Quick answer
Define exactly which two events your response-time metric measures. For first human response, use inquiry receipt and evidence of the first genuine human reply. Do not substitute review completion, client acceptance, assignment, or an automatic acknowledgement. Keep inquiries with no response visible rather than giving them a zero-minute result.
Agree on the event vocabulary
An intake workflow can record several useful milestones. The form receives the inquiry. A reviewer assesses it. A decision maker accepts or declines it. Someone takes ownership. A person then contacts the prospect. These milestones may happen in a different order in another business, so document the actual process before building a report.
Choose one primary metric and a few diagnostic intervals. For example, receipt to first human reply measures the prospect's wait, while receipt to review and acceptance to assignment explain internal delays. A single total completion metric can hide a fast review followed by a long period with no owner.
Find evidence for every timestamp
The entry creation time can establish receipt. Workflow history can establish recorded step transitions. A notification event can show that the system attempted a message, but it does not by itself prove inbox delivery or that a human wrote a reply. First-response evidence may need a CRM activity, help-desk message, or mailbox record.
Make a field dictionary containing event name, source, identifier, timezone, and meaning. Mark unavailable events as unavailable. Do not invent a first-response timestamp by copying the last-modified date, because a typo correction or background integration can update the record without contacting the prospect.
Join records using stable identifiers
Use an entry or lead identifier carried into the downstream system. Matching only by email address can combine separate inquiries from the same person. When a direct identifier is missing, flag the join as uncertain and review a small sample before trusting an aggregate.
Preserve the original timestamps and normalize a separate calculation column to one timezone. Record whether exported values include an offset. Around a daylight-saving transition, naive local timestamps can produce misleading intervals. A timestamp with a clear origin is more useful than a precisely formatted number whose timezone nobody knows.
Work through a small fictional example
Suppose an inquiry arrives at 09:00, a reviewer finishes at 09:20, an owner accepts assignment at 10:00, and a human replies at 10:35. Receipt-to-review is 20 minutes. Receipt-to-first-human-response is 95 minutes. Reporting 20 minutes as response time would describe an internal action, not the prospect's experience.
A second inquiry with an automatic acknowledgement at 09:01 and no human reply remains unanswered for this metric. Keep it in an open-inquiry list with its current age. Excluding all unanswered inquiries can make the average look better precisely when the process is failing.
- Use one record per inquiry, not one record per email address.
- Label automatic acknowledgements separately from human responses.
- Show missing timestamps and uncertain joins as exceptions.
- Keep declined and accepted cohorts separate when their processes differ.
Choose calendar time or business time deliberately
Calendar elapsed time is straightforward and includes nights and weekends. Business-hour time needs a named operating schedule, holidays, timezone, and treatment of partial days. Do not compare the two without a label. A Friday-evening inquiry can have a long calendar wait and a short business-hour wait.
Start with a transparent calendar-time report and a count of unanswered inquiries if business-hour rules are not yet agreed. Add the more complex calculation only when the business owner can explain it. Test known examples at opening time, closing time, weekends, and holidays before using the result as a service commitment.
Review distributions and exceptions
Show the median and a slower-response percentile alongside the count of inquiries and the count still unanswered. With a tiny sample, show the individual intervals instead of treating a percentile as a stable benchmark. Break the report down by service or assigned team only when each group has enough records to interpret.
Audit several rows back to the original evidence before automating the report. Check that a reopened inquiry, repeated workflow step, or second customer message does not overwrite the first valid response. The finished report should help somebody act on delays today, not merely produce an attractive average that conceals them.
Illustrative diagnostic example, not customer measurements.When to bring in help
For help connecting form and workflow evidence, bring a fictional report sample and your event definitions. HandL WP can check the technical joins and timestamps without treating an automated email as a completed human response.
Related troubleshooting
Keep intake fields complete when building from a PDF.
Helpful references
Client intake workflow example. Notification step and event behavior. User Input step settings.