The leak is structural, not personal

Walk into most CNC shops in Ludhiana or Coimbatore with 20 to 200 people on the rolls and ask what a turning insert costs per component. You will get a machine hour rate, a shrug, or a number the costing engineer built from the insert price alone. The quote sheet shows one blended rate per hour. Inside that rate sits depreciation, power, floor space, operator wages and, folded in without a separate line, tooling wear. That is the standard way Indian job shops build a rate, and it is also the reason a batch that consumed three edges and a batch that consumed nine look identical on the quotation. The loss never appears as a line anyone can act on. It appears as a month-end tooling bill that is higher than last month, and a vague instruction to the supervisor to keep an eye on the inserts.

Tooling is a principal cost constituent in any machining cost model, alongside material, labour, machine overheads, energy, inspection and scrap, as the cost-modelling literature sets out (Singh et al., IJCOPE, 2026). In milling specifically, the cost per cutting edge, the insert indexing frequency and the cutter body investment are all items a cost model must carry. When those items are buried inside a single hourly rate, the shop has no way to tell a worn edge from a chipped one, and no way to price either into the next RFQ. The fix is not a new ERP. It is three numbers per insert, recorded at the machine, reviewed weekly.

Why the blended rate hides the leak

A published Indian CNC cost-per-part formula is a useful mirror here. It defines the machine hourly rate as the fully-loaded rate — machine depreciation, power, floor space, tooling wear and operator labour — not just the operator’s wage (CI Suite Pro, CNC machining cost guide). That is correct accounting for a quote. It is useless for control. Once tooling wear is inside the rate, the only way to recover an overspend is to raise the rate for every job, including the jobs that are running clean. The shop that chipped nine edges on an interrupted cut subsidises itself from the shop that ran three edges on a smooth shaft. Nobody sees it, and the insert grade gets blamed for a problem that lives in the holder runout, the coolant pressure or the incoming bar stock.

There is a second reason the leak persists. In most plants, tools are changed when the operator thinks it is time, or when parts start going out of size. Some tools are thrown away with life left in them; others run past their life and cause rejections, poor finish or breakage (TnG Smart Factory, tool life management). Both habits cost money, and both are invisible inside a blended rate. The operator is not being careless. He has no counter in front of him and no reason to record why an edge came off.

The three numbers you need per insert

Before any log format or software, settle what you are measuring. For every insert on every machine, you need three figures and one word.

1. Price per usable edge. Take the invoice price of the insert and divide by the number of edges the insert actually gives you. For an insert with four edges and no regrinds, the cost of one tool life is simply the price divided by four (MfgCalcs, cutting tool life tracker). A CNMG insert bought at ₹125 with four usable corners is ₹31.25 per edge. Write that number on the tool crib card, not the pack price.
2. Standard parts per edge. This is your own number — the count the tool-life counter on the control is set to, or the planned parts per edge on the process sheet. It is not a catalogue life. The tracker method is explicit that the standard life is the shop’s number and the workbook does not use catalogue or published tool lives (MfgCalcs). If your control has a tool-life page, read the count off it. If it does not, take the process sheet the planning engineer signed and use the planned figure.
3. Parts actually made on that edge. Counted at the machine, per tool-part pair, not estimated at the end of the shift. This is the number most shops do not have, and it is the one that turns the other two into a cost.

The word is the change reason. Group every change as Planned, Wear, Premature failure or Other. That grouping is what tells worn-out tools from broken ones (MfgCalcs). Without it, a chipped edge and a worn edge average into the same cost per component, and the average is a lie that points at the wrong fix.

Worked arithmetic, clearly hypothetical

The following example is constructed to show the method. It is not a measured result from any plant.

Take a CNMG insert at ₹125 with four edges, so ₹31.25 per edge. The process sheet sets a standard of 120 parts per edge on a mild steel shaft. Over one week the log shows the following for that tool-part pair:

• Edge 1: 118 parts, changed for Wear
• Edge 2: 121 parts, changed for Wear
• Edge 3: 84 parts, changed for Premature failure (operator noted chipping on an interrupted cut)
• Edge 4: 119 parts, changed for Wear

Total parts made: 442. Total standard parts for four edges: 480. Life achieved is the sum of parts made divided by the sum of standard lives — 442 ÷ 480 = 92.1%. Note the method: across several changes you divide the sums, you do not average the percentages (MfgCalcs). Averaging the four percentages would give 92.1% here by coincidence, but on uneven edge counts it drifts, and the sum method is the one that holds.

Cost per component on this tool-part pair is the tool spend divided by the parts made. Tool spend is four edges at ₹31.25, or ₹125. Cost per component = ₹125 ÷ 442 = ₹0.283. At standard life the same ₹125 would have covered 480 parts, or ₹0.260 per component. The gap is ₹0.023 per component on this one tool. Multiply by the annual volume on that part number and you have the tooling line that should sit in the quote. On a part running 60,000 pieces a year, that gap is ₹1,380 — small on its own, but this is one insert on one part. A shop with forty active tool-part pairs and a mix of premature failures is looking at a number worth chasing.

Two cautions on the arithmetic. First, do not add cost per component across tools as if the totals were additive in a simple way. A part cut by four tools carries all four costs, and a log that does not know which tools each part number uses cannot safely sum them (MfgCalcs). Compute per tool-part pair, then have the planner who knows the routing add the pairs deliberately. Second, the gap above is tooling spend only. It leaves out machine time, scrap and set-up, which a broken tool also costs (MfgCalcs). Do not present the tooling gap as the full cost of a premature failure. It is the floor, not the ceiling.

Read the change-reason column before you read the cost

The highest-value column in the log is the one most shops omit. If Edge 3 above had been logged simply as “worn”, the conclusion would be that the grade is short-lived on this material and the buyer would go shopping for a different insert. Because it was logged as Premature failure with a note about the interrupted cut, the correct conclusion is different: the grade is fine, and the problem is the cut, the holder runout or the coolant. That is a machine-side fix, not a purchasing decision.

This is also where the standard life gets tested honestly. After four to six weeks of logging, if life achieved sits above 100% with no wear failures, the standard can be raised. If it sits below 100% and the premature-failure share is high, the standard is not the problem — the process is. Importing a catalogue life at the start creates a variance that measures the catalogue, not your shop, and you will spend the review meeting arguing about the wrong number.

A one-page log a shift supervisor will actually fill

You do not need software to start. A paper sheet at each machine, or a single sheet at the tool crib, with five columns is enough:

1. Date
2. Machine number
3. Tool and part number (the tool-part pair)
4. Parts made on that edge
5. Change reason — Planned / Wear / Premature failure / Other, with a short note if Premature

The supervisor fills one row each time an edge is indexed or a tool is changed. The costing engineer transfers the rows to a spreadsheet once a week and computes price per edge, life achieved and cost per component for each tool-part pair. The weekly review is fifteen minutes: which pairs are below standard, what is the premature-failure share, and is the cause on the machine side or the buying side. Software that counts tool life from the machine’s recorded production, for each tool assigned to a part and machine, is a real option and removes the manual tally (TnG Smart Factory). But buying it before the change-reason discipline exists gives you accurate counts of an unclassified loss. The paper log teaches the discipline; the software scales it.

One practical note on the tool crib. Indexable inserts are replaceable tips with multiple cutting edges that can be rotated or flipped to expose a fresh edge once one becomes dull (A-Tec Corp, carbide inserts). That is the whole reason the price-per-edge arithmetic works. If the crib issues a pack of ten and nobody records how many edges went to which machine, the per-edge cost is a fiction. Issue against the machine number, not against the department.

When to change the quote

Not after one bad batch. A single batch of chipped edges is a process problem to be fixed at the machine — check the holder, the coolant, the bar stock — not a costing signal to be pushed into the next RFQ. Wait for four to six weeks of counted data across the tool-part pairs that matter, then revise the tooling line in the quote for those pairs only. The rest of the rate card stays where it is until its own data arrives. This is slower than the instinct to reprice everything after a bad month, and it is the only way the revised number survives a customer’s audit or a family review.

The structural point stands. As long as tooling wear lives inside the machine hour rate, the shop cannot see the leak, cannot price it, and cannot fix it. Three numbers per insert and one word for why the edge came off will not transform the business. They will, however, turn a month-end bill into a per-component figure you can defend across the table — and that is the difference between a quote and a guess.

Questions that come up in the review meeting

Do we need a tool-life counter on the control to start? No. The process sheet’s planned parts per edge is a valid standard. The control counter is better because it counts automatically, but the paper log works without it.

What if the operator does not know why the edge came off? Then the row says Other, with a note. An honest Other is more useful than a guessed Wear. Over a month, the Other rows tell you where the training or the inspection is missing.

Should we track every insert or just the expensive ones? Start with the tool-part pairs that carry the highest annual volume or the highest premature-failure suspicion. A ₹31 edge on a 60,000-piece part matters more than a ₹200 edge on a 400-piece job.

Can we use the insert supplier’s recommended life as the standard? You can record it, but do not set the standard from it. The standard should be your own control count or process sheet, revised only after your own logged data shows life achieved above 100% with no wear failures.

Does this connect to GST or statutory records? Not directly. Tooling purchase is an input cost recorded through your normal purchase and accounting process. The per-component figure is a management number for quoting and control, not a statutory return.