
AI IN MARKETING: LESS MAGIC, MORE CRAFT
Ask marketing departments whether they use AI and almost every hand goes up. Ask what exactly they use it for and the room often falls silent. A few texts, the odd image, now and then a summary. It usually depends on one person who is enthusiastic about it. When that person is on holiday, so is the AI.
That’s not a criticism. It simply shows that trying something out and working with it are two different things.
WHERE AI REALLY HELPS TODAY
The most useful applications are rarely the most spectacular. They are the tasks nobody likes doing but that come up every day:
- Creating product copy in several variants and languages from existing data
- Turning briefings, minutes and research results into a usable form
- Adapting campaigns for different channels and formats
- Pre-sorting enquiries so that sales can focus on the right ones
- Monitoring market and competitor information continuously instead of once a year
That sounds unspectacular. But this is exactly where the time goes that teams are missing for strategy, ideas and customers.
WHY IT SO OFTEN STAYS AT THE EXPERIMENT STAGE
In our experience, it’s rarely the technology. It comes down to three things.
First, the data is missing. AI can only write as well as the information it receives. If product data is spread across five Excel lists and three people’s heads, even the best model won’t help much.
Second, the rules are missing. What can the AI do on its own, where does a person check, who approves? Without this clarity, nobody dares to actually use the results.
Third, the process is missing. A tool you have to operate from scratch every time doesn’t replace any work. Only when AI is built into a fixed process does it noticeably take the load off.
START SMALL, BUT START RIGHT
We almost always advise starting with a single, clearly defined use case. One that comes up often, is measurable and where a mistake wouldn’t be a disaster. Build it properly, with data, rules and process. Only once it works in everyday business does the next one follow.
That’s less exciting than a grand AI strategy. But it means that after six months something is actually up and running.
For implementation we work closely with mds. Agenturgruppe, which has been running automated product data and publishing processes for years. If you’d like to know where the first sensible use case lies in your marketing: let’s spend 30 minutes talking about it.