Case study: an email campaign to 33,100 contacts

The project began with a straightforward request: contact a large list without losing visibility over the sending process. The client had the contacts, content and schedule. We needed a process to prepare the list, test messages, spread out sending and keep a usable record of each result.

I developed a dedicated mailing tool and used it for several campaigns. The largest used an adjusted dataset of 33,100 contacts. The main mailing ran over 18 days so we could monitor server responses and avoid a sudden increase in volume.

The starting point: a large contact list and little room for improvisation

A list of tens of thousands of addresses is not just a file you import before clicking Send. It is important to know how many contacts are really unique, which lines should be excluded, which address is sent, where the answers arrive, and how a person can unsubscribe.

The requirements included several subject line variants and French or English content. The tool had to select the right message and subject line, and retain enough information to investigate a rejection without unnecessarily exposing the contact list.

What has been developed

The tool was designed for controlled sending, with a monitoring interface and a local database to record statuses. It did not attempt to reproduce every feature of a large marketing platform. It covered the exact process needed for this project.

• import and deduplicate the contact list;
• selection of content according to the language available;
• management of 7 subject line variants;
• a reply-to address separate from the technical sending address;
• pauses between messages and daily limits;
• Detailed log of SMTP acceptances and refusals;
• archiving messages sent when the infrastructure permits;
• opportunity to stop and resume the campaign without starting again from scratch.

Five tests before the main campaign

Before the first live batch, five test messages checked the full process: sender, reply-to address, content, subject line, links, rendering and whether the message appeared in the expected sent folder.

A useful test does not only check if the HTML is nice. It checks what the recipient receives, what the sender sees, and what the system records in case of a problem.

This passage through a small list also enabled the first parsing deviations to be corrected before reaching the main base.

Gradually increasing volume over 18 days

The campaign was sent in stages. Volume increased across days and batches, with an adjusted maximum of about 2,400 accepted messages in one day. This let us monitor immediate errors and retain room to stop.

Sending server accepted 31 040 messages Of the 33,100 addresses processed. The 2,060 immediate refusals correspond in particular to addresses or domains rejected during the SMTP transaction. The calculated acceptance rate is 93,8 %.

What these numbers prove, and what they don't prove

Logs measure the messages accepted or rejected during sending. They cannot establish that 31,040 people received the message in their main inbox. SMTP acceptance can still be followed by a delayed bounce, spam classification or recipient-specific filtering.

I therefore do not publish an opening or conversion rate that has not been verified on a comparable basis. The client reported positive qualitative feedback after the campaigns, but the published technical evidence is deliberately limited to what the logs establish.

The tool was then taken over for a second organization

When this article was published in July 2026, the software had also become the basis for another campaign project. This reuse led to improvements in several areas: a secure public unsubscribe link, bounce handling, adaptive sending rates, stopping on critical errors and a clearer audit log.

This second campaign was then in preparation. It is therefore not presented here as a commercial result. However, it shows that a useful tool must be able to be taken over, cleaned and adapted to a new context without copying the settings of the first client.

When this service can help

A custom emailing campaign may be relevant when the base is important, the standard tool does not cover workflow, multiple languages or variants have to be applied, or the client wants more accurate tracking of the sending steps.

If an existing platform already covers the need, it often remains the best choice. Development becomes interesting when specific constraints begin to cost more time and risk than adaptation itself.

Conclusion: mailings followed at each stage

Monitoring shows what was tested, which messages were accepted or rejected and where the campaign can resume.

To study a comparable project, indicate the approximate volume, origin of the list, the current tool and the expected frequency on the page. email campaign preparation and management.

Frequently asked questions