# Outbound campaigns built and run by the assistant

> The assistant talks to the prospecting database, builds lists on real business criteria, writes the sequences and runs them, with a person keeping control.

- Client: 20-person consultancy
- Sector: Geomarketing and site selection
- Period: 2026
- Headline result: End to end (from targeting to execution, with no list built by hand)
- Source: https://www.praedic.com/en/case-studies/ai-driven-outbound-campaigns

Outbound prospecting rarely stops because it does not work. It stops because nobody has time to prepare it. Building a clean list, checking the contacts, writing a sequence that does not read like a mail merge, scheduling it, following up the replies: that is several days of work for one campaign, and those days do not exist in a company of twenty people.

The outcome is familiar. You run a campaign once, you never run the second, and the conclusion drawn is that outbound does not work.

## Move the effort from assembly to decision

The engagement connected the assistant to the prospecting database, so that targeting could be expressed in business language rather than as a combination of filters in an interface.

You describe who you are looking for the way you would to a colleague: the type of company, the size, the area, the role of the person, and the signals that make now the right moment. The assistant turns that description into a search, builds the list, enriches it, and reports on what it found.

It then writes the message sequence from the real offer and from the context of each recipient, runs the campaign and tracks the replies, which come back into the CRM alongside the rest of the sales pipeline.

## What stays with people

The sales lead keeps control of two things, and they are the right two: who gets contacted, and what they are told. He approves the list before anything is sent, he reads the sequences, and he can stop a campaign in flight.

That limit is deliberate. An outbound campaign puts the company's name in front of people who do not know it yet, and a badly judged message costs more than no campaign at all. What is automated is the preparation, not the judgement.

The regulatory side is dealt with in the same movement: strictly business-to-business prospecting, a subject related to the recipient's role, clear identification of the sender, and opt-out honoured from the first request.

## What actually changes

The gain is not sending more messages. It is that a campaign takes hours to assemble instead of days, which makes it feasible, and therefore repeatable.

Outbound only pays off over time, through iteration. You test a segment, look at what replies, adjust the message, and go again. As long as each iteration costs three days, nobody does the second one. When it costs two hours, the learning loop starts turning.

## What transfers

This work assumes a clear offer and an identifiable segment. If a company cannot say precisely who it is addressing, no tool will make up for that, and the place to start is the positioning work.

It also assumes accepting that outbound feeds the same pipeline as inbound, under the same follow-up rules. A campaign that produces replies nobody handles is money thrown away, with reputational damage on top.