Rework Tracking: Measuring and Reducing Remakes in Stone Shops

Every fabrication shop has a corner where the mistakes go. A left-hand return that came back two inches short. An island top with a chipped sink cutout nobody wants to claim. A job where the seam opened three weeks after handover. Almost any owner can walk you to that corner and point. Very few can say what it cost last quarter, which failure modes generated that cost, or whether this month was better than the last.

That gap is why remakes get managed emotionally rather than systematically. Somebody gets shouted at, everyone tightens up for a fortnight, and the same defect returns because nothing structural changed. Rework tracking converts a heap of bad parts into a ranked list of causes with dollars attached. It needs agreed definitions, capture at the moment of failure, an honest costing rule, a few ratios you trust, and a weekly conversation aimed at the process rather than the person.

Remake, Repair or Callback: Get the Definitions Right First

Decide what you are counting before you count anything. Quality-cost accounting separates internal failure, meaning defects caught before the customer receives the work, from external failure, meaning defects the customer finds. Internal failures cost material and hours; external failures also cost freight, disruption and reputation.

A remake is a piece scrapped and produced again from new material. It consumes slab yield twice, machine hours twice and handling twice, and it lands on a schedule with no slack. Record the piece, not the job: a kitchen with one remade backsplash and a kitchen entirely recut are not the same event.

A repair leaves the original piece in the job but needs unplanned work to make it acceptable: a chip filled and repolished, a seam reopened, an edge reprofiled after a burn, a cutout enlarged on site. Repairs are cheaper and far more common, so shops tracking only remakes stay blind to most of their rework hours. Same fields, separate bucket.

A warranty callback is any defect found after sign-off: seam separation, adhesive discoloration, a loose undermount sink, staining that traces to a missed sealer. Capture the elapsed time between install and complaint, because a failure at eight days and one at eight months point at different root causes. Yield loss, meaning material written off for reasons that are not workmanship, deserves its own type.

Capture the Event at the Moment It Happens

The biggest reason rework data fails is that it gets reconstructed at month end from a stack of job folders. Memory smooths the record toward the most convenient story, small events vanish, and categories drift because whoever filled in the sheet was not at the machine. Data captured after the fact is an opinion with a date on it.

Who Logs It

Whoever discovers the defect logs it, not the supervisor and not the office. A polisher who finds a chip writes it down before starting the repair. A templater whose field dimension does not match the drawing writes it at the site. Delegating capture upward adds a filter, and filters trim whatever is politically uncomfortable.

What the Record Must Contain

Keep the list short enough to finish in under a minute: date, job number, piece identifier, material and finish, square footage affected, stage of discovery, one category from the fixed list, a line of description, the disposition, and initials. Stage of discovery gets its own field rather than being folded into the category, because the same error costs little at layout and a great deal at install.

Where the Record Lives

The medium matters less than the friction. A laminated card on a clipboard works; so does a spreadsheet on a tablet near the saw. A form in your job system works best because it links the event to the job automatically. Past roughly thirty seconds, capture gets skipped under pressure, and busy weeks generate the most rework.

A Defect Taxonomy That Fits Stone Fabrication

A good taxonomy has three properties. Categories are mutually exclusive, so every event has one home. The list is short, because thirty categories collapse into an unused dropdown. And every category maps to a controllable process step, so ranking produces a work order rather than a complaint. A bucket named other becomes a landfill within a month.

Defect category	Typical root causes	Control that prevents it
Template or field measurement error	Out-of-square walls not captured; missing appliance or plumbing dimension; template taken before cabinets were final	Field checklist with mandatory sign-off; photographs of every wall junction; template verified against the drawing before leaving site
Programming or layout error	Mirrored part; wrong hand on an L-shape; cutout placed from the wrong datum; veining ignored in nesting	Second-person review against the approved drawing; no programmer approves their own file; customer-signed layout on veined material
Material flaw	Fissure or resin void opening under load; weak vein through a narrow sink rail; mesh delamination; lot mismatch	Photograph slabs on receipt; probe suspect areas before cutting; rod narrow rails; record lot numbers against the job
Breakage in handling or transport	Narrow section lifted flat; worn suction cups; poor A-frame loading; corner impact in a doorway	Lifting rules by part geometry; scheduled vacuum lifter inspection; dedicated transport frames; two-person rule by size
Cutting or cutout dimension error	Wrong cutout template revision; tolerance stack-up across cutouts; machine offset drift	Revision control with obsolete drawings destroyed; first-article measurement each job; scheduled offset verification
Seam and adhesive failure	Contaminated bond surface; wrong adhesive for the substrate; bad mix ratio; bonding outside the manufacturer temperature range	Adhesive chart posted at the station; surface preparation standard; batch and date logged on every seam; support detail set at quoting
Polish or finish defect	Worn or wrong-grit pads; a skipped grit; insufficient water; excessive dwell causing burn	Documented pad sequence per material; pad changes tracked by interval, not by eye; water flow checked at shift start; raking-light inspection
Damage on site during installation	Overhang unsupported during set; cabinets not level; forcing a tight fit; other trades over finished tops	Documented cabinet go or no-go; support blocking standard; protective covering at handover; installer stop-work authority
Specification or communication error	Wrong edge profile or finish; wrong sink model; a change agreed verbally; an outdated revision released	One written change-order path, no verbal exceptions; profile and finish confirmed before cutting; old job packets pulled on every change

The control column is dominated by verification steps and written standards rather than requests to be careful. Telling people to concentrate harder decays as attention moves on, while a checklist that must be signed keeps working when the shop is busy and tired.

Costing a Remake Honestly

A cost figure that counts only the slab is a comfortable fiction. It understates the event, and because it understates unevenly it distorts the ranking and hides categories whose damage sits in labour and schedule. Write one formula and apply it mechanically. Consistency beats precision: a rule that is slightly wrong but applied identically still ranks causes correctly.

Start with material. Charge the square footage actually consumed by the replacement part rather than the net area of the finished piece, since the offcut is rarely usable, and value it at replacement cost. If the remake forces a slab from a different lot and the colour no longer matches, the cost is the whole slab.

Add labour at a loaded rate covering wages, burden and overhead, multiplied by the hours consumed at every station the part passes a second time. Include the hours that produced the scrapped piece; those were paid for and delivered nothing. On cheap material with a simple profile, that sum often exceeds the slab itself.

Add logistics next: freight for a replacement slab, fuel and vehicle time for a return trip, rigging, and an install crew standing idle or returning on a separate day. Then price the schedule, since a remake consumes bottleneck hours that were sold to somebody else. Charge the contribution margin of the displaced work, add any customer concession, and the result hurts enough to make prevention spending easy to justify.

Rework Rate, Denominators and Trends

One ratio is never enough and four is too many. Remakes as a percentage of jobs completed is the figure a customer would recognise. Remade square footage as a percentage of square footage produced normalises for job size, so a run of small vanities cannot flatter the numbers. Total rework cost as a percentage of revenue belongs in the management report.

The two volume ratios usually diverge, and the direction is diagnostic. A high job-based rate with a low area-based rate means failures cluster on small parts, pointing at handling, splashes and narrow rails. The reverse means a few large pieces are failing, implicating programming, layout or seam decisions. The gap often locates the problem faster than either ratio alone.

Be strict about the denominator: count jobs when they are invoiced, use the same rule every period, and print the rule on the report. Then read the numbers as a trend rather than a verdict, since one bad week can double a small shop's monthly figure. Plot a rolling three-month average and react when the trend moves.

Pareto Ranking and the Vital Few

Joseph Juran popularised the observation, drawn from Vilfredo Pareto's work on income distribution, that most quality problems trace to a small minority of causes, a pattern he named the vital few and the trivial many. The practical value is the instruction it carries: rank your causes, work the top of the list, and decline to spend effort on the tail until the head is fixed.

Build the chart from cost, not event count. Counting events treats a chipped splash and a recut island as equivalent, so count-based charts float the cheapest, most visible failures to the top. Sort categories by total cost, plot them descending with a cumulative line, and let the leading two or three become the improvement agenda for the quarter.

Then drill one level deeper inside the leading category. If seam and adhesive failure heads the list, split those events by material, installer, adhesive batch, shop temperature at bonding, and whether the seam had structural support beneath it. A second Pareto inside the first is where the actionable cause usually appears, converting a category name into a condition you can change.

Running a Weekly Review That Does Not Become a Blame Session

Set a fixed half-hour slot each week with the shop lead, the programmer, a fabricator, an installer and whoever owns scheduling. A standing agenda keeps it short: the week's events by category, the running Pareto, progress on open actions, and one category for deeper discussion.

Enforce two ground rules from the first meeting. Events are described in terms of process and conditions, never in terms of who ran the machine, and anyone may report a defect they caused with no consequence attaching to the report. The moment a log entry becomes disciplinary evidence, logging stops and you will not get it back for a long time.

Every discussion ends with a written entry: the change, a named owner, a date, and the measure that will show whether it worked. Two or three actions per week is sustainable; ten is a wish list abandoned by the third week. Read open actions aloud at the start of each meeting, because visible follow-through is what convinces the floor this is real.

Corrective Action That Actually Sticks

Separate containment from correction. Containment stops the bleeding now: quarantine the suspect slabs, add a temporary inspection, call the customers whose jobs used the same adhesive batch. Correction removes the mechanism. A shop that only ever contains accumulates inspection steps forever and never gets cheaper, because inspection scales with volume while prevention is mostly a one-time investment.

Rank candidate corrections by durability rather than ease. Strongest is eliminating the step that fails, such as moving bonding into a controlled area. Next is mistake-proofing: a fixture, a file naming rule or a software constraint that makes the wrong action impossible. Then standard work with a signed verification, then training. Weakest is a sign asking for more care, which should never be the only response to a costly category.

Close the loop with a scheduled recheck. Sixty to ninety days after a corrective action, pull the same category and compare count and cost against the period before the change. If nothing moved, the diagnosis was wrong, and the honest response is to reopen it.

How Remake Data Changes Purchasing and Training

Once material flaw and breakage events are attributed to specific suppliers and lots, purchasing conversations change character. You stop arguing about price per square foot in the abstract and present a documented record of the yield loss that material produced. The same logic applies to adhesives, blades, profiling wheels and pads.

Training becomes targeted for the same reason. Rather than a general session on quality, the log shows that seam preparation on one engineered material drives a large share of repair hours, clustered around a process step rather than a person. That produces a focused session with a written standard attached.

Rework data also improves quoting, since job types that generate repeated remakes are usually underpriced for their real complexity. Note too that rework means extra cutting, grinding and polishing, and therefore extra exposure. OSHA sets a permissible exposure limit for respirable crystalline silica of 50 micrograms per cubic meter as an eight-hour time-weighted average, with an action level of 25 micrograms per cubic meter, so unplanned dry touch-up carries a compliance dimension too.

The controls in the table depend on having the right consumables in the building when the standard calls for them, so pair the tracking system with a stocking policy. Review the full range of fabrication and installation supplies at Dynamic Stone Tools, or browse blades, profiling wheels, pads, adhesives and handling equipment together in the complete product catalog when rebuilding a station kit around a newly written standard.