How marketing can actually be automated: what works, what doesn't, and what it costs. Written for the people who have to implement it.
A realistic course month by month — with the dip in month two, the three points where most people stop, and the signs that it is holding.
When a contact counts as inactive, what a reactivation looks like that does not read as desperation, and when deleting is required.
How to measure the actual waiting times, which five points cost the most time, and which measures work at each of them — without pressuring the customer.
Three alternatives compared — server logs, consent-free analytics services and self-hosted solutions — with what each one gains and gives up.
How a fixed image recipe builds a recognisable visual language, and what to settle legally and technically before publishing.
What the common stage terms mean, why they do harm without a definition of your own, and how to settle a shared list in an hour that both sides can sign.
Why content moves around while loading, which six causes are behind it, and how to fix them one by one without breaking the layout.
Why most editorial plans get abandoned after six weeks, which five columns are enough, and what a plan looks like that keeps running through interruptions.
The distinction between provider and deployer, which marketing applications fall into which risk class, and the four real obligations.
Why the click price is the wrong basis for comparison, under which three conditions ads pay off, and what minimum budget it takes.
How duplicates arise, which rules govern merging, what can get lost in the process — and how to stop them coming back.
Why classic A/B tests say nothing below a few thousand visitors, which five approaches work instead, and how to assess changes defensibly anyway.
When to ask, how to lower the hurdle, which four questions produce a usable reference — and what has to be settled legally before publication.
What belongs in an AI policy, what expressly does not, and why one page achieves more than a twenty-page rulebook — with a complete outline to adopt.
Why the open rate has measured little since automatic image preloading, and what actually decides whether a message gets opened.
How tool sprawl arises, which four questions to ask per tool, and how to get from twelve to five in three steps without losing data or processes.
What goes into an llms.txt, how it differs from robots.txt and sitemap.xml, how widespread support actually is — and an honest assessment of the effort.
Why internal links achieve more than most external ones, which five rules apply, and how to connect your existing pages sensibly in two hours.
Why most positioning statements never get repeated, which three parts a workable sentence has, and how to develop it over four rounds.
Why industry and size do not make an audience, which five characteristics actually separate, and how to get there in one afternoon.
What an entity is, why consistent details across several sources matter more than any single optimisation, and the seven steps to a clear company profile.
Why choosing the tool is the smallest decision, which four skills actually have to be taught, and what an introduction over twelve weeks looks like.
What MCP does differently from a direct integration, where each one's strengths lie, and the four questions that carry the decision.
Why monthly comparisons are systematically wrong with long cycles, how to match costs and returns in time, and which two proxies report earlier.
Three routes compared — a classic CMS, generated static pages, a site builder — with the four questions that carry the decision.
Ten derived formats from one thorough article — with the difference between reusing and merely splitting up, and a process that runs in two hours.
The six properties cited sources have in common — and the three reasons substantively strong pages still never appear in an answer.
Which four fields stay, which go, why some cost disproportionately much — and the ten technical points that cause abandonment.
Which crawlers exist, how collectors and runtime fetches differ, and which three strategies suit which business model.
How many attempts are defensible, at what intervals, and what separates helpful following up from irritating — with the stop rules.
Why a percentage of revenue misleads, which three quantities actually determine the amount, and how to split the budget across the four items.
How to read the intent behind a query off the results page, which four intents exist, and why the wrong intent never reaches the top.
Four questions that carry the decision — the three middle options between a full price list and silence, and the effect on enquiries.
Why language models produce plausible-sounding false details, which five kinds of detail are most affected, and which measures actually help day to day.
Why four segments usually beat twenty — with the three characteristics worth separating on, and the question of when a segment stops being worth it.
Why lists of adjectives do not work, which six sections a usable style document has, and how a language model can actually use it.
Six starting points from the form field to changing channel — with effort, expected effect and the order to actually work through them in.
When a consent banner is legally required, what measurement remains possible without consent, and what you actually give up.
Fixed spelling rules, a short list of values and one central register — so the analysis still tells you something two years later.
What both routes actually cost — including the items that appear in no quote: onboarding, coordination and knowledge loss.
Who the obligation applies to, what the requirements actually demand, and which ten measures cover most of it.
Address structure, hreflang annotations and language selection — with the six mistakes that make a multilingual site invisible.
Five messages with purpose, spacing and content — why the first has to arrive within minutes, and what must not appear in any of them.
A points model with eight criteria that holds up even with few closed deals — separating fit from behaviour, and using negative points.
Structure, linking and order of a topic cluster — worked through on a real example, with the question of when it pays off and how to tell it is holding.
Answer first, evidence after — with right-and-wrong comparisons, the six phrases that make a paragraph unquotable, and a checklist before publishing.
The step from a single request to a repeatable process — the three readiness conditions and the places a human has to stay.
Sorted by the task to be done rather than by vendor — with the properties that matter per working step, and when switching costs time instead of saving it.
How to recognise AI traffic, how to separate it cleanly from other sources, what the server logs add — and where the measurement honestly ends.
A repeatable procedure with a fixed question list, clean testing conditions and an evaluation that turns into concrete tasks.
Every system cites differently — different sources, different currency, different willingness to name small providers.
Sorted by task rather than by vendor — with an assessment of effort, benefit and risk, and a clear recommendation on which three to start with.
Permissions, logging, scope and the handling of injected instructions — the six questions that have to be answered before the first connection.
Step by step from the empty configuration field to a working connection — with the pitfalls that appear in no documentation.
Three questions decide whether personal data may go into an AI system: contract, place of processing, and use for training.
From 58 to 100: the concrete changes to our own website, with before and after measurements — and the mistake that cost us 24 points.
SPF, DKIM and DMARC explained plainly: what the three records do, the order to set them up in, and why emails still fail to arrive when they are in place.
Sorted by task, each with an explanation of why it is built that way — plus the four building blocks you can assemble any prompt of your own from.
The differences between the revised Swiss Act and the GDPR, and the six points a Swiss business website has to have settled.
The difference between more text and more effect: a five-stage production process and the three warning signs of dilution.
Four types cover most business sites — three you can skip. Where they go, how to check them, and the mistake that quietly invalidates everything.
The break point where most automation fails — with the definition both sides have to sign, the three handover models, and what happens to rejected leads.
What each number answers, how often to collect it and when it misleads — plus the three metrics small companies can drop with a clear conscience.
Section by section, with reasoning instead of rules of thumb: what belongs at the top, and which five mistakes cost the most enquiries.
Search engine and answer engine optimisation side by side — the overlaps, the real differences, and one concrete list of measures for each.
Licence, setup, upkeep and the items that appear in no quote — with a worked calculation for three company sizes and the question of when it starts to pay.
Five signs the spreadsheet no longer holds — with a transparent cost calculation, the limits of both approaches, and a way to move without losing data.
A complete five-step process using free sources, a sound way to judge keywords without volume data, and a clear line for when paid tools start to pay off.
A concrete plan for founders and small firms: what to do in weeks 1 to 12, what is deliberately left out, and how to tell it is working.
Google's actual position on automatically produced content, why detection tools do not work — and the five properties that bad text really fails on.
Eleven formats compared, sorted by buying stage and effort. Which lead magnets actually bring in contacts, which only create work, and why.
The order of adoption decides whether the rest works: which building block comes first, and which four to leave out at the start.
Answer engines cite sources by different rules than search engines rank. What changes compared with classic SEO — and the six measures that work.
MCP is the open standard through which AI systems access your own data and tools. How it differs from a normal interface, and what to settle first.
How Claude and ChatGPT differ in day-to-day marketing work, which system solves which task better, and where each one falls short — with prompts to copy.
