Automating Lead Follow-Up Without Sounding Like a Robot

Automation should buy you time to have better conversations. Where to automate, where to stay human, and the consent rules that apply to anything sent automatically.

Figures in this article describe the wider market and are drawn from the third-party sources listed at the end. They are not Lead Foundry results, and nothing here is a projection of what any individual broker will achieve.

The short version

Automation in a broker practice has an obvious failure mode and a less obvious one. The obvious one is sounding like a machine. The less obvious one is automating the wrong half, so the broker keeps the administrative work and hands the relationship to a template.

What should be automated?

Everything that is a decision about timing, routing or reminding. None of those benefit from human judgement, and all of them fail when a human is busy.

AutomateWhyWhat it prevents
Lead intake into the CRMNo judgement involvedA lead waiting for someone to notice
Owner assignment on arrivalA rule does it better than a personAn unowned lead
Creating all six follow-up tasksThe cadence should not depend on memoryStopping at attempt two
Appointment confirmations and remindersFully mechanical, high yieldMissed appointments
Reassignment when an owner is awayLeave is predictable, forgetting is notA fortnight of unworked leads
Logging attempts and timestampsNobody logs reliably by handNot knowing your own response time

The one to build first: Cadence creation on arrival. Research across close to 3.5 million leads found 93% of converted leads are reached by the sixth attempt while 50% of leads are never called twice. Automating the six tasks closes most of that gap on its own.

What should stay human?

The content of anything the consumer reads, and every decision about their situation.

The failure to avoid is an automated sequence that continues after a person has replied. It is the single most common way automation damages a relationship, and it happens because the sequence was built to fire on time rather than to stop on engagement.

How do you keep automated messages from sounding automated?

Three things, and the first is worth more than the other two combined.

The tell that gives it away: A message that could have been sent to anyone. If removing the name still leaves a message that works for any recipient, it will read as bulk regardless of how the name was inserted.

What are the consent rules for automated messages?

Automation does not change the legal character of the message. In New Zealand the Unsolicited Electronic Messages Act 2007 requires consent, accurate sender identification, and a functional unsubscribe on commercial electronic messages. In Australia the Spam Act 2003 applies and ACMA has published detailed expectations.

ACMA expects reliable consent records covering the method used to provide consent, the terms of that consent, and the date and time it was obtained, and can require their production on complaint. It also names specific non-compliant practices, including relying on old consent and requiring consumers to log in to an account in order to unsubscribe.

That last one is worth checking in whatever tool you use, because it is a default in some systems and it is a design decision you are responsible for.

What should the automation stop on?

Build the stop conditions before the send conditions. A sequence with no stop rules is a complaint waiting to happen.

What lead follow-up should a broker automate?

Automate intake, owner assignment, creation of the full six-attempt cadence on arrival, appointment confirmations and reminders, reassignment during leave, and activity logging. Those are timing and routing decisions that do not benefit from human judgement and that fail reliably when a human is busy.

What should not be automated in broker follow-up?

The words a consumer reads, any assessment of their situation, and the decision to stop. A message works on a purchased enquiry because it names their specific enquiry, and an assessment of need is regulated advice rather than a workflow step. Sequences must also stop on any reply.

Do consent rules apply to automated messages?

Yes, identically. Automation does not change the legal character of a commercial electronic message. New Zealand applies the Unsolicited Electronic Messages Act 2007 and Australia applies the Spam Act 2003, both requiring consent, accurate sender identification and a functional unsubscribe. ACMA expects consent records showing method, terms and timestamp.

How do you make automated messages sound human?

Include a real detail from the enquiry, meaning the brand, the date and what they asked about, rather than relying on a name token. Write in the voice you speak in, keep it short, and send from a person with a monitored reply address. If the message would work for any recipient with the name removed, it will read as bulk.

Sources