Lyon Waugh Auto Group — Engagement Audit

Window: April 25 – May 01, 2026 (7 days; today is partial) Channel: Acura of Peabody (Jennifer Beal) Generated: May 1, 2026 21:35 UTC

Snapshot of how Jennifer Beal engaged visitors on acuraofpeabody.com — visitors landing, sessions, scroll/idle behavior, conversations created, leads escalated to the dealership team, and emails delivered.

At a glance

Numbers reflect distinct counts across the whole 7-day window — visitors who returned on multiple days are counted once.

How escalation works (so the metrics make sense)

When Jennifer Beal decides a conversation needs a human, she calls “set needs attention” — at that moment three things happen at once:

1. The conversation is tagged needs_attention (durable — the tag stays as the audit trail)
2. An email is delivered to the right dealership inbox immediately (sales → leads@driveacurapeabody.com + dcurrie@lyonwaugh.com; service → service team; parts → parts)
3. The conversation status is set to needs_attention momentarily

As soon as the customer keeps replying, the status flips back to pending_user_reply so Jennifer can keep helping — but the tag persists and the email has already been sent. So:

• The number of leads the team got emailed about = the count of escalations that fired (the daily “emails” column below).
• It is not the count of conversations whose status currently reads needs_attention — that resets quickly as customers continue chatting.
• It is not the count of generic email.delivered events — those include inbound, transcript copies, etc.

Daily breakdown

Apr 25–27 the proactive engagement layer was off (or rules disabled). Apr 28 it switched on and the funnel woke up. The cadence has held since.

Leads escalated to the dealership team

Every entry below is a conversation Jennifer Beal escalated. The escalation email was sent the moment it happened to the right dealership inbox — formatted as DriveCentric ADF/XML for sales and HTML summary for service.

Sales (11)

Service (6)

Voice calls (2)

What’s working

• Proactive idle engagement is converting at ~27% reply rate when it fires — strong number for an unprompted outreach.
• Rage-click recovery converts ~13% of frustrated clicks into a conversation.
• Voice handling end-to-end is solid: the inbound voice lead this week (Sneesh Sharma, +1 (857) 258-6545) escalated to Doug with the model question and phone number for callback.
• DriveCentric ADF format is being used on every sales escalation — leads land in CRM ready to ingest.
• Highest-intent customers got full handoff context — the Sarah Lopes lead (May 1 evening) captured VIN, payoff, payment target, credit band, and a confirmed Saturday 9:30am appointment before handoff.

What needs attention (action items)

1. Conversations list filter — needs_attention now matches by tag (just shipped). The filter previously only matched conversations whose status was still needs_attention. Because that status flips back to pending_user_reply as soon as the customer keeps replying, most escalated leads were invisible. The filter has been updated to match by tag, which is the durable signal. You’ll see all 18 flagged leads now.

2. One-click “Mark as cleared” added on row hover. A green check icon appears on any row that’s currently flagged. One click removes the tag, resets the status, and removes the tag from the matched contact — the row leaves the bucket immediately.

3. One day this week opened many conversations but most visitors didn’t reply. A rule fired aggressively at the wrong moment. Worth tightening the trigger conditions on the heaviest-firing rule (Idle on used inventory page) so it doesn’t flood quiet sessions.

4. Cooldown / message-cap throttle hits show as “failure” in the rule dashboard. When the engine correctly throttles a visitor (because we already messaged them recently), it logs the result as failure instead of skipped. This makes the dashboard look like a high error rate when the actual error rate is closer to 5%.

5. Duplicate escalation emails — bug fixed. A spot check against Postmark showed several conversations had been escalated multiple times across turns, sending the dealership team duplicate copies (Sarah Lopes received 6 rounds; one other conversation triggered 8 rounds). The cause was that change_conversation_status: needs_attention could be called repeatedly on the same conversation, with no guard. The function now suppresses the email if the conversation already carries the needs_attention tag — the team is told once, future re-calls update the reason silently. Clearing the tag with the new “Mark as cleared” button re-arms the next escalation. Jennifer Beal’s prompt has also been updated to instruct her not to re-escalate the same conversation.
6. Auto-tag opportunities for sharper review. Conversations that capture phone + appointment + 3+ exchanges should auto-tag as hot_lead; conversations with a VIN should auto-tag vin_captured. Lets the dealership filter for highest-value items first.

Filter recipes the team can use today

• All open leads — filter needs_attention (now matches by tag, catches all 18 above).
• Just sales pipeline — needs_attention + tag sales.
• Just service — needs_attention + tag service.
• Voice calls only — Communication method = Voice (now fixed; finds all 7 calls).
• Hot recent — sort by created_at desc, filter status pending_user_reply or needs_attention.

Bottom line

Jennifer Beal’s last 7 days delivered 18 distinct conversations escalated to the dealership team — each one triggered an email at the moment it fired and went straight to the right inbox. 18 of those leads are still flagged for follow-up right now, and now visible under the “needs attention” filter and one-click clearable from the row. The strong full-detail captures on the highest-intent buyers (Sarah Lopes appointment, Sneesh Sharma callback, Daniel Card service+loaner) show the prompt is working.