Own tool and client project · Sales 7 min read

Automated client acquisition system

Find companies, gather contacts, approach them and know who reacted — all without manual work. This is the system I use myself, and a narrower version of it is already with a client.

Author: Maksym Kamenchuk, Nexflow — Banská Bystrica Published Updated

So it is clear from the start which is which:

1. The full system is my own tool. I built it for my own outreach and it runs the whole cycle: it finds companies by criteria, scrapes contacts from their public pages, sends a personalised approach, tracks opens and classifies replies as interest or refusal. It is not a theoretical design — it is a tool I actually use.

2. A client in Spain received a narrower version of it. A paid job: a web application that finds companies and gathers the publicly available contacts for them. Neither sending nor reply tracking was part of that delivery — the client does the approaching himself. At his request I do not name him.

I write it split out like this deliberately. Showing my own tool at full strength and letting the reader assume the client got the same thing would be dishonest — and that is exactly the kind of thing you only get to verify after signing.

What work this replaces

Approaching companies by hand is work that in a small company happens between other tasks — which is why it happens irregularly, or not at all:

One good approach takes roughly ten to fifteen minutes this way. At that pace three or four approaches a day is realistic, and the worst part is not the time but the irregularity: when work is busy the outreach gets skipped, and two months later the jobs are missing.

The time estimates describe the manual procedure, not a measured before-and-after result.

The full system — what it does end to end

This is the version I use myself:

  1. Finding companies. From the given criteria — sector, location, type of activity — a list of relevant companies is produced.
  2. Collecting contacts. The system goes through their publicly available pages and extracts company contacts including email addresses.
  3. A personalised approach. The text is based on what that company actually does — it is not a template with a name dropped in.
  4. Sending. The email goes out automatically and the contact is recorded together with what was written to them.
  5. Open tracking. You can see which approaches the recipient opened and which went nowhere.
  6. Reply classification. Incoming reactions are sorted into interest and refusal, so they are not lost among the rest of the mail.

The point of those last two steps is not statistics for their own sake. Without them you cannot tell whether it is the list that is not working or the text — and without that, outreach cannot be improved, only repeated.

What the client received out of it

A company in Spain ordered a narrower scope, and that is what was delivered:

Part of the systemMy own toolDelivered to the client
Finding companies by criteriayesyes
Collecting public company contactsyesyes
Own web interface and loginyesyes
Personalised approachyesno
Automated sendingyesno
Open trackingyesno
Reply classification (interest / refusal)yesno

The client wanted a tool that saves him the searching, while keeping the approach itself in his own hands. That is a legitimate choice — for a first contact with a partner, plenty of companies have good reason to write personally.

Sending and tracking can be added at any time. But I recommend it only once it is proven who is being written to and with what offer — automating a bad approach only means putting off more companies faster.

What stays with a person even in the full version

Answering the answer. The system approaches and sorts the reactions. When somebody shows interest, a person takes over the conversation — automating the sales conversation itself would be a mistake.

Setting the criteria. Who it makes sense to write to is a commercial decision, and it shifts with what works.

Checking the tone. A sample of the generated texts has to be read regularly. The model occasionally drifts from how you want to sound, and without a check nobody catches it.

Respecting a refusal. Anyone who asks not to be written to must be excluded permanently. That is a rule, not an optional feature.

Contact quality. An automatically found email address may be out of date or belong to the wrong department. The list is source material, not a guarantee.

The legal side — stated, not glossed over

A tool that collects contacts and sends cold approaches deserves clear boundaries. I would rather write them out myself than have anyone guess at them:

Nexflow does not provide legal advice. Before deploying cold outreach, check the rules with your DPO or lawyer — I will gladly supply a technical description of what data the system processes and where it reads it from.

What you can and cannot verify

Honestly: there is no public review for this job, the application sits behind a login, and the client does not yet wish to be named. My own tool is, again, mine — it is not a reference, it is a demonstration of what I can build.

What you can verify are my other jobs:

I put it this way on purpose. A project that cannot be verified carries less weight than one that can — and it is better that I say so than that you work it out for yourself.

Frequently asked questions

Does the system send emails automatically?

The full version I use myself does — including open tracking and sorting replies into interest and refusal. The client in Spain was delivered a narrower version focused on finding companies and collecting contacts, without sending. The scope is agreed at briefing.

Is this a mass mailing?

No. The text of the approach comes from what that particular company does, not from a template with a name filled in. That is a fundamental difference in how people react to it — and it is the part that decides the result.

Does the system answer the emails too?

No, and deliberately so. The system approaches and sorts reactions into interest and refusal; when somebody shows interest, a person takes over. Automating the sales conversation itself would be a mistake — that is precisely where a live contact has its value.

Where does the system get contacts from?

From the publicly available pages of the companies it finds — from what companies themselves publish as a contact for business partners. It is not designed to get around logins or technical restrictions on websites.

Is automated cold outreach fine under GDPR?

Working with publicly available company contacts and a relevant B2B offer with a simple unsubscribe is common business practice. The rules do differ by country and by whether the contact is a company or a personal one, and open tracking is additionally processing of data about the recipient. Check it with your DPO or lawyer — Nexflow does not provide legal advice.

What does a system like this cost?

A web application for searching and collecting contacts starts at €1,200. The full cycle including personalised approaches, sending and reaction tracking is above that, depending on scope. On top of deployment, budget for running costs — the search API, the language model and the email service, by volume.

Why do you not name the client?

The client does not yet wish to be named. So I state only that it is a company in Spain. When it becomes possible I will add it — I am not going to invent a reference.

Related services

ServiceBusiness process automation ServiceAI agent for business

Related articles

DecidingAI agent vs chatbot vs automation GuideWhat to automate in a company first

Related work

Client projectE-shop product automation (WooCommerce) Client projectDesktop personal finance application

Looking for clients by hand?

Tell me who you approach and how you find them today. I will tell you what of that can be automated, where the limit is, and what should stay with you.

Get a free consultation →