Speed to dial is the whole game

Agencies argue about lead quality constantly and about response time almost never. That is backwards. Lead quality is largely bought; response time is entirely operational, costs nothing to change, and is usually the difference between two desks getting completely different results from identical records.

This article does not cite a study. There are widely circulated statistics about lead response time, most of them from outside insurance, most of them repeated without anybody checking the original. You do not need them, because the mechanism is legible without a citation and because the only number that should change your behavior is the one from your own floor. So: the mechanism first, then how to measure it yourself.

Why minutes matter, mechanically

Four things happen to a prospect in the minutes after they submit a form, and all four work against you.

Intent decays. The prospect filled out that form because something prompted them: an ad they saw, a bill that arrived, a conversation with a family member, a health scare. That prompt has a half-life. Ten minutes later they are making dinner and the thing that made coverage feel urgent has receded. You are no longer calling someone who wants to talk about insurance; you are interrupting someone who did, an hour ago.

Context is lost. Call within a minute or two and you can open with the thing they just typed. "You asked about burial coverage around fifty dollars a month" is a sentence that lands completely differently when the prospect remembers typing it. Call the next morning and you have to spend the first thirty seconds re-establishing who you are and why you are calling, which is the exact window where people hang up.

The phone is still in their hand. This is the most underrated one and it is purely physical. Someone who just submitted a form on a phone is holding that phone. The same person two hours later has put it on a counter, in a bag, or on silent. Reachability is not a fixed property of a prospect; it is a property of a moment, and the moment right after a submission is the best one you will ever get.

Competition, where it exists. If the record was shared or resold, other agents are dialing it too, and the ordering of those calls is decided by whoever is fastest. Even on exclusive records, the prospect may have filled out more than one form, and you are still in a race you cannot see.

What a realistic first-touch cadence looks like

Speed to first dial is necessary and not sufficient. One fast call that goes to voicemail and is never followed up is barely better than a slow call. The shape that works is fast first contact plus a disciplined cadence that front-loads attempts into the window where reachability is highest, then tapers.

• Attempt one, immediately. Not when an agent frees up. The lead should be in front of somebody, or in a dialer, within seconds of arriving. This is the attempt that does most of the work.
• Attempt two, within minutes. A missed first call is frequently a prospect who did not recognize the number, not a prospect who declined. A second attempt shortly afterward catches a meaningful share of them.
• Attempt three, the same day, different hour. Someone who cannot talk at 2pm may be free at 6pm. Varying the hour matters more than varying the day.
• Then taper, and vary the daypart. Subsequent attempts spread across several days, each at a different time of day, in the prospect time zone rather than yours.
• One channel is not a cadence. Where you have consent for it, a text or an email between call attempts gives the prospect a way to respond on their own terms and makes the next call a returned call rather than a cold one.
• Have an end. A defined number of attempts with a defined disposition at the end, so leads do not sit half-worked forever and your reporting can tell "we never reached them" from "we never really tried".

The exact numbers are yours to set, and they should come from your own data rather than from an article. What is not negotiable is that the cadence is defined, written down, and the same for every agent, because otherwise your contact rate is an average across several different processes and tells you nothing.

How batch delivery destroys contact rate

Some lead vendors deliver in batches: a file in the morning, or a drop every few hours. This is convenient for the vendor and it is quietly destructive for the buyer, for a reason that is worth spelling out.

Consider a batch that arrives at 9am containing leads submitted between 5pm yesterday and 9am today. The newest record in that file is a few minutes old. The oldest is sixteen hours old. Your agents work the file top to bottom, so the average record gets dialed hours after submission, and some get dialed most of a day late.

Now notice what this does to your reporting. The batch has a single average contact rate, and that average is made of two completely different populations: a handful of records dialed while the intent was still warm, and a majority dialed long after. You cannot see the difference, so you conclude the leads are mediocre. They were not mediocre. They were delivered in a way that guaranteed a mediocre average.

This is why real-time delivery is not a feature bullet. A lead posted to your endpoint within seconds of submission gives you the option of being fast. A file gives you no such option, no matter how good your floor is. It is the one thing that cannot be fixed downstream.

How to instrument your own speed to dial

This is the practical section. You can have this measured by the end of the week, and most desks that do it are surprised by the answer.

Four rules for making that comparison mean something. Hold the source constant, because comparing fast dials on one vendor against slow dials on another measures the vendors, not the speed. Hold the period constant, because seasonality moves contact rate on its own. Hold the script constant. And use enough leads that a handful of outliers cannot swing the result, which for most desks means weeks rather than days.

Then look at three cuts of the same data. By hour of delivery, which tells you whether your delivery window matches your staffing. By agent, which tells you whether the delay is structural or is one or two people. And by day of week, which usually exposes a Monday problem nobody knew about.

The operational changes that actually move it

Once you can see the number, the fixes are usually mundane and usually cheap.

Delivery in seconds, not batches. This is a vendor question, and it is the one to settle before you optimize anything on your side. If leads arrive in a file, nothing below this line will help much.

Match the delivery window to the staffing, not the business day. Leads that arrive when nobody is on the phone will be dialed hours later by definition. If you cannot staff a window, close it. A smaller number of leads worked fast beats a larger number worked late, and the arithmetic is not close.

Cap to what the floor can actually work. A daily cap that exceeds your capacity does not produce more sales; it produces a queue, and a queue is just a slow batch you built yourself. This is the single most common self-inflicted speed problem.

Remove the human handoff. If a lead lands in a queue that somebody assigns manually, or in an inbox that somebody checks, you have built a delay into the system and its length depends on how busy that person is. Route automatically on the fields already on the record: state, county, vertical, plan interest, budget band.

Let the dialer own the first attempt. Human beings are good at conversations and bad at noticing that a new record appeared while they were on a call. A dialer that pulls new arrivals to the front of the queue removes the entire category of problem.

Dial in the prospect time zone. A lead delivered at 8:15am Eastern from a prospect in Oregon is not a lead you can dial yet. Our records carry the prospect time zone for exactly this reason, and a floor that ignores it generates complaints instead of conversations.

Make the number visible. Put median time to first dial on the same board as calls and closes. Anything measured and displayed improves without anybody being asked to try harder, and nothing improves when it is only mentioned during a bad month.

What this does not fix

Speed does not rescue a bad record. If the phone number is wrong, dialing it in forty seconds does not make it right. If the prospect is not eligible for the product, a fast call reaches the wrong person quickly. Speed multiplies whatever the lead already was, which is precisely why it deserves its own measurement rather than being folded into a general argument about quality.

It also does not replace a cadence. Fast first contact plus one attempt is a worse program than moderately fast first contact plus a disciplined sequence. Both matter, and they are separate things to measure.

On our side, delivery happens on submission rather than in batches, and the delivered_at timestamp is on every payload so you can compute this without asking us for anything. If you deliver into an AgentTech Dialer campaign, there is no integration hop at all, which removes the seconds you would otherwise spend in a queue. The measurement, though, has to happen on your floor. It is your number, and it is almost certainly worth more attention than your next vendor comparison.