Customer Disputes a Crack After Cure Response

Purpose

The customer says new slab is already cracking translation tree in this library tells you how to classify a crack once you are standing over it: on a joint, mid-panel, a corner, or structural. What it does not cover is the shop's own standing process for getting from a customer's phone call to that classification, correctly, every time, whether the person who takes the call is the owner or the newest hire in the office. A dispute mishandled at intake, no data taken, a verbal promise made before anyone measured anything, no pour records pulled, turns a five minute reassurance into a warranty fight regardless of how sound the classification method itself is.

Scope

Covers intake, evidence gathering, site visit, liability determination, and resolution for a customer-reported crack or defect on a completed, cured pour, for residential and light commercial flatwork, from the first call through record closure.

It does not cover the technical severity classification itself, owned by the crack width and pattern and hairline crack condition trees; the physical repair technique, owned by the crack injection and patch repair technique or panel replacement practice; or damage a crew discovers on its own during the cure window rather than reported later by a customer, owned by the early traffic damage to a fresh pour response.

Roles and responsibilities

Procedure

1. Take the call to a stated response-time standard, not an open-ended one. Get the three phone or photo data points the classification tree uses, crack location relative to any joint, a width read against a credit-card edge, and whether a step or flush face is felt, then assign urgency and schedule a visit inside a stated window: two business days for a cosmetic-sounding report, same day for anything describing a step, rocking panel, or safety concern. Acceptance: the three data points captured, urgency tier assigned, visit scheduled inside the stated window. Wrong looks like scheduling "whenever someone's free" with nothing captured first; stop rule, no visit goes on the calendar until at least an attempt at all three data points is made. Hazard: none, a phone call.

2. Pull the job's own pour records before the visit, not after. Retrieve the joint layout record, delivery log, finishing log, and cure log for that specific pour date and area, reading each for anything already relevant, a re-cut joint, a rejected load, a logged intrusion. Acceptance: records located and reviewed, or their absence explicitly noted, before the tech leaves the shop. Wrong looks like showing up with no history and relying on memory of the job; stop rule, pull whatever exists first, since a joint layout or cure record settles half these calls before a tape measure comes out. Hazard: none, a records review.

3. Classify the crack on site using the shop's stated method, and measure every point the method requires. Apply the crack width and pattern or hairline condition tree, whichever fits, taking the comparator reading at multiple points along the full length rather than only at the point that looks widest, and check for a step at each of those points, not just the obvious one. Acceptance: a classification reached with a measurement recorded at every point the method calls for, photographed at scale, and a monitor set where the disposition is to monitor. Wrong looks like a single reading at the most visible spot standing in for the full protocol, which is exactly how a narrower, unremarkable-looking stretch with a real step goes unmeasured; stop rule, a classification based on fewer points than the method requires is not valid, go back and finish the readings. Hazard: kneeling or crouching to read a crack near a driveway or street edge carries ordinary traffic exposure, so the tech works facing traffic and off the travel lane where the site allows it.

4. Determine liability against the pulled record, not against how upset the customer sounds. Cross-check the classification against the step 2 records: does the crack's location match the recorded joint layout, was the delivery log clean for that pour, did a logged intrusion or a flagged repair already exist for this area. Acceptance: a written determination, covered, not covered and behaving as designed, or indeterminate pending an engineer, with the record citations that support it. Wrong looks like a verbal "we'll take care of it" offered on site before the records have even been checked, which commits the shop before it knows what it is committing to; stop rule, no coverage promise, spoken or implied, before the written determination exists. Hazard: none, a determination made from records.

5. Communicate the determination with the shop's script, and get the resolution in writing before any tool touches the slab. Use the tree's own translation language for a joint or shrinkage crack, that it cracked exactly where concrete is designed to crack, and be direct rather than soft where the finding is structural. Get the customer's agreement to the resolution, repair now, monitor with a logged re-inspection date, or engineer referral, before scheduling any work. Acceptance: resolution stated, agreed, and logged with a date. Wrong looks like a repair scheduled based on an on-site conversation with nothing written down, which is how the shop and the customer later remember two different agreements; stop rule, no repair work is booked until the resolution is in writing. Hazard: none at this step.

6. Execute or schedule the resolution, and close the dispute record. Route a repair to the appropriate technique, log a monitor case's re-inspection date onto the schedule per the tree's own cadence, or hold for the engineer's written report before any further action. Acceptance: the chosen path actually executed or scheduled, and the dispute record closed with the final disposition and date. Wrong looks like a monitor case verbally agreed to and never actually placed on anyone's calendar, which is how a genuinely active crack goes unchecked for a year; stop rule, the record does not close until the resolution path has a concrete date attached to it, not just an intention. Hazard: none beyond whatever the chosen repair technique itself carries, already covered by that technique.

The record this produces

One dispute record per reported crack. Fields: job and customer identifier, intake date and the three data points, urgency tier and visit date, pour records reviewed with what they showed, classification with measurements at every point and photographs, liability determination with its basis, resolution agreed and the date it was communicated, and the closure disposition with any re-inspection or engineer referral date.

It sits with the job's other pour records once closed. The PM reads it if the same customer calls again about a different crack, since a second dispute on one job is a pattern worth a site-wide look rather than a second isolated call. The owner reads a structural or disputed finding before it goes out, and the estimator reads dispute records across a season, since a shop whose disputes cluster on one crew's pours has a placement problem the sibling classification tree alone will never surface.

One pass, with a step that failed

A customer calls about a crack in a driveway poured three weeks earlier, describing it as "you can see it clearly, running kind of across." Step 1 took the three data points by phone: location roughly mid-panel rather than on a visible joint line, width described as "wider than a coin edge," no mention offered of a step. That description put it above hairline and outside a clean cosmetic read, so the visit was scheduled for the next business day, inside the two-day standard.

Step 2 pulled the joint layout record for that driveway ahead of the visit: the panel in question measured 11 feet along the driveway's length, with the recorded crack location roughly 3.5 feet from the nearest joint, meaning if it is ordinary restrained shrinkage it formed mid-panel rather than along a joint that failed to catch it.

Step 3 is where the first pass failed. The tech took one comparator reading at the point that looked widest, 0.010 inch, hairline-adjacent, and was ready to call it a monitor case without walking the crack's full length. The stop rule caught it: the method requires multiple points, not one. Re-measured properly at three points along the run, the readings came back 0.006, 0.009, and 0.022 inch, the last point well past the 0.018 inch ceiling the classification tree treats as the top of tolerable cosmetic width, with a slight step felt at that same 0.022 inch point that the single-reading pass had walked right past.

Step 4's liability determination followed the corrected classification, not the first one. A mid-panel location, inside spec joint spacing per the layout record, ruled out a joint-cutting defect, but the measured step moved this into the structural bucket regardless of location, which the tree routes to an engineer rather than a shop-level disposition. Step 5 communicated that plainly: not a coverage decision yet, an engineer referral, with the reasoning stated rather than softened. Step 6 scheduled the engineer visit and logged the dispute as open pending that report, rather than closing it on an incomplete read.

Had the first single-point reading stood, the crack would have been logged as a hairline monitor case, a customer told it was nothing to worry about, and a genuinely stepped, structural-bucket crack left unaddressed on the strength of one measurement taken at the wrong point on the line.

When more than one thing is going on

More than one defect shows up in the same visit, a crack plus a separate stain or spalled area: route the additional defect to the several defects on one slab triage tree rather than trying to resolve every finding through this single dispute record. The pour predates the shop's own record-keeping or the records were lost: proceed with the classification and note plainly that no pour record exists to cross-check against, which affects the liability determination's confidence but not the classification itself. The customer refuses a site visit and wants a repair or a credit based on photos alone: hold the line that classification requires an in-person measurement per step 3, and offer the visit again rather than resolving from photos. The crack was found by the shop's own crew on an unrelated visit rather than reported by the customer: this procedure still applies from step 2 forward, since the records pull and classification steps do not depend on who noticed it first.

References

• The customer says new slab is already cracking translation decision tree, in this library, for the crack classification method this procedure applies rather than restates.
• The hairline crack stable repair now versus monitor condition tree and the several defects on one slab triage tree, for the specific classification paths this procedure routes to.
• ACI 224.1R, causes, evaluation, and repair of cracks in concrete structures, for the tolerable width figures the classification method relies on.
• See related: the crack injection and patch repair technique and the early traffic damage to a fresh pour response, for the repair and discovery paths this dispute record hands off to.