at line end × the SAM on the costing sheet. "Minutes on the line" is often the minutes left after logged downtime, off-standard work and anyone "not on the line" have been taken out. Apparel industrial engineers call this on-standard efficiency. The same source notes it is good for operator skill analysis, because it doesn't blame an operator for a line that had nothing to sew. Its counterpart, overall efficiency, keeps the lost time in the denominator.

Dispatch (or overall) efficiency is good pieces that left the gate × a realistic SAM ÷ every paid minute of direct labour. This is the number your P&L feels.

The same split exists outside garments. In a machine shop the shift log says "machine ran 7 h 30 min" and nobody records the forty two-minute stops in between. OEE.com, a reference site run by the monitoring-equipment maker Vorne, says minor stops usually last well under five minutes and that "most companies do not accurately track" them. Evocon, an OEE software vendor, describes the top three OEE "blindspots" it sees among its own customers: too many stops classified as "planned" and dropped from the calculation, optimal cycle speeds underestimated so performance shows above 100%, and quality assumed to be 100%. Both are vendors, neither offers India-specific data, but the three blindspots map closely onto the gaps below.

The seven gaps, in the order you should check them

1. Shift edges. The register starts at "line start", but payroll starts at punch-in. Warm-up, tea and bundle hunting at each end appear nowhere. We cover this in detail in late start and early wash-up.

2. Logged downtime that disappears. A supervisor honestly writes "no feeding 45 min" or "power cut 1 h". The MIS then takes those minutes out of the denominator, because it was "not the line's fault". Fair for judging the operator, but the plant still paid for those minutes.

3. People outside the count. Floaters covering absentees, helpers, and operators sent to off-standard jobs are often missing from the line headcount, yet they are on the muster roll. See what floaters cost on a sewing line.

4. Made but not dispatched. A piece counted at line end may still be in finishing, packing or the godown at month-end. That is timing, not loss, and it reverses next month; separate it or you will chase a ghost.

5. Rejects counted as output. The line-end counter records a piece before final inspection. If it is later rejected, the register has already credited it.

6. Rework counted twice. A piece goes back for alteration, is re-offered, and gets counted again. See what rework really costs.

7. Loose standards. The SAM on the costing sheet, or the cycle time in the machine plan, is often older and looser than the real work content. Every piece then "earns" more minutes than it took. The register looks good, and you overestimate capacity and underestimate cost per piece.

A worked ₹ example: the bridge from 85% to 60%

Assume a garment unit with 300 direct sewing operators, 480 paid minutes a day and 26 working days. That is 3,744,000 paid minutes a month. Assume an all-in labour cost of ₹18,000 per operator per month, covering wages, PF/ESI and bonus. This is an illustrative figure. Each state notifies its own minimum wages by trade and skill level, and most revise the DA part twice a year, so use your state's current notification. That gives ₹1.44 per paid minute, and one efficiency point = 37,440 minutes ≈ ₹54,000 a month.

The register averages 85%. Dispatch-based efficiency works out to about 60%. Here is the bridge, applying one correction at a time (all inputs are assumptions):

Read the table this way:

• Real, recoverable loss: about 16 points, roughly ₹8.8 lakh a month of paid time that produced nothing that shipped. Steps 1–3 are flow problems; 5–6 are quality.
• Timing: about 2.2 points. Check that it reverses next month. If WIP keeps growing, it is cash locked up on the floor, not timing.
• Costing error: about 7 points. You don't recover it by working harder. You recover it by re-timing the styles and re-pricing.

Steps 1–3 assume the register's denominator was 85.8% of paid minutes (paid minutes minus 20 minutes a day, 7% downtime and 3% floater time); steps 4–7 scale the numerator. Pricing each step at ₹54,000 a point is a rough conversion. Your own split will differ: this table is arithmetic on assumptions, not survey data.

The same gap in other plants

The mechanics hold wherever a register is written by hand. Per-point values below use assumed inputs, so substitute your own.

On packing lines, the micro-stop gap is usually the largest one. It is the subject of FFS packing machine micro-stops.

How to reconcile register vs dispatch every week

You need four numbers you already have, on one sheet:

1. Paid direct minutes from attendance and payroll, not from the line register.
2. Good pieces dispatched from invoices or gate passes.
3. Change in stock of WIP and finished goods, counted on the same day each week and valued in standard minutes.
4. One standard per product, set by a fresh time study, not the costing sheet from three seasons ago.

Then: dispatch-based efficiency = (dispatched + Δstock) × fresh SAM ÷ paid minutes. Set the register average beside it, work through the seven steps, and see what's left unexplained. Anything over two or three points that you can't explain is a measurement problem in its own right.

Then make the fix part of the daily shift report: one line for "minutes paid but not on the register", so the gap is visible every morning instead of once a month.

What cameras can and can't tell you here

What cameras can see: whether a machine or station is running or stopped, minute by minute, including the short stops nobody logs. When the line actually starts and stops at each end of the shift. Whether a line is starved, with work missing between stations. Piles of WIP building up. Aggregate headcount at stations compared with the muster roll. That covers most of gaps 1, 2, 3 and part of 4.

What cameras can't see:

• Whether your SAM is right, which needs a time study.
• Whether a piece passed inspection, at least not reliably from a general floor view.
• What was invoiced.
• What anyone is paid.

Gaps 5, 6 and 7 come from your QC, dispatch and IE records, not from video. Cameras show where a line was starved, not why, and racks, poor lighting or a badly aimed lens leave holes.

And a point on dignity. All of this is about flow. "Line 4 sat idle for 50 minutes after lunch" is a feeding or planning problem, not a verdict on the operators on line 4. Keep counts aggregate, rank no individuals, and tell workers what is recorded and why. See the DPDP worker notice.

Where Mama fits

Mama reads your existing cameras, or cameras it adds, and every morning sends the owner a short note in plain language: which machines and lines stopped yesterday, for how long, and how the running time compares with what the register claims, valued at your own cost per minute. It watches machines and flow, not named people. Mama can draw a camera plan for the gaps your register misses. Leave an email or phone in the form below.

FAQ

Is 85% register efficiency with 60% dispatch efficiency a sign that supervisors are fudging? Usually not. On-standard efficiency is designed to leave out lost and off-standard time, so it runs higher than overall efficiency. The problem comes when the owner reads it as if it were the dispatch number.

Which number should I use for quoting and capacity planning? Use the dispatch-based figure with a freshly timed SAM. Use the register figure to coach lines and judge operator skill.

How big a gap is normal? We found no national Indian survey, and no IIM, NPC or CII study, that measures this gap. The closest published observation is Prasanta Sarkar's 2012 note on 10 NCR garment factories: lines reached up to 65% on a given day, but monthly sewing efficiency fell below 40% (his range across the factories was 30–45%, with one at 60%). He blamed long line-setting times, peaks and troughs in orders, and planning problems, not bookkeeping. Measure your own bridge.

Does work-in-progress make the month look worse than it really is? Yes, if stock grew during the month, which is why the bridge adds the change in WIP and finished goods. If it grows every month, that is cash locked up on the floor.

Can OEE software or cameras close the gap automatically? They can close the time-based gaps: shift edges, unlogged short stops and excluded downtime. They can't fix loose standards or rework counting. Those need a time study and a single, agreed counting point after inspection.