SiliciumHex FieldKit

Problem Solving & Quality · Observe

Check Sheet

Design a simple tally form — defect type by time, machine or location — that operators fill in at the source with one stroke per event.

  • Time30 min
  • FormatSolo
  • StageObserve

Check Sheet: what it is and why it works

A check sheet is a simple, purpose-built form for counting events at the point where they happen. Categories (defect types, stoppage causes, locations) run down the rows; time slots, machines or batches run across the columns; and the person on the spot adds one stroke per event. After a defined collection period, rows and columns are totaled. A useful variant, the location check sheet, is a sketch of the part or equipment on which each defect is marked where it occurs.

Many improvement efforts stall because the problem exists only as recollection: 'it happens a lot on nights', 'mostly on the left side'. A check sheet converts that into counts at almost no cost, and because operators fill it in themselves it also builds shared ownership of the data. It is quicker to set up than a change to the MES or historian, and it captures categories that automated systems often miss. It is one of the seven basic quality tools and feeds the others directly: Pareto analysis ranks its categories, stratification compares its columns, and a prior gemba walk usually tells you which categories are worth counting in the first place.

What you need

  • The question the data must answer
  • A list of categories agreed with the people who will record
  • The columns to split by: time, machine, batch, shift
  • A defined collection period and recording point
  • Operators willing to test and refine the sheet

What you get

  • A tested one-page check sheet
  • Tallies by category and by column for the collection period
  • Row and column totals
  • Input data for a Pareto chart or stratification
  • Operator feedback on missing or unclear categories

When to use it

When data exists only as memories and complaints, never as counts.

How to do it, step by step

  1. Decide the question the data must answer, and the categories to count — defect types, locations, causes of stoppage.
  2. Draw a grid: categories in rows, time slots, machines or batches in columns.
  3. Test the sheet for one shift with the operators and simplify whatever slowed them down.
  4. Collect for a defined period, one stroke per event, at the point where it happens.
  5. Total rows and columns, then feed the result into a Pareto chart or stratification.

Worked example: Weld defects on pallet-rack uprights

Illustrative scenario — figures are realistic but not from a real company.

A fabricator of steel pallet-rack uprights reworks about 4% of the robotic welds that join bracing to the columns. Supervisors are convinced spatter is the main issue, but nobody has ever counted defects by type.

  1. The quality engineer framed the question: which defect types occur, on which of the two robot cells, and where on the upright?
  2. The sheet listed five defect types in rows (porosity, undercut, spatter, missing weld, burn-through) against columns for cell A and cell B by shift. A side sketch of the upright showed its 14 weld joints, numbered.
  3. Welders tested it for one shift. 'Spatter' was split into 'spatter on weld' and 'spatter in bolt holes', because only the second one caused problems at the customer.
  4. For two weeks, inspectors added one stroke per defect and a dot on the sketch at the joint where it occurred.
  5. Totals: 312 defects. Missing welds made up 41%, almost all on cell B, and the sketch showed them clustered at joints 13 and 14, at the top of the upright.

Result. The pattern pointed to cell B's upright shifting at the far end of the fixture. A worn clamp was found and replaced, and the robot path was touched up. Weld rework fell to about 1.5%. Spatter, the assumed main issue, came third on the list.

Common pitfalls and how to avoid them

  • Too many categories.Keep categories few and distinct, with an 'other' box that you review; split later if 'other' grows.
  • Designing the sheet without the people who will use it.Test it for a shift with the operators and remove anything that slows them down.
  • Recording after the fact.Keep the sheet where the event happens and mark it immediately; end-of-shift memory reintroduces the bias you are trying to remove.
  • Collecting data without a question.Write the question at the top of the sheet and stop collecting when it has been answered.

Frequently asked questions

What is the difference between a check sheet and a checklist?

A checklist confirms that required steps were done, with each item ticked once. A check sheet counts how often events occur, with one stroke per event, so it produces frequency data. A pre-start equipment inspection form is a checklist; a tally of defect types by shift and machine is a check sheet. Both are useful, but only the check sheet feeds a Pareto chart.

How long should you collect data on a check sheet?

Long enough to capture the normal variation of the process and enough events to see a pattern, often one to four weeks covering every shift, product and condition that might matter. For rare events, extend the period or widen the scope. Decide the period in advance so collection does not stop the moment a favorite theory seems confirmed.

When should you use a location check sheet?

When the position of a defect on a part or piece of equipment may reveal its cause: paint defects on a panel, leaks along a pipe run, wear across a conveyor belt, porosity on a casting. Marking each defect on a sketch shows clusters that a table of counts cannot, such as defects always near a clamp, a gate or one side of a mold.

Origin

Check sheet — one of the seven basic quality tools promoted by Kaoru Ishikawa and JUSE (Guide to Quality Control, 1968).

Used in these playbooks

Chronic scrap reduction month 1 month

One month against a loss everybody has learned to live with: count at the source, rank the losses, split the data, list the causes and prove the real one before spending money.

  1. Check Sheet
  2. Pareto Analysis
  3. Stratification
  4. Ishikawa Diagram
  5. On/Off Cause Verification

Related methods

More in “Observe”

Go where the problem happens and collect facts, counts and timelines — not opinions.