Speed to lead is a routing problem

A web form lands mid-morning. It is a weight-loss consult inquiry, from a location that has agents on shift. The first call goes out mid-afternoon. Ask the team what happened and you get four true answers. The lead went to a shared inbox. Somebody assumed the morning agent had it. The morning agent was on inbound. Nobody was wrong, and the lead waited most of a working day.

The instinct is to coach the agents to be faster. But nobody in that story was slow. The lead was late because for most of a day it had no owner.

Speed to lead is time to first attempt

Define the metric precisely. Speed to lead is the time from a new lead arriving to the first attempt to reach them, by agent, location, and queue view. It is not time to first conversation, and it is not time to booking. It measures one thing: how long the lead sat before anyone tried.

Once you define it that way, the causes stop looking like effort problems. A lead sits because:

• It arrived somewhere that is not a queue: an inbox, a spreadsheet, a CRM list nobody sorts.
• It arrived in a queue that several people could claim and nobody had to.
• It arrived at a location that had no coverage at that hour, and nothing moved it.
• It arrived, someone claimed it, and then got pulled onto inbound with the claim still held.

Each of those is a routing decision that was not made. The agent was the last step in a chain that had already lost the time.

Make the lead a routed event

In DialSplice a new web lead in the CRM does not wait to be noticed. It arrives as a routed event and becomes one canonical work item, placed by location, with its first-attempt SLA attached. In the example distribution rules for Northgate, the row reads: New web lead · CRM · Power Hour · Location from the form · first-attempt SLA. The lead is in the Power Hour view as soon as the form is submitted, at the right location, with a clock already running.

Because there is one work item and views are only views, it cannot also be sitting in an inbox waiting for a second person. Nothing is duplicated between Power Hour, Callbacks, Inbound, Coverage, and Overnight.

Give it exactly one owner

An agent claims the lead and holds a single-winner lease. While the lease is held, no one else can take it. When the lease expires or is released, the lead is available again with its clock still running. If the agent who claimed it gets pulled onto inbound, the lease expiring is the system’s way of saying “this is nobody’s again” out loud, rather than letting the claim rot.

Ownership is what makes the number coachable. When speed to lead is broken down by agent, the figure means something because each lead had one name on it. In the illustrative analytics fixture, median speed to lead ranges from 2m 50s for one agent to 6m 10s for another. A manager can look at that and have a specific conversation, because the leads were not shared.

Route around the gaps

Coverage is where speed to lead is usually lost quietly. A location’s agents go to lunch, or a small clinic has no one on shift after mid-afternoon, and its leads wait until morning.

The Coverage view exists for this. Managers move work across locations, and the move is scoped: an agent at Riverside covering Northgate receives Northgate work under Northgate’s scope and policy, including the location’s caller ID, not Riverside’s. The queue row reads “Coverage · reassigned from L. Nguyen”, so the handoff is visible to everyone who looks.

Leads that are still unworked at close become the overnight batch. The batch is assigned automatically in a location-scoped set, with fairness evidence attached, and a manager reviews and releases it in the morning. The lead does not vanish overnight; it waits with a name pending on it.

Check contactability before the attempt, not after

Routing fast is only useful if the first attempt is allowed. A lead that arrives at 7:40 AM into a location where SMS quiet hours run until 8:00 local can be called but not texted. The contactability preflight evaluates consent, quiet hours, attempt caps, and holidays per channel at the moment of the touch. It tells the agent that voice is OK and SMS is blocked until 8:00 local, so the first attempt goes out on the channel that is open rather than waiting for a channel that is not.

Measure from the ledger

Because every claim, preflight, and disposition is a ledger row, speed to lead is computed from timestamps that were written as the work happened. It is not self-reported. The illustrative KPI tile reads “Speed to lead · 4m 12s · −38s wk/wk”. If that number improves, it is because leads had owners sooner, and the ledger can show which ones.

The same rows feed connection rate and schedule rate, so the three numbers agree with each other. An attempt that was never made cannot connect, and a connection that never happened cannot book. When the first number moves, the others tell you whether it was worth it.

What this changes on the floor

• A new lead is in the Power Hour view at the right location before anyone has to notice it.
• One agent holds it. The lease expiring is visible; a forgotten claim is not.
• Coverage moves work between locations under the receiving location’s policy, with the reassignment on the row.
• Unworked leads become a reviewed overnight batch, not a lost afternoon.
• The speed-to-lead number per agent is real because the leads were not shared.