Look at the journey before buying the fix
The inbox contains unanswered questions, the calendar contains meetings, and the ad dashboard contains leads. None of those screens alone tells you where the customer journey broke.
A lead-response audit joins them. Pick a manageable sample of recent enquiries and trace what happened next. Twenty is a starting workload for inspection, not a statistically representative sample or a universal audit standard. If volume is low, use the available enquiries and show the count.
Choose one channel and a defined arrival period. Mixing referrals, cold enquiries and paid leads can hide differences in intent. Keep enquiries that failed to book in the sample; they are often the most useful records to inspect.
Make one row per enquiry
Assign an internal identifier instead of copying customer details into a shared report. Record arrival time, first acknowledgement, first useful answer, qualification outcome, next action, owner and booking outcome.
Use explicit definitions. An automatic “we received your message” is an acknowledgement. A useful answer addresses the question or asks for a detail required to answer it. A booking is a provider-confirmed appointment. An attended appointment needs a separate attendance record.
| Check | Evidence to look for | Failure example |
|---|---|---|
| Capture | Enquiry appears in the team's working queue | Form submission exists only in email |
| Useful response | Question answered or necessary detail requested | Instant greeting followed by silence |
| Ownership | One named person owns the next action | Everyone assumes someone else replied |
| Qualification | Fit is recorded with a reason | A booking link is sent to every enquiry |
| Next step | Customer knows what to do and why | “Let us know” with no useful direction |
| Booking | Confirmed identifier, time and event type | Link click counted as an appointment |
| Closure | Booked, closed or handed off with a reason | Follow-up continues after cancellation |
Mark each check pass, fail or unknown. Unknown is valuable. It shows where the team lacks evidence; do not convert it into a failure count just to make the score look decisive.
Calculate two response times
Calculate acknowledgement time and useful-response time separately. Then show the median and the longest waits in the sample. A single average can hide a few customers who waited much longer than everyone else.
Keep both elapsed time and business-hours context if the team has a stated coverage schedule. An enquiry received on Sunday should not be silently removed from the report. Show when it arrived and when the team was expected to respond.
Also inspect the response itself. A fast wrong answer can create extra work. A slower but complete answer may resolve the enquiry. Time measures the delay; conversation review measures whether the next step helped.
Turn observations into a repair order
Imagine a fictional sample of twenty enquiries. Four never entered the working queue. Six received a greeting but no useful answer. Three had no assigned owner. Those groups may overlap, so do not add them and claim thirteen separate lost customers.
Fix capture first if enquiries are disappearing. Fix ownership next if nobody knows who should act. Then improve the question path or response coverage. Sending faster messages into an unowned queue will not repair the later steps.
For each proposed fix, write the evidence, the change and a verification check. “Add notifications” is vague. “Create one owner task when a new form enquiry has no useful response by the team's agreed deadline” can be tested.
Run the same audit after the change
Take another sample using the same channel, definitions and selection rule. Record the arrival period and any changes in offer, traffic or staffing. If those changed, say so; a before-and-after difference does not isolate the effect of automation.
Compare the failure you meant to repair. Did missing enquiries reach the queue? Did ownership become clear? Did useful-response delays fall? Then inspect qualification, booking and attendance for unintended effects.
Do not promise recovered revenue from an audit count. A missed enquiry may have been unsuitable, duplicated or unlikely to buy. The audit identifies preventable process failures; paid outcomes require their own records.
Keep the first report small enough to use
Finish with three observations, one repair owner and one review date. Attach the definitions so the next operator can repeat the audit. A long scorecard without a decision is another piece of administration.
Rainlight's lead-response scorecard can help structure the discussion. Treat its result as a diagnostic aid rather than an industry benchmark. If the audit points to a broken capture-to-booking path, use the service-business briefing to review the connections that need repair.