Marketing automation

Your marketing keeps working through the night. In the morning you read what happened.

Newsletters, social, ad accounts, search notifications and content: marketing is the half that creates demand. We automate it so that every step is logged — including every step the machine deliberately did not take. Nothing runs in the dark, and nothing runs without a rule.

NewslettersListsOpensSocial plannerRule setApple AdsGoogle AdsGuard railsIndexNowSitemapsPR & LinkedInDashboard

01What runs by itself in marketing

Five building blocks, all in operation

None of this is an announcement. Every one of these blocks runs today — across platforms of our own and across client projects.

Newsletters

A full send at the push of one button — and figures that hold

The list is built from consent and tags, never from a raw address dump. Sending is paced and capped. Then comes the part hardly anyone does: the measurement is cleaned up.

  • Opens are measured, not estimated
  • Link-scanner clicks are taken out — otherwise every send looks magnificent
  • Block list and bounces are reconciled daily
  • Separate lists for customers and for prospects
Social

A planner that fills its own weekly slots

The week has fixed slots, and the planner finds its own material for them: from the topic pool, from the discussions currently running on the platform, and from the in-house magazine. The AI writes, a rule set checks.

  • Checks cover own links, forbidden mentions and length limits
  • Published to Facebook and Instagram on a rhythm
  • Until it goes out, any post can be pulled with one click
Ad accounts

Apple Ads and Google Ads, steered by rules

Both accounts are driven through their APIs, not through the interface. Keywords, bids and targets follow fixed rules, and what one platform learns feeds into the decisions on the other.

  • The figures from both app stores feed into the same calculation
  • Runs several times a day in the cloud — no machine in the office has to stay on
  • Every intervention is logged with its reason
Search & content

Write once, publish everywhere — and notify

New addresses are reported to the search engines daily and the sitemaps follow along. And one source produces the versions for the press release, LinkedIn, the newsletter and social — each in the register of its channel.

  • Notification goes through IndexNow — and only for what actually changed
  • Four channels, one state of play: the same thought is not reinvented four times
  • Reach figures appear in documents only as far as they are documented
Dashboard

Even the decision not to act is logged

Every step is recorded — and so is every step the automation deliberately did not take, together with the figures that led to holding back. That is the difference between a machine you believe and a machine you trust.

  • In the morning, read what happened overnight
  • Whatever waits for approval sits in one place

What you read here is built and running — it is not a list of what could be built.

02How a post comes about

From the topic pool to publication

The same path applies to the social post, the newsletter and the press release — only the register changes.

  1. 01

    The slot calls

    The week has fixed slots. When one is free, the planner goes looking for material itself — it does not wait for somebody to have an idea.

  2. 02

    The material is found

    From the topic pool, from the discussions currently running on the platform, and from the magazine. Anything already used does not come round again.

  3. 03

    The AI writes

    In the register of the channel and in the voice of the house — not in the voice of a tool. Image and text are made together, not one after the other.

  4. 04

    The rule set checks

    Own links only, no other company named in the text, lengths respected. Whatever fails is not published but sent back — with the reason.

  5. 05

    Published — or pulled

    The post goes out on the rhythm. Until that moment one click is enough to stop it; afterwards it is in the log, together with everything that led to it.

03Guard rails

What the automation is not allowed to do

Ad accounts are the place where unsupervised automation burns real money. So what matters here is not what it can do, but what it is explicitly forbidden to do.

Budgets stay untouched

The automation adjusts bids and keywords, never the budget. How much money may be spent in a month is decided by a human — and only by a human.

Small steps, no jumps

Every change has a limited step size. An automation allowed to double a bid in one go is not a control system, it is a risk with access to an account.

Learning phases are left alone

A young campaign is neither judged nor touched. Intervene during the learning phase and in the end you are only measuring your own interventions.

What a human switched off stays off

The automation switches nothing back on that somebody turned off by hand. It does not know the reason — and a machine that does not know the reason has no business touching the switch.

Each of these guard rails is a rule you can look up, and each is checked by a guard of its own. A guard that has never raised an alarm does not count with us — so each one is deliberately made to trigger once.

04And the other half

Marketing brings the demand. Somebody has to collect it.

Replies to a newsletter land in the same mailbox as replies to an opening message — and are treated the same way there. Both halves share contacts, consent and block list. Each half can still be booked on its own.

Sales automation

Inbox, prospecting, follow-ups, orders, invoices and renewals — the half that turns a reply into a customer.

Both together

One dashboard, one approval principle, one set of contacts. The overview shows what the two halves share and where the handover happens.

05Questions about marketing

What we are asked about this most often

Does the AI simply write our posts and put them online by itself?

It writes them, yes — but between writing and publishing there is a rule set, and until publication one click is enough to pull a post. How much the machine may decide alone is up to you; during the trial run it decides nothing at all.

Why do you clean up the click figures? Higher numbers look better.

Because recipients' mail servers click every link in a message as a precaution to inspect it. Unfiltered, you are mostly measuring scanners — and then you base the next decision on a figure nobody actually clicked. We take those clicks out and document how.

Does the automation touch our advertising budgets?

No. Budgets are explicitly excluded. The automation works on bids and keywords, in limited steps, and leaves campaigns in their learning phase alone. What a human switched off, it does not switch back on.

Where do the newsletter addresses come from?

From consent and from tags held in the contact data — never from a bought-in raw list. Unsubscribes take effect immediately and in every list, and bounces are retired instead of being mailed again.

Do we need a machine running in the office for this?

No. The runs happen in the cloud, several times a day, regardless of whether anyone at your place is logged in. You notice it because something happened overnight — not because a device was humming.

Can we automate marketing only?

Yes. The marketing half can be booked on its own and runs independently. If the sales half is added later, both share the same contacts and the same block list — bringing them together afterwards is not a rebuild.

Tell us what goes out by hand every week at your place.

Newsletters, posts, ads, notifications to search engines: write to us about what repeats at your place. You will get an honest assessment — including when the answer is: this is not worth it for you.

The quickest way is by email or through the contact form.