Blog · Board and management reporting · Customer service
Using data to improve customer experience starts with a specific recurring problem. Follow a survey-and-ticket workflow that defines populations, compares segments, and turns the findings into a measurable service review.
Using data to improve customer experience starts with recurring service failures that a team can investigate and change. Compare structured feedback with ticket outcomes, choose one problem, and define how you will check it afterward. A segment difference is a useful lead, not proof of what caused the experience.
An illustrative US commercial-products supplier wants to reduce avoidable repeat contacts. Its experience manager must choose which recurring issue to review with operations first. The question is specific: which segment-and-issue group combines weak reported satisfaction with substantial repeat-contact work?
Set the reporting window to September 2026. Use the segment recorded at the customer interaction, rather than today's segment. The existing guide to B2B customer segmentation from the ledger covers how to establish purchase-based groups; here the groups are already approved.
Write down what qualifies as a repeat contact. In this example, it is a customer initiating another contact about the same unresolved case. Automatic notifications and internal handoffs do not count. Count each case once, even if the customer contacts the company several more times.
Bring an invitation-and-response export and a support export. The first needs invitation ID, interaction date, segment, approved issue code, question version, and numeric answer. The second needs ticket ID, created date, segment, primary issue code, repeat-contact flag, closed date, and elapsed resolution hours.
The survey covers eligible delivery, billing, and setup interactions; not every interaction produces a ticket. Keep the survey and ticket populations separate. Match their period and grouping rules rather than attaching every answer to every ticket for the same customer.
Keep the question, scale, and invitation timing consistent. Pew Research Center explains that survey trends need consistent wording and a comparable question context. If those changed, mark the break before comparing periods.
Use approved issue codes supplied by the business. Do not infer sentiment or classify free-text messages as if those were observed numeric answers. For scoring and exclusion rules, see customer satisfaction survey data analysis.
The manager selects one recurring issue for each segment from the existing reason codes. Each eligible survey invitation and ticket belongs to only one displayed group. Answers of four or five on the five-point satisfaction scale count as satisfied.
| Segment and issue | Invitations | Valid answers | Satisfied answers | Tickets | Repeat-contact tickets | Closed tickets | Resolution hours |
|---|---|---|---|---|---|---|---|
| Small business: delivery | 120 | 60 | 30 | 50 | 20 | 40 | 200 |
| Midmarket: billing | 80 | 40 | 32 | 30 | 6 | 24 | 48 |
| Enterprise: setup | 60 | 30 | 21 | 20 | 4 | 16 | 64 |
| Total | 260 | 130 | 83 | 100 | 30 | 80 | 312 |
The totals describe these selected groups, not all company activity. Closed tickets are those created in the window and closed by the reporting cutoff. Resolution hours include only that closed subset.
In Excel, put this table in A1:H5. Calculate satisfaction, valid response rate, repeat-contact incidence, and mean resolution hours in I2:L2, then fill down:
=IF(C2=0,"",D2/C2)
=IF(B2=0,"",C2/B2)
=IF(E2=0,"",F2/E2)
=IF(G2=0,"",H2/G2)
Format the first three columns as percentages. Show the final column in hours. A missing denominator should produce a blank or an explicit unavailable label, rather than a manufactured zero rate.
For underlying ticket rows, Microsoft's COUNTIFS documentation explains counting records that meet multiple criteria. If Tickets column B contains the approved group label and C contains numeric repeat flags, the repeated-ticket count is:
=COUNTIFS(Tickets!B:B,A2,Tickets!C:C,1)
For a row-level resolution check, with the same group in B, status at the September 30 cutoff in D, and validated hours in E, use:
=AVERAGEIFS(Tickets!E:E,Tickets!B:B,A2,Tickets!D:D,"Closed")
Microsoft documents AVERAGEIFS as the arithmetic mean of values meeting multiple criteria. The ticket sheet must already be restricted to the agreed reporting cohort. Covirage is not affiliated with Microsoft or Google.
The calculations return the following figures. Percentages are rounded to one decimal place and hours to two decimals.
| Segment and issue | Satisfaction | Valid response rate | Repeat-contact incidence | Mean resolution hours |
|---|---|---|---|---|
| Small business: delivery | 50.0% | 50.0% | 40.0% | 5.00 |
| Midmarket: billing | 80.0% | 50.0% | 20.0% | 2.00 |
| Enterprise: setup | 70.0% | 50.0% | 20.0% | 4.00 |
| Selected groups | 63.8% | 50.0% | 30.0% | 3.90 |
Delivery is the review candidate: 20 of its 50 tickets required repeat contact, and 30 of its 60 valid answers were satisfied. An operations owner should inspect those cases for a recurring, actionable failure before changing a process.
Do not conclude that delivery is inherently worse than billing. Each row has a different customer segment as well as a different issue. Also, the survey respondents need not be the customers whose cases had repeat contacts. The table identifies a place to investigate; it does not identify an individual cause.
Overall repeat-contact incidence is 30 / 100 = 30.0%. Overall satisfaction is 83 / 130 = 63.8%, not the simple average of the three displayed scores. The guide to weighted averages in Excel explains why totals need their original denominators.
Resolution hours reconcile as 200 + 48 + 64 = 312, and closed cases as 40 + 24 + 16 = 80. Therefore mean resolution time is 312 / 80 = 3.90 hours. The other 20 tickets are open at cutoff and excluded from that average; review their age separately.
Record a named operations owner, the proposed process change, its start date, and the next equivalent observation window. Keep the original definitions. If delivery repeat-contact tickets were 10 rather than 20 with all denominators unchanged, delivery incidence would be 10 / 50 = 20.0%, and the selected-group total would be (10 + 6 + 4) / 100 = 20.0%.
That arithmetic states what success would look like under fixed inputs. Real follow-up counts must come from the next export, with changes in volume and customer mix disclosed. Use a customer experience dashboard to display the agreed measures after deciding which action they support.
Upload your structured survey and ticket summaries, or bring a Google Sheet with approved segment and issue definitions. Covirage maps the columns; its tools compute and check the comparisons, and the external AI model explains the result with every figure traced to its rows. Create a free Covirage account and bring a sample of your own data.
See Covirage for customer service teams for the supporting industry context.
Start with numeric satisfaction answers, survey invitations, ticket reason codes, repeat-contact flags, and resolution timestamps. These fields connect feedback with observable service outcomes. You can usually remove names and message bodies while retaining the IDs needed for matching and duplicate checks.
Look for a recurring, material problem with clear ownership and reliable supporting records. Consider affected volume, repeat contacts, survey coverage, and the effort of a reversible intervention. Avoid ranking problems from satisfaction percentages alone.
No. Respondent mix, seasonal demand, other service changes, and question wording may also change the score. Compare equivalent populations and record what else changed. A stronger causal conclusion requires a suitable evaluation design, not just a before-and-after chart.