Broken BDC callback cadence is the single most-underinvested loss in dealer pipeline. A $500 CPL becomes a $2,000 CPL when the callback slips past 30 minutes. Nobody in the group notices because the loss shows up in the "did not convert" column, not in any operational dashboard.
This playbook is a 14-day rebuild. It works whether your BDC has three agents or thirty, and whether your CRM supports automation or barely handles emails. The discipline is more important than the tooling.
Step 1. Audit current lead-source → callback-window latency
Before you change anything, measure what is broken. Pull the last 60 days of lead events. For each lead, calculate the elapsed time from lead-arrival to first outbound (call or SMS, whichever came first). Group by lead source: Google Ads, Meta, TikTok, third-party (AutoTrader, Cars.com, CarGurus), direct website.
Then look at the 50th, 75th, 90th percentile per source. Almost every audit finds the same shape: median around 45 minutes, 90th percentile past 4 hours, and the 4-hour leads convert at half the rate of the 30-minute leads. That gap is the money.
Step 2. Set the four callback windows
Every lead gets classified into one of four windows the moment it lands:
- Window 1 (0-5 min) — high-intent VDP form fills, phone opt-ins, financing pre-qual submissions. Direct call, no SMS-first.
- Window 2 (5-30 min) — inventory ad clicks that resulted in a lead form, third-party lead marketplaces. SMS-first, phone second.
- Window 3 (30 min - 2 hr) — general contact form, service scheduling, generic "get more info". SMS-first.
- Window 4 (2 hr - 24 hr) — repeat leads (same customer, existing record), non-urgent service inquiries. Email or SMS, no phone unless the customer requests it.
The window classification is fixed by lead source + intent signal at arrival. It is not the agent's judgment call. Judgment calls at 4pm on Friday are how the cadence breaks.
Step 3. Route by intent signal, not by round-robin
Round-robin assignment is the single most common BDC pattern and the single most damaging. A high-intent Window-1 lead should go to the strongest closer available. A Window-4 lead should go to the newest agent. Round-robin gives you the opposite half the time.
The routing rule: Window 1 → top-third of agents by conversion rate. Window 2 → middle-third. Window 3 → bottom-third or newest. Window 4 → new agents for training. Track agent conversion rates monthly; re-tier quarterly.
Step 4. Wire the SOP into CRM automations
Every callback window has a hard deadline. The CRM should surface breached leads to the sales manager, not to the BDC agent. Agents cannot police their own missed windows — the sales manager owns the escalation.
Two automations that catch 80% of the leakage:
- Window breach alert. When a lead sits past its window without an outbound attempt, page the sales manager. Not "notify" — page. SMS to the manager's phone.
- Repeat breach escalation. When the same lead breaches its window a second time, escalate to the GM. This is a routing failure, not an agent failure.
Step 5. AXIOM policy: manager overrides require an audit log
Every SOP has exceptions. Every exception is a place where the cadence starts to erode. The rule: the sales manager can override any window classification, but every override generates an audit row with the lead ID, the original window, the override window, and the reason. Weekly review — if the same reason appears more than three times, the SOP needs updating, not the exception.
Step 6. Rollback plan: previous cadence in cold storage for 30 days
Before you flip the new cadence live, snapshot the existing routing config and store it. If the first week shows the new SOP is worse than the old one, roll back on day 8. Do not let ego make the group live with a bad rebuild for a quarter to "give it a chance." Two weeks is a fair trial.
Failure modes we have seen
- The SOP survives one week. Windows drift by day 5, agents revert to habit by day 10. Fix: the sales manager sits with the BDC for the first two weeks and enforces the windows in real time. The cadence is a habit; habits need a coach.
- Window 1 gets abandoned on Friday afternoon. The strongest closers are with deals; nobody covers the 5-minute window. Fix: a Friday-afternoon backup rotation, one agent designated as Window-1 fallback until 6pm.
- Alerts get ignored. After a week of noise, the sales manager silences the pages. Fix: only page on the second breach in the same day. The first breach is an alert to the CRM; the second is an alert to the human.
What to build first
If you have one day: run the 60-day audit and look at the 90th-percentile latency by source. You will find the money in three of the four sources. The rebuild follows the audit.
How AUTONOMi Runs This Playbook For You
The 14-day BDC rebuild described above — the 90-second first-touch SLA, the multi-channel cadence, the reconciliation report, the daily standup — is what LANE, AUTONOMi's follow-up automation rail, ships as a subscription. LANE stamps the correlation ID at the lead-source layer, dispatches the first-touch call within the SLA (or auto-hands to your BDC's escalation queue), runs the cross-channel cadence with real STOP-word handling on SMS and one-click unsubscribe on email, and writes every touchpoint back to the CRM with the correlation ID intact.
The Friday reconciliation report the playbook calls for is generated at 6am Friday, reviewed by AUTONOMi's AXIOM dealer reviewer against the rebuild's own SOP, and delivered with a compliance signature you can hand a compliance officer. When a rooftop drifts on first-touch SLA, LANE opens the ticket itself — no waiting for the next standup to catch it.
A rebuild that takes a strong internal ops lead 14 days plus follow-through discipline takes AUTONOMi under a week to switch on, with the SOP and reviewer running from day one. If your BDC is currently missing the 90-second SLA on 30%+ of inbound and you know that's the problem, start now — or talk to us if the group scope makes a case-by-case scoping call the better first step.
The BDC cadence is not a technology problem. It is a routing problem with a technology assist.
