Most reports and surveys of generative AI adoption in business puts the marketing function either at, or very near, the top of the adoption curve. Since the LLMs deal with words and communication, same as marketing, this makes sense, especially when you look at the relentless demand for constant, current content across an insane number of channels that marketing is racing to fulfil these days.

AI tools are used to generate copy for ads, emails, landing pages, blog posts and social content. The brief to generate the copy will often be informed by AI-created personas and insights, AI-curated brand books and voices, and/or AI-queried market, trend, and competitor research. Then. the results of these campaigns and communications are reviewed and analysed by AI, digging out patterns and nuggets of information in collections of data too big for a human brain to be able to process.

In addition to the copy, AI produces on-brand images and basic (for now) videos. Plus, it makes personalisation easier, and more personal – frankly, the biggest challenge these days can be to make it relevant without getting creepy. (This isn’t actually new, Target US was better at winnowing out pregnancies than your nosiest co-worker back in 2012).

Generative AI also massively accelerates and simplifies the localisation process. Translations are now a single prompt away although (speaking from experience) you DEFINITELY need a local, native speaker of the target language, with relevant product/industry knowledge to make sure the output isn’t saying something weird (just as you do for any AI comms). The thing is though, you now only need one quality checker, not an entire department of translators.

This new generation of AI tools help marketers:

• come up with more ideas (think social media posts, Google ads, banner ads on news sites, and so on),
• produce text and images more quickly, and more cheaply,
• make A/B testing easier and faster, and
• (as mentioned) make it possible to come up with more focused, relevant and personalised content for direct channels such as email.

The idea is to free up the marketers to be able to think more strategically, come up with genuinely new innovations and directions that the AI can then support, and to use the data analysis and AI-created reports to make decisions based on previously unavailable information.

The reality for too many people at the moment is that the hamster wheel has simply accelerated.

As always, the tools can help, but are painful when they’re allowed to take over. Human-created or curated quality is always going to win out over quantities of AI slop. It’s just a bit too easy to get overwhelmed and forget that.

Using AI tools (LLMs) in Marketing

Some quick tips/reminders first:

• Give context: paste in your persona, brand voice, or a sample of your existing writing before asking for anything.
• Specify the format – “give me three options” or “write this as bullet points” gets cleaner output than open-ended requests.
• Iterate, don’t regenerate: if the first draft isn’t right, tell it specifically what to change rather than starting over. There is SO much value in the conversation with an LLM.
• Use it as a thinking partner, not a word processor: “What am I missing from this campaign brief?” tends to end up with better, more specific and useful results than “write me a campaign brief.” Plus it means you put in some thinking effort first, and that’s important.

Customer Persona Creation

A halfway decent persona turns generic copy into something that feels like it was written for one specific person. AI can help you build one in minutes rather than scheduling a workshop with people you can’t get hold of for weeks.

Example prompt: I run an executive coaching consultancy targeting intelligent, successful women in their 50s, who are navigating significant life and/or work changes. Based on this, create a detailed customer persona including: demographics, job role or life stage, goals, frustrations, preferred content channels, and the language they’d use to describe their own problems. Give the profile a name and make it feel like a real person.

Remember to iterate…

Ask the AI to write a sample email, social post, or product description “as if speaking directly to [Persona Name].” This is a brilliant way to check whether a persona is actually useful or just fun to play with.

Ongoing: Save your persona as a reusable block of text. Paste it at the top of any new chat as context before asking for copy. Or, better, use Claude or ChatGPT Projects, or Microsoft Copilot Agents (or Gemini Gems, but I find Gemini to not be quite as good at persona stuff as the others) to give you a permanent place to come back to this information.

Search Engine Optimisation (SEO) – Keywords and Content Briefs

SEO is the art and science of being found in Google (organic results). The work is time-consuming, detailed and repetitive; exactly the kind of thing AI handles well. This is an area where Gemini tends to shine (funny that).

Prompt for keyword ideas: I write content about improving your writing craft as an aspiring romance novelist. Suggest 15 long-tail keyword phrases my target audience might search for.

Then: Give me 5 blog article ideas for a post about writing craft focused on the following keyword: quick-read, mountain man romance (yes, that is a thing, not really my thing, but a friend writes it) Note: only use one of the ideas per keyword, otherwise you’ll end up competing against yourself, and potentially end up with both articles ranking too low on Google to be useful.

Once you’ve decided on a post idea: Create a detailed blog post brief on how to build a male lead character in 5 easy steps for the keyword ‘quick-read, mountain man romance’. Include: suggested title, target audience, key points to cover, suggested word count, and internal linking opportunities I should consider.

Ongoing: Expand that Project you’ve created for your persona to include examples of your brand voice (paste in the text of your 5 best previous blogs etc), and major target keywords (romance writing craft, how to write romance, blah, blah, blah). Using that to come up with your content brief each time will give you way better results than a standard query.

Social Media Content

Keeping multiple channels fed with fresh, on-brand content is where people run out of steam fastest. AI won’t replace your voice, but it’s a great place to start, both for topic ideas, and first draft posts.

Take the blog post you wrote from the brief above: Based on this blog post [paste text, attach file, or add URL], write five LinkedIn posts, three Instagram captions, and two Facebook posts for [target audience]. Vary the angle for each — one educational, one opinionated, one with a question to prompt comments, one behind-the-scenes, one promotional. Suggest an image prompt for each of the Instagram entries.

If you’re heading straight for the socials: Write 10 attention-grabbing opening lines for a LinkedIn post about the challenges of finding a new job as a female over 50. Avoid clichés like ‘In today’s world’ or ‘I’m excited to share’.

Ongoing: Use a tool like Zapier, Relay or Make to send a new blog post URL to an AI assistant via webhook, which then drafts a set of social posts and drops them into a Google Doc or Notion page for review. (This isn’t as tech-intense as it sounds, let me know if you’d like me to run through how to set this up in an upcoming post.)

Emails

Example prompt: Write a three-email welcome sequence for new subscribers to AI for Squishy Humans, including recommended interval timing. Email 1 should introduce who I am and what to expect. Email 2 should deliver a quick win or useful tip. Email 3 should make a soft offer or invite a conversation. Keep the tone light, practical, helpful and cheerful and the email copy no more than 3 short paragraphs each.

Competitor and Market Research

I really like Perplexity for this, but all the LLMs can do a decent job.

Example prompt: Summarise the key marketing messages and positioning of these three competitors: Oxford, Cambridge, University of London. What themes and benefits do they emphasise? What are they not saying? Where might there be a gap in the market?

I’m going to stop there, next edition will be customer service and support.