Complaint Pareto: group by product, failure mode or customer?

This guide is for business-to-business complaint records at a chemical supplier or distributor, where each complaint names a product, a customer, a date and a described failure. It does not replace a regulated complaint procedure, such as quality defect handling for medicinal products under EU GMP Chapter 8, and a Pareto ranking does not by itself prove a root cause.

Which grouping should a chemical supplier complaint Pareto use?

A chemical supplier complaint Pareto should use the grouping that matches the decision being made: failure mode to choose corrective actions, product to assign owners, customer to judge account risk, and lot or production period to test whether a change caused the complaints. A Pareto chart ranks categories so a team can focus on the "vital few" factors that warrant the most attention (IHI, accessed 2026), but which factors appear depends entirely on the grouping you pick, so write the grouping into the analysis plan first.

The decision table below matches the complaint Pareto grouping to the question a chemical supplier is asking.

The first row is the usual starting point: when the goal is to choose and fund corrective actions, group by failure mode. When the complaints concern how cases were handled rather than the product, group by handling stage instead; How do you find where your technical support cases get stuck? covers that analysis.

When should a complaint Pareto group by failure mode instead of product?

A complaint Pareto should group by failure mode instead of product when the next step is choosing a corrective action, because a single cause such as water contamination, poor mixing instructions or a packaging defect can produce complaints across many products. A product Pareto spreads that one cause across several bars, so none of them looks large enough to act on.

The Minitab blog gives a non-chemical version of the same surprise: one company believed most customer complaints were about product defects, but its Pareto chart showed many more complaints about shipping delays (Minitab Blog, 2016). The grouping changed the conclusion.

Use product grouping when ownership is the question. If the decision is which product manager, formulation chemist or plant must respond, a product Pareto assigns the work cleanly. A common pattern is to run both: a failure-mode Pareto across the whole portfolio, then a product Pareto inside the top one or two failure modes.

Minitab's Pareto tool supports this with a BY variable, which draws one Pareto per group using the same ordering of bars across all groups (Minitab Support, 2026). In practice that means one failure-mode Pareto per product family, side by side, so a reader can see whether the same failure leads in every family or only in one.

Define failure modes as observed symptoms with one-line definitions, not as suspected causes. "Cloudy", "hazy" and "milky" become one code; a cause code like "contamination" would decide the answer before the analysis.

Should a complaint Pareto rank by count or by cost and impact?

A complaint Pareto should rank by count when complaints are of similar severity, and by cost or impact when they are not, because a count Pareto treats a line stoppage and a late certificate of analysis as equal. Practitioners on the Elsmar Cove quality forum put it plainly: a single instance of problem A can be more costly than 100 iterations of problem B, so "follow the money" (Elsmar Cove forum, 2012).

The same thread describes what impact includes for a supplier: the value of rejected parts, missed shipments, line shutdown costs charged back by the customer, and future business that may be lost because of the customer's view of the supplier (Elsmar Cove forum, 2012). For a chemical supplier, add replacement product, freight, site visits and laboratory time.

Build the impact score before seeing the ranking. A simple approach is a 3-level severity weight, for example 1 for an inquiry with no loss, 3 for a product replaced and 9 for a customer line stoppage or safety concern (illustrative). Multiply each complaint by its weight and rank the weighted totals.

Show both charts when they disagree. If odor complaints lead on count but lead time to reformulate makes a rare separation failure the costliest, the count chart tells you what customers notice most and the weighted chart tells you where the money goes. Write the weights into the analysis plan so the next period uses the same scale.

How do you stop one complaint being counted twice in a Pareto?

To stop one complaint being counted twice in a Pareto, assign each complaint one primary category and record any secondary categories in a separate field that is reported but not added to the bars. Without that rule, a complaint about both foaming and residue appears in two bars, and the bar percentages no longer describe complaints.

A Pareto chart's cumulative percentage line runs from the first bar to the last and is meant to add up the contribution of each category (Minitab Support, 2026). That arithmetic holds only when each record sits in exactly one category. If double tagging is allowed, state the unit in the chart title, for example "failure tags" rather than "complaints", and report how many complaints carried more than one tag.

Three rules keep the count clean:

1. Pick the primary failure mode as the one that drove the customer's action, such as stopping use or rejecting a delivery.

2. Merge duplicate records of the same event, such as an email and a call about one delivery, into one complaint with one ID.

3. Count repeat complaints from one customer about one event once; count a new event from the same customer again.

Check the coding before trusting the ranking. Have two people code the same 20 records independently and compare (illustrative). Where they disagree, rewrite the category definition rather than the record.

What period should a complaint Pareto cover?

A complaint Pareto should cover a fixed period of complete months that holds enough complaints for the top bars to be stable, and every comparison should use the same period length and the same category definitions. A quarter is a common choice for a supplier with steady complaint volume, and a rolling 12 months suits low volumes (illustrative).

EU GMP Chapter 8 asks for quality defect records to be reviewed and trend analyses performed regularly, without fixing an interval (European Commission, 2014, section 8.19). The interval is yours to set, but once set it should stay unchanged so each review compares like with like.

Count-based comparisons assume a stable base. The NIST/SEMATECH e-Handbook defines an inspection unit for counting defects and notes that its size may depend on the recording facility, measuring equipment and operators (NIST/SEMATECH e-Handbook, section 6.3.3.1). For complaints, the equivalent base is shipments or active customers. If shipments doubled between two quarters, compare complaints per 100 shipments rather than raw bar heights.

Record anything that changes the base during the period: a new complaint form, a new logging team, a new product launch or a lost large account. A bar that rises on the day the logging changed may be a recording effect, not a product effect.

How do you turn the top Pareto bars into improvement priorities?

To turn the top Pareto bars into improvement priorities, take the bars that make up roughly the first 80% of the weighted total, check each for a shared cause, and open one investigation per bar with an owner and a success measure set before the action starts. The 80% line is a guide; the Elsmar Cove thread notes that the ratio "isn't hard and fast by any means" (Elsmar Cove forum, 2012).

For each top bar, drill one level down with a second Pareto: by product inside a failure mode, or by lot inside a product. EU GMP Chapter 8 frames the same step for medicinal products, saying root cause analysis should be applied at an appropriate level and, where the true cause cannot be determined, the most likely root causes should be identified and addressed (European Commission, 2014, section 8.16).

Keep the remaining bars visible. The thread's practitioners describe the smaller categories as a view of what is trending, such as training or material issues (Elsmar Cove forum, 2012). EU GMP Chapter 8 also says the effectiveness of corrective actions should be monitored and assessed (European Commission, 2014, section 8.18), so rerun the same Pareto, same period length and same categories, after each action.

When the top bars hold only a handful of complaints each, or one complaint could sit under two groupings, have a statistician check whether the ranking would hold with another period before funding actions. Lubinpla's technical support performance analysis fixes the denominator and exclusion rules first, then sets out repeat failures and improvement priorities from your own case records in one cited document for your review.