A fast reply needs a useful next step.
Speed to lead is the time between a real borrower inquiry and a relevant response. An automated greeting alone does not prove that your team answered the question or arranged the next conversation.
Start the clock at the right event
A campaign entering a queue, a model producing text, Bonzo accepting a request, and a borrower receiving a message are different events. Define which event starts your response-time measurement and which finishes it. For inbound conversations, preserve the time the borrower actually wrote, not just when a background import found the record.
LoanWick supports Bonzo-connected replies using available history and officer settings. Actual timing depends on eligibility, contact windows, human activity, upstream services, and processing. This guide does not promise a fixed response-time guarantee.
Give the reply a job
Fictional contextA purchase inquiry already names the assigned officer and says the borrower is browsing homes.
BorrowerCan someone explain what I need before making an offer?
AssistantYes. Would you like to set up a quick call with your loan officer to go through the next steps?
Illustrative script, not a production transcript or lending advice. Confirm the officer and prior context before using any example.
Design coverage before expanding traffic
Separate inbound replies from campaigns
Identify the conversations that need an answer rather than measuring every queued background update as a new inquiry.
Check the permitted scope
Review the officer, stage mode, pause, contact window and recent human activity before expecting an assistant response.
Make the next action concrete
Answer within the assistant’s scope, ask one useful question, or coordinate the assigned officer’s next step.
Inspect outliers
Look at slow or failed eligible conversations, including delivery evidence, instead of relying only on an average.
Measure speed without rewarding empty messages
Report eligible inbound volume, the median and slowest response bands, unanswered cases, and the outcomes after the reply. Keep excluded, paused and opted-out records separate from an actual response failure. A small pilot with honest exception review is more useful than a large campaign with an attractive but incomplete average.
| Measure | What it tells you | What to check |
|---|---|---|
| First generated reply | Shows model activity | Not proof of delivery or a useful answer |
| First delivered relevant reply | Shows a completed response step | Needs an explicit delivery and relevance definition |
| Confirmed consultation | Shows the next commitment | Must be distinguished from a suggested time or link |