Check the quality you shipped

First-pass yield and the cost of rework

Why this step

How many units passed the site audit the first time, which failure modes recur, and what the rework cost. Broken down by profile series, this is the report that tells you a supplier or a design is the problem, not the crew.

Open the full article

What this does

Site QC analytics live under Installation › Analysis & Reports as five screens over the checks installers answer on site: Site QC Results (every answered check, pass or fail), Site QC Failures (straight to what fails most), Site QC First-Pass Rate (the share of openings that passed their very first inspection), Site QC Rework (re-inspections, with the hours between first failure and eventual pass), and Site QC Measurements (how far measured values drift from tolerance).

Two record streams feed them: one row per check (results, failures, measurements) and one row per inspection attempt (first-pass, rework). Check rows appear when a sheet is opened and fill in as the installer answers; submitting the sheet stamps the inspector and date and appends the attempt row.

Before you start

The Site QC toggle enabled in the Installation settings, and an installation role — the same access as the rest of the Installation app.
Checklists configured (Installation › Configuration › Site QC › QC Checklists) and inspections actually submitted. One sample checklist ships; answers start at zero, so a fresh site shows greyed sample data on the graph and pivot screens (the list-first screens simply open empty).

Steps

Site QC Failures — failed checks counted per check, the site-quality Pareto

01
Open Analysis & Reports › Site QC Failures first — it is the manager's report: only failed checks, grouped by check, so the top bar is your biggest site problem.
02
Cross-check in Site QC Results: every answered check, the pivot crossing checks against passed/failed counts. Filters cover Failed, Passed, Critical and This Month; group by project, work package, inspector, product or severity.
03
Open Site QC First-Pass Rate: only each opening's first inspection, grouped by inspector by default — read the passed-versus-failed split per group as the rate (there is no single stored percentage).
04
Open Site QC Rework: attempts after the first, with Rework Hours — the elapsed time from the first failed inspection to the eventual pass — as the pivot measure per inspector or project.
05
Open Site QC Measurements for the dimensional checks: the Deviation measure is how far outside tolerance each measured value landed — rising breach counts or magnitudes per check are the early warning (values inside the band record a deviation of zero).

The five screens

Screen	Rows	Scope and default
Site QC Results	All check rows	Everything, grouped by check — including still-blank rows from half-filled sheets, so apply the Passed/Failed filters for answered-only counts. The list shows the measured value, deviation, severity and the pass toggle.
Site QC Failures	Answered checks	Pre-filtered to failures only, grouped by check.
Site QC First-Pass Rate	Inspection attempts	First attempts only, grouped by inspector; the graph can also split passed-versus-failed per week.
Site QC Rework	Inspection attempts	Attempts beyond the first, grouped by project, measuring Rework Hours.
Site QC Measurements	Answered checks	Measurement-type checks only, graphing Deviation per check.

Site QC Results pivot crossing each check against passed and failed counts

First-Pass Rate grouped by inspector — first attempts only

How a check passes — the rules behind the numbers

Pass/fail and yes/no checks pass on the obvious answer; choice checks pass when the selected option is in the checklist's passing list; photo checks pass when a photo is attached.
Measurements pass inside the tolerance band — and a check whose tolerances are both zero is informational: it always passes and records the value only.
A check that requires an evidence photo fails without one, whatever the measured answer says.
Deviation is recorded only when a value lands outside tolerance — the distance beyond the nearer bound.
First pass = attempt number 1 and passed. Rework Hours is wall-clock time between the first failed attempt and the eventual passing one — elapsed hours, not labour booked and not money; it lands on the passing re-attempt only, so intermediate failures carry zero and a job never re-passed contributes nothing.

Measurement Spread — the dimensional checks, ready to graph deviation from tolerance per check

Troubleshooting

The graphs show smooth grey data on a new site	Sample mode on the chart and pivot views — no inspections submitted yet. It disappears with the first real submission; the list-first screens just open empty. Note the breadcrumb titles differ slightly from the menu labels (Failing Checks, Rework Loop, Measurement Spread) — same screens.
A measurement inside tolerance shows as failed	The check demands an evidence photo and none was attached — missing evidence fails the check regardless of the value.
An "always passing" measurement check	Both tolerances are 0.0 — that makes it informational by design. Set real bounds if it should gate.
Expected rework costs, found only hours	Correct — the module measures elapsed hours between failure and pass. Cost it outside if you need money.
First-pass numbers look too good	Only submitted inspections count. Crews fixing issues before submitting the sheet never register a failed first attempt.
The five menus are missing	The Site QC toggle is off in the Installation settings, or you lack the installation role — the entries live inside the Installation app and follow its access.

Common mistakes

Reading rework hours as technician time on the job — it is calendar time between two inspections, including nights and weekends.
Comparing inspectors' first-pass rates without comparing what they inspect — hard checklists depress the rate.
Leaving tolerances at zero on dimensional checks and believing the site "never fails a measurement".
Building the Pareto while crews free-type answers into the wrong checklist — the analytics are only as clean as the checklist discipline.
Judging quality on Results counts alone: an opening inspected three times contributes three rows; use First-Pass Rate for the per-opening truth.

Checkpoint

You can attribute site failures to a profile series or a failure mode rather than to a person.