Insights · 3 min read
How to use AI in marketing: where it saves time, and where it costs quality
Pignatari Marketing Innovations ·
AI is very good at the slow, repetitive parts of marketing: research, first drafts, variants, repurposing and summaries. It is bad at the parts that decide whether people trust you: the promise, the positioning, the numbers and the voice. The teams that get real value from it use it in the first group, keep humans in charge of the second, and write down a few simple rules. Below: where we use AI, where we do not, and the rules we give every team we train.
- Research on your customers' words
- First drafts and variants
- Repurposing content
- Summaries and reports
- Objections from call transcripts
- The promise and positioning
- Numbers, results, testimonials
- The final voice
- Anything personal
- Sensitive data
Where AI saves real time
Research on your own customers. Reviews, comments, survey answers and sales call transcripts contain the exact words your clients use. AI can read hundreds of them and group the recurring problems, doubts and phrases in minutes. In our experience it is the most valuable use, and the most overlooked.
First drafts and variants. Ten versions of a headline, five angles for an ad, three structures for an email. Not to publish as they are, but to have more options on the table before a person chooses.
Repurposing. A webinar becomes a set of emails, short posts and FAQ answers. The thinking has already been done; AI helps change the format.
Summaries and reporting. Turning a week of numbers and call notes into a short written summary for the team, so the meeting starts from the same facts.
Preparing and following up sales calls. A short brief before each call, built from the application answers; a summary of the call for the CRM; a first draft of the follow-up email. A person checks each one before it is used.
Where it costs you quality
The promise and the positioning. What you offer, to whom, and why you rather than someone else. These are decisions. AI will produce something plausible and generic, which is the opposite of positioning.
Numbers, results and testimonials. AI can invent a statistic or a quote that sounds real. In marketing that is a trust problem, and in some markets a legal one.
The final voice. Text that has not been rewritten by a person tends to sound like everyone else's. Your audience notices, even if they cannot say why.
Anything personal. Founder stories, client relationships, delicate messages. If the reader would care that a person wrote it, a person should write it.
Sensitive data. Client lists, contracts, health or financial information: check what your tools do with the data before pasting anything in.
Five rules we give every team
- A person decides the promise. AI can suggest; the founder or the marketing lead chooses.
- Never publish a number, result or quote that came from AI. Every figure in published content has a source someone can show.
- Keep a written voice guide. A page with the words you use and the ones you do not, with examples. Give it to the AI and to new team members.
- Every published line is read by a person. No exceptions for "small" posts.
- Save the prompts that work. A shared library turns individual tricks into a team process.
Example: turning sales-call objections into pre-call content
One example we like, because it connects AI to sales:
- record and transcribe a month of sales calls;
- ask AI to list every objection, with how often it appears and the exact words used;
- pick the three most frequent;
- write each answer yourself, without AI;
- turn each answer into a short video or a paragraph placed before the call.
The aim: fewer repeated objections on calls, and content based on what prospects actually say.
Frequently asked questions
Will AI replace copywriters?
It replaces some of the slow parts of their work. It does not replace the judgement about what to say, to whom, and what not to promise. Good copywriters who use AI well become faster.
Does Google penalise AI-written content?
Not for being AI-written. Google's Search guidance says it focuses on the quality of content, not how it is produced, and its spam policies act against pages generated at scale mainly to manipulate rankings rather than help users. The real risk is publishing generic text that helps no one.
Which AI tools should a marketing team start with?
The tool matters less than the process. Start with one general assistant, a transcription tool for calls, and the rules above. Add more only when a specific task needs it.
How long does it take to train a marketing team to use AI?
The basics, a few sessions. The habits, a month of use with someone reviewing the output. That is why our workshop includes a follow-up session after 30 days.
Want your team to use AI without losing quality?
The AI Marketing Workshop runs in 2–4 live sessions, with a follow-up session after 30 days.
Apply to work with us
