The Multi-Skilling Matrix: What to Do When an Operator Doesn't Show Up (EMS Lines)

On an EMS line, an absent operator costs you almost nothing if the gap is at the packing table and a whole shift if it is at the hand-solder or test station. A multi-skilling matrix is the sheet that tells your supervisor, at 7:15 a.m., which absence is which and who can fill it. Built properly, it turns a lost shift into a reshuffle.

By the Mama Editorial Team — we scope camera projects for Indian factories. Last verified: 6 October 2026, against the Uttar Pradesh minimum-wage order of 17 April 2026, the OSH Code, 2020, ESD Association program guidance, and the Shahi Exports absenteeism study cited inline.

Key takeaways

• Absence is normal, not an exception. In a study of four Indian garment factories (apparel, not electronics), 10–11% of workers were absent on a typical day, almost all of it unauthorised. Plan for it daily, not as a crisis.
• Not every station is equal. On an EMS line, a few stations (certified hand-solder, functional test, AOI review) need a qualified person; most others can be filled in minutes. The matrix should rank stations by how hard they are to fill.
• "Just send someone from packing" can break your own rules. ESD control programmes typically require training before anyone handles sensitive parts, and certified-solder stations often require a valid IPC certificate — which, per an IPC-authorised training centre, is valid for two years.
• Overtime is the expensive fix. Under Section 27 of the OSH Code, overtime is paid at twice the rate of wages and needs the worker's consent. A trained floater is usually cheaper than covering gaps with overtime or a stalled line.
• Cameras see the gap, not the skill. Video can show an unmanned station or WIP piling up behind it. It cannot tell you who is certified, and it should never be used to rank individuals.

Why absence hurts an EMS line more than you think

The best Indian evidence comes from apparel, not electronics, but it is the closest we have for labour-intensive assembly. In worker-level data from four Shahi Exports garment factories, supervisors absorbed absence by reshuffling until it passed about 10%; beyond that, line efficiency fell, because they ran out of people who could step in. Our sewing-line absenteeism piece works through that study and its ₹ consequences in detail.

An EMS line is less forgiving than a sewing line, for two reasons:

1. Some stations need a qualification, not just a pair of hands. A substitute at the touch-up bench without solder training may produce boards that pass the bench and fail at the customer.
2. The line rate is set by the slowest station. If the gap falls at the bottleneck, every operator downstream waits. We covered the arithmetic in UPH, UPPH and line balancing. An absence is simply line imbalance that arrives without warning.

So the question is not "how many are absent?" but "which stations are empty, and who is qualified to fill them?" That is what the matrix answers.

What a multi-skilling matrix actually is

A skill matrix is a grid: operators down the side, stations across the top, and a skill level in each cell. Gemba Academy describes it as a visual tool that shows "at a glance how much cross-training you have" between people and tasks. Plants use anything from two to seven levels. Four levels are enough for most EMS lines:

On an EMS line, add two columns that a garment matrix doesn't need:

• ESD training: date completed / next due. The ESD Association's sample control-programme document based on ANSI/ESD S20.20 makes initial ESD training a condition before an employee handles any ESD-sensitive device, and has untrained employees in the protected area escorted by ESD-certified staff. Your own programme sets the exact rule, but under one like this an untrained person cannot fill a gap inside the EPA, however willing.
• Certificate expiry, where your customer requires one. An IPC-authorised training centre states that J-STD-001 certification is valid for two years from the exam date, with renewal in the six months before expiry. If your customer requires certified operators at hand-solder or rework, a substitute whose certificate lapsed last month does not count. Check your customer's quality agreement for the exact rule.

Tier your stations before you need to

First rank stations by how hard they are to fill.

"Three per tier-A station" is our rule of thumb, not an industry standard: with one person absent and one on leave, a third is still there. Set your own number, but set it per tier.

Then read each operator's row. Someone at level 2 on four tier-C stations gives you less cover than someone at level 2 on one tier-A station. Train people up into tiers A and B, not further across tier C. And update the sheet the day a skill changes — a laminated March matrix is wrong by June.

The 7:15 a.m. fill order

When attendance comes in short, the supervisor needs a fixed sequence, not a debate:

1. Fill tier A first, from the matrix. Move a level 2 or 3 person to the empty tier-A station, even if that opens a gap at their own tier-C station.
2. Collapse tier-C stations. Two packing or labelling stations can often be run by one person at a slightly lower rate.
3. Borrow from the neighbouring line, and keep it fair. In the Shahi study, managers borrowed mostly from adjacent lines through a handful of informal partnerships. A matrix that spans neighbouring lines makes borrowing systematic rather than personal.
4. Use the floater pool. Floaters are trained operators kept off fixed stations for exactly this purpose (see the worked example below).
5. Overtime last. Under Section 27 of the OSH Code, overtime is paid at twice the rate of wages, needs the worker's consent and stays within the cap set by the appropriate (usually state) government. It works for a short gap, not as a staffing plan.
6. If tier A cannot be filled, change the plan, not the rule. Run the jobs that do not need that station, or move the tier-A work to a later shift. Do not put a level-1 operator alone on a certified station to protect the day's count.

Worked example: is a floater cheaper than the gap?

Take a 20-operator manual and box-build line in Noida, single 8-hour shift, 26 working days a month. Wages are the Uttar Pradesh Category-I (Gautam Buddha Nagar and Ghaziabad) minimum rates from order No. 374/36-2-2026 of 17 April 2026, an interim-relief order effective 1 April 2026: ₹15,059/month semi-skilled (₹14,173 basic + ₹886 VDA) and ₹16,868/month skilled. The VDA is revised periodically, so check your current rate. These are minimums before PF, ESI and bonus — treat them as a floor.

Assumptions (replace them with your own): all 20 operators are paid at the semi-skilled rate; absence runs at 10% (borrowed from the garment study above; measure your own), and each operator is absent independently. We cost all 20 wages, though an absent worker is usually unpaid, which overstates the gap by about 5%. The line has 3 tier-A stations. When a tier-A station is filled by a level-1 substitute, the line runs at 70% of rate for the shift, because that station is the bottleneck.

On these assumptions a floater costs less than the gaps, and the table counts wages only — not late dispatch, lost margin, or rework from a level-1 substitute at a solder station. The overtime row looks cheaper, but it assumes a qualified person works a double shift seven times a month, which is neither realistic nor within normal hour limits.

Real absence is not random: it clusters around festivals, so some days lose two or three tier-A stations at once. That is when a floater who is level 2 on all three tier-A stations earns their salary. On full-attendance days, the floater covers breaks at the bottleneck and trains level-1 operators. Every operator who quits also resets part of the matrix; what each new hire really costs prices that.

What cameras can and cannot tell you here

What a camera can show. It can show which stations were unmanned and for how long — as a count per station, not a list of names. It can show WIP piling up after a substitution — a new bottleneck seen at 9 a.m., not at evening dispatch. Over a month, it shows how much slower the line runs on short-staffed days: the number that justifies, or rules out, a floater. This is the same blind-zone measurement we describe in where a mid-size EMS line loses hours.

What a camera cannot show. It cannot tell you who is certified, who is level 2, or why someone did not come in. It cannot judge a solder joint; that is what AOI, SPI and your test jig are for, as covered in cameras in EMS plants. And it is the wrong tool for attendance by name — your muster roll does that; aggregate station counts are more useful and easier to defend. If cameras cover the line at all, give workers the notice the DPDP framework expects (template here).

Where Mama fits

Mama reads the cameras you already have (or a few we add). Each morning it sends you a short WhatsApp note saying where yesterday's hours went: which stations sat empty, where WIP piled up, and how output on short-staffed hours compared with normal ones. It watches machines, stations and flow, counts people only in aggregate, and never names or ranks anyone. The matrix stays your supervisor's tool; Mama tells you whether it held up. To start, send a short phone video of your floor and we will send back a camera plan.

FAQ

How many people should be trained on each station? There is no statutory number. Our rule of thumb is at least three people at independent level per shift for qualified (tier-A) stations, two for judgement stations, and anyone ESD-trained for standard-work stations. Set the target by how hard the station is to fill, not one number for the whole line.

Can I put a contract worker on a certified-solder station for a day? Only if they hold the training or certification that your quality system and customer agreement require, and have completed ESD training. If they don't, run other jobs or move the work to another shift. A day's count is not worth a field failure.

Does high absenteeism always hurt output? Not linearly. In the garment study above, effects were small until absence reached about 10%. A matrix helps you absorb bad days, though it cannot stop people being absent.

Should we use cameras or face recognition to track who is absent? No. Attendance by name belongs in your muster roll and access system. Use cameras for station-level counts (manned or empty, WIP building or not), and give workers a clear notice under DPDP.