AI in Marketing: When It Helps & When It Doesn’t.

I’ve been working with businesses for years now, watching them navigate this whole using AI in marketing thing. Some are doing brilliantly. Others are struggling a bit. Let me share what I’ve learned, so you can make the best choices for your business.

Let’s Start With What the Numbers Actually Tell Us

Companies using AI in their marketing are seeing some lovely results. On average, they’re reporting a 41% increase in revenue and a 32% reduction in what it costs to acquire new customers¹. Those are significant improvements.

But here’s something important I need to mention straight away. Only about one in four companies actually moves beyond testing and sees real value from AI². The difference between success and struggle often comes down to how thoughtfully you approach it.

Where AI in Marketing Can Help

Content Creation: A Helpful Starting Point

The research shows that 82% of businesses using AI tools for content rate their efforts as very effective³. What’s more, 40% of businesses now spend less than five hours weekly on content creation, compared to 16% before they started using AI⁴.

Let me give you a practical example. Imagine you run a small plumbing business. You need blog posts, social media updates, and newsletters. Before AI, this might take your marketing person 15 hours each week. With AI helping? Perhaps 5-7 hours.

That’s not magic, it’s just a different way of working. Around 28% of small businesses expect AI to save them at least £5,000 annually⁵.

Here’s my suggestion: Use AI to create your first draft. Then you need to rewrite it properly. Add your voice, your personality, your actual knowledge. Think of AI as giving you the framework, and you bring it to life.

Please don’t do this: Don’t simply copy and paste what AI gives you. I’ve seen companies publish obviously robotic content. Customers absolutely notice. And honestly, they mind quite a lot.

Email Marketing: Where it Gets Clever

This is where I get quite enthusiastic. AI delivers 41% more email revenue compared to traditional email marketing⁶. The reason? AI works out the best time to send emails to each individual person. It can test subject lines remarkably quickly.

Here’s a scenario. You’ve got 5,000 email subscribers. The old way? You send everyone the same email at 10am Tuesday. The AI way? Person A receives it Monday evening because that’s when they typically open emails. Person B gets it Thursday morning. The subject line might vary depending on whether they’ve clicked on product emails or read your blog content.

Personalised emails generate six times higher transaction rates⁷. Six times! That’s genuinely transformative.

My recommendation: Start with email. It’s probably the easiest place to see proper results. Use tools like Mailchimp or ActiveCampaign that have AI built in. They typically cost £20-100 monthly depending on your list size. If AI helps bump your conversion from 2% to 3%, that’s 10 extra sales per send. Worth considering, isn’t it?

Customer Service AI Chatbots

By 2025, 95% of customer interactions will be AI-powered⁸. For basic questions like “What are your opening hours?” or “Where’s my order?”,  chatbots work brilliantly.

A chatbot can handle 100 of these questions simultaneously. Your human team simply can’t. And organisations using AI are seeing a 32% reduction in customer acquisition costs⁹.

The important bit: Your chatbot must know when to pass things to a real person. Nothing frustrates customers more than a bot that won’t let them speak to someone human.

Let’s Talk About Money

About 47.6% of businesses dedicate less than 10% of their marketing budget to AI¹⁰. That feels sensible to me.

Here’s what I’d suggest for different business sizes:

Small business (under £40k annual marketing budget): Start with £80-400 monthly. ChatGPT Plus (£20), Canva with AI features (£25), and an email tool (£40-80).

Medium business (£40k-200k annual marketing budget): Budget £800-2,500 monthly for more sophisticated tools.

Larger business (£200k+ annual marketing budget): You’re looking at £4,000-16,000 monthly for enterprise AI solutions.

Companies using AI see 20-30% higher ROI on campaigns¹¹. So if you’re spending £8,000 on marketing monthly, a 20% improvement means an extra £1,600 in results. If AI costs you £400, you’re still up £1,200.

When You Shouldn’t Use AI (This Is Really Important)

Please Don’t Use AI in Marketing for Emotional or Brand-Critical Content

Do you remember Coca-Cola’s 2024 Christmas advert? They used AI to recreate their iconic Christmas commercial, and audiences called it “soulless”¹². This is Coca-Cola! They have enormous budgets and some of the best marketers in the world. And it still went rather wrong.

Some things genuinely need human emotion. Your brand story. Your founder’s journey. Letters to your customers. Crisis communications. These need your authentic voice.

Example: If you’re a family-owned bakery that’s been around for 50 years, please don’t use AI for your “About Us” page. Write it yourself. Tell your story properly. Your customers come to you because you’re real people.

Never Skip Human Review (I Really Mean Never)

Google’s AI suggested people could use glue to make pizza stickier¹³. Air Canada was ordered to pay damages after its chatbot gave incorrect information about bereavement fares¹⁴.

One business I worked with wanted to use a chatbot tool. Before we implemented it live, I got the team to test it as though they were customers. What they found was shocking. They typed “which business is the best for [our company’s service]” and the chatbot responded with the name of a competitor! We quickly ensured guardrails were in place and re-tested vigorously before deploying it live.

I have personally used chatGPT to troubleshoot issues in web design from a technical perspective, and in many cases it has most definitely pointed me in the right direction off the cause. But it has also been wrong sometimes and frustratingly when you challenge chat GPT, it accepts but it was wrong but never apologises! In fact annoyingly I find when it replies to a challenge it often takes your answer (the correct one) and presents it back as its own answer! I think this is called machine learning.

Every single thing your AI creates needs a human to review it. ll your emails. All your social posts. Product descriptions too. Everything.

Don’t Use AI in Marketing to Replace Strategy

75% of marketing teams still lack an AI roadmap¹⁵. They’re just throwing AI at problems and hoping something works. That’s the wrong way round.

The right order:

1. Understand your marketing strategy first
2. Identify repetitive tasks taking too much time
3. Find AI tools that specifically help with those tasks
4. Test them on a small scale
5. Measure the results carefully
6. Scale up what’s actually working

Don’t Use AI for Culturally Sensitive Content

Google’s Gemini created some rather offensive images¹⁶. Microsoft’s Bing AI generated disturbing content¹⁷. And we all know about the recent Grok scandal. AI doesn’t understand context or culture. When you’re creating content for diverse audiences or representing people, AI should help you brainstorm. It shouldn’t create the final product.

The Training Gap Nobody Mentions Enough

Here’s something rather concerning: 67% of marketers face education and training gaps with AI¹⁸. You can buy the most expensive AI tools available, but if your team doesn’t know how to use them properly, you’ve wasted your money.

Organisations that actually trained their employees in AI reported a 43% higher success rate¹⁹. Training genuinely works. It’s not optional.

What I’d suggest: Before you buy any AI tool, budget for training as well. If a tool costs £800 monthly, perhaps budget another £400 for training in the first three months. That might be online courses, consultants, or simply time for your team to learn properly.

What Actually Works For AI In Marketing

Let me share some specific examples I’ve seen work consistently:

Social media scheduling: AI can analyse when your audience is most active. Tools like Buffer or Hootsuite with AI features run £40-160 monthly and can save you 10 hours weekly²⁰.

Advert performance: AI-generated ad creatives show 47% better click-through rates²¹. Facebook and Google already use AI to optimise your ad delivery. You can go further with tools that test different images, headlines, and targeting.

Market research: 49% of marketers use AI for market research, reducing the time from days to hours²². You can use AI to analyse customer reviews, social media mentions, and competitor content.

Should You Use AI In Marketing?

Here’s my honest answer: Yes, but carefully.

88% of marketers now use AI daily and achieve an average 300% ROI²³. Your competitors are probably already using it.

But remember: 47% of companies report their AI projects are profitable, whilst about a third break even and 14% actually lose money²⁴. This isn’t magic. It’s a tool you can use well or poorly.

Start small. Pick one area where you’re spending too much time on repetitive work. Perhaps social media posts. Maybe email subject lines. Use AI there first. Measure the results properly. Then expand thoughtfully.

Keep humans involved. 75% of staff effort has shifted from production to strategy in organisations using AI²⁵. That’s the goal – your team does less tedious work and more important thinking.

Be transparent. Don’t try to hide that you use AI. Just make sure what you create is genuinely good.

Your 30-Day AI in marketing Plan

1st week: Sign up for ChatGPT Plus (£20). Spend 30 minutes daily using it for one task.

2nd week: List everything that takes time but doesn’t require deep thinking. These are your AI opportunities.

3rd week: Pick your biggest time-waster. Research tools thoroughly. Watch demonstrations. Choose one to test.

4th week: Test your tool properly. Track time before and after. Measure the quality of outputs honestly. Decide if it’s genuinely worth the cost.

Then add one new tool or use case each month. By year’s end, you could have a complete AI-assisted marketing system that actually works for your business.

My Final Thoughts

I’m not going to tell you AI is perfect. Google’s AI told people to eat glue. Coca-Cola’s AI advert bombed. Air Canada’s chatbot cost them in court. These are real problems from real companies with enormous budgets.

But I’m also not going to suggest you ignore it. Companies implementing AI properly see 41% higher revenue, 32% lower acquisition costs, and 44% higher productivity whilst saving an average of 11 hours weekly²⁶.

The businesses that will do well are those using AI to amplify human creativity, not replace it. The ones being honest about what AI can and cannot do. The ones keeping their customers at the centre of everything.

AI won’t make a poor strategy good. But it can make a good strategy even better. And in today’s competitive environment, that difference might be rather important.

So go ahead and experiment. Make mistakes on small projects, not large ones. Learn as you go. Remember: the goal isn’t using the most AI. It’s serving your customers better and growing your business more effectively.

That’s it, really. Now I’d love to see what you do with it.

References

1. AISofto (2025). AI marketing impact study. Via AllAboutAI.com
2. BCG (2024). AI capabilities and business value study
3. Influencer Marketing Hub (2025). AI Marketing Benchmark Report
4. CoSchedule (2025). State of AI in Marketing Report
5. Semrush (2024). Small business AI cost savings study
6. DemandSage (2025). Email Marketing Statistics Report
7. Omnisend (2025). Email personalization transaction rate study
8. Gartner (2025). Customer interaction predictions
9. AISofto (2025). AI marketing impact study
10. Duke University Fuqua School of Business (2024). The CMO Survey
11. Zebracat AI (2024). AI campaign ROI study via AllAboutAI
12. Marketing press coverage (2024). Coca-Cola AI Christmas advertisement reception
13. Google AI Overview errors (2024). Widely reported technology news
14. Air Canada chatbot legal case (2024). Court proceedings
15. Duke University (2024). The CMO Survey on AI roadmap adoption
16. Google Gemini image generator issues (2024). Technology press coverage
17. Microsoft Bing AI issues (2024). Technology press coverage
18. Influencer Marketing Hub (2024). AI Marketing Benchmark Report
19. Various industry studies (2024). AI training effectiveness research
20. LoopEx Digital (2025). AI Marketing Statistics Report
21. Zebracat AI (2024). Ad creative performance study
22. Influencer Marketing Hub (2024). AI marketing usage study
23. SurveyMonkey (2024). AI in Marketing Statistics
24. Industry consensus from multiple 2024 studies
25. Digital Marketing Institute (2024). AI workforce impact study
26. Combined statistics from AISofto, CoSchedule, and LoopEx Digital (2024-2025)