A customer complaint analysis presentation should explain what the records count before it recommends a fix. Contacts, complaint cases, affected orders and customers are different things. Mixing them can make a service problem look larger, smaller or more widespread than the evidence supports.
Begin with one decision: which complaint category deserves investigation or a proposed response? The broader data storytelling workflow for presentations helps connect a finding to that decision. Here, the work starts one step earlier: making the complaint counts comparable.
Choose a case definition before drawing a chart
Imagine a customer emails about a delayed delivery, phones for an update and then replies to the original email. That can be three contacts about one complaint case. If the customer also reports damaged packaging, your system may treat that as a second case or a second label on the original case. State which rule your analysis uses.
Keep an anonymous case identifier in the source table, alongside the opening date, primary category, status and any severity flag. Add an order reference only when the relationship has been checked. The presentation needs the aggregated results, not a copy of customers’ contact details or private messages.
Use one primary category per case for a part-to-whole chart. If several labels can apply, call the chart “issue mentions” and explain that its total may exceed the number of cases. Neither approach is inherently correct for every analysis; the label must match the counting rule.
A worked example: 210 contacts, 150 cases
This fictional example describes one month of an online shop’s complaint records. The team has grouped 210 contacts into 150 complaint cases. Each case receives one primary category. All figures below are constructed for the explanation, not reported customer outcomes.
| Primary category | Cases | Share of 150 cases | Cumulative share |
|---|---|---|---|
| Late delivery | 90 | 60.0% | 60.0% |
| Refund delay | 25 | 16.7% | 76.7% |
| Damaged item | 20 | 13.3% | 90.0% |
| Unclear return policy | 15 | 10.0% | 100.0% |
The category counts sum to 150. Late delivery accounts for 90 ÷ 150 = 60% of cases, not 60% of customers or orders. The first two categories account for 115 cases, or approximately 76.7%. Keep the unrounded numbers for calculations and round only for display.
A descending bar chart is enough to show which category is largest. Add a cumulative percentage line if it helps the audience choose how many categories to investigate together. ASQ’s Pareto guidance describes ordering categories by a stated measure, such as frequency or cost. A count-based ranking should not quietly become a ranking of harm or financial impact.
Do not force the data to fit an “80/20” headline. In this example, the two largest of four categories account for about 76.7% of cases. The actual distribution is the point.
Explain why more complaints can coexist with a lower rate
Now suppose a second month has 180 cases and 1,500 orders, compared with 150 cases and 1,000 orders in the first month. The case count rises by 30, or 20%. The count per 100 orders falls from 15 to 12. Both statements can be true.
| Measure | Month A | Month B | Change |
|---|---|---|---|
| Complaint cases opened | 150 | 180 | 30 more cases |
| Orders placed | 1,000 | 1,500 | 500 more orders |
| Cases opened per 100 orders placed | 15 | 12 | 3 fewer cases per 100 orders |
This is a workload ratio. It is not the percentage of orders that generated a complaint. A case opened this month may refer to an earlier order, and one order may generate several cases. To report a true share of affected orders, link cases to distinct orders in a defined order cohort and allow an appropriate observation period.
A useful slide headline is “Complaint workload grew, while cases per 100 new orders fell.” It preserves both observations. A headline such as “Customer satisfaction improved by 20%” would invent a different measure and a conclusion the records do not establish.
Keep frequency and severity visible
The highest bar offers a place to investigate. It does not decide every priority. A smaller group of damaged-item cases may require urgent attention even while the team studies delivery delays. Review serious incidents through the organization’s existing escalation process; they should not have to wait until their category becomes the largest.
A separate status view can also change the discussion. Twenty-five refund-delay cases include cases at different stages: some may be resolved, some awaiting payment and some awaiting customer information. Show the definitions before comparing age or resolution time. An open case has not yet reached its final resolution time.
Use two adjacent panels: complaint frequency on the left, unresolved or escalated cases on the right. They answer different questions. Avoid combining them into a weighted “priority score” unless someone can explain the weights and the decision they support.
Use five slides for the operational review
- Decision and scope. State the reporting window, which channels are included and the investigation requested.
- Counting rule. Explain the 210-contact to 150-case grouping and the one-primary-category rule.
- Pattern. Show category counts and the actual cumulative shares. Label the measure on the chart.
- Context. Show the order-volume comparison beside unresolved or escalated cases. Keep the units separate.
- Next investigation. Name the records needed to distinguish plausible explanations and who will review them.
For late delivery, the next step might be to join the affected cases to dispatch and carrier records, then separate dispatch delays from transit delays. A complaint saying “the parcel arrived late” establishes the reported problem; it does not identify which stage caused it.
The team can request that analysis without already promising a new carrier, extra staff or a measured reduction. For a full improvement project after the problem is defined, use the existing DMAIC presentation structure. Keep this meeting focused on what the complaint records can currently support.
Prepare a text brief for the slide draft
Once the counts and definitions have been checked, the Presenti text-to-presentation workflow can organize the written brief into slides. Calculate the figures in the source worksheet first. Generating a presentation is not a substitute for deduplicating case records or deciding which denominator is valid.
Create a five-slide operational review using this fictional complaint analysis. Month A contains 210 contacts grouped into 150 cases, with one primary category per case: late delivery 90, refund delay 25, damaged item 20, unclear return policy 15. Show counts, shares and cumulative shares without asserting an 80/20 split. Month A had 1,000 orders; Month B had 180 cases and 1,500 orders. Call the comparison cases opened per 100 orders placed, not percentage of affected orders. Keep frequency, unresolved status and severity distinct. Request a review of dispatch and carrier records for late-delivery cases. Do not invent causes, customer quotes, savings or improvement results.
When reviewing the draft, inspect the chart labels as carefully as the arithmetic. A correct value under the wrong label can still mislead the meeting. The reader should be able to trace the recommendation back to a defined set of cases and understand what further evidence is needed.
