Generative AI for marketing: where it helps, where it's overrated, and how to actually use it

Generative AI for marketing is useful across five jobs: research, content, personalization, ad creative, and ops. Most teams only use it for one (content), and even there, the results are mixed. Seventy-eight percent of organizations now use AI somewhere, but only about 5% see a real impact on profit (McKinsey State of AI 2025). The gap between “I have a ChatGPT subscription” and “this is changing how we grow” is enormous. This post maps where generative AI genuinely delivers, where it’s overrated, and how to close that gap. It’s one of the core jobs in the one-person marketing stack, part of the bigger picture of running marketing with AI.

What generative AI actually changes in marketing

When people say “generative AI in marketing,” they usually mean “ChatGPT writes my blog posts.” That’s like saying a car is a radio because you only use it to play music.

Generative AI (software that creates new text, images, or code based on what you ask for) touches five distinct marketing jobs: research, content, personalization, ad creative, and operations. The teams getting real results use it across several. The rest are fiddling with prompts in one corner and wondering why nothing changed.

HubSpot’s 2026 State of Marketing found that 87% of marketing teams use AI somewhere. That sounds impressive until you see the next number: only 26.5% report a big productivity increase. The rest? “Slight to moderate.” Which is a polite way of saying nothing really changed.

My take: The problem is almost never the tool. It’s that teams plugged AI into an existing workflow instead of redesigning the workflow around it. You can’t bolt a jet engine onto a bicycle and expect a plane.

The McKinsey State of AI 2025 report puts it in harder numbers: out of 1,933 organizations surveyed, just 109 (about 5.5%) attribute more than 5% of their earnings to AI. Marketing and sales is actually the function most likely to report revenue gains. But “most likely” still means a small minority.

This is the adoption-value gap. Closing it is the whole point of this guide. I mapped the full workflow-redesign approach in how to use AI in digital marketing, which covers all five areas end to end.

Five marketing jobs where generative AI delivers

I’m going to walk through each one with a concrete example, so you can see where it might fit into your day. If you want a deeper list with specific tools, I wrote a separate post on the best AI tools for marketing. For companies using AI for marketing with real case studies and outcomes, that post covers the brand stories. And for 15 examples of artificial intelligence in marketing with real costs and effort levels, that’s a separate guide. If you’re selling to other businesses, the dynamics shift enough to deserve their own breakdown — here’s how AI in B2B marketing plays out differently.

Research and insights

This one is underrated. Before generative AI, competitive research meant weeks of reading, tagging, and summarizing. Now you can drop three competitor landing pages into Claude or ChatGPT and get a structured comparison in five minutes. Not perfect, but a solid starting draft.

What works well: summarizing survey data, scanning competitor messaging, drafting customer personas from interview transcripts, pulling out patterns from reviews. Basically anything where you’re reading a lot and looking for themes. If positioning is the job, the guide on AI for product marketing walks through how to use AI for messaging, launch planning, and competitive positioning end to end.

Harvard Business Review reported that AI is genuinely changing market research. The speed advantage is real. The danger is trusting the summary without checking the source material. AI compresses information, and sometimes it compresses the important stuff out. If you want to see this in action, I walk through the full prompt sequence for using AI research to create a full marketing plan in an afternoon.

Content creation and repurposing

The obvious one. And the most overhyped.

GenAI content marketing gets the most attention, but here’s the honest picture: HubSpot found that 71% of marketers say AI helps them create more content. In the same survey, 52% say AI has made content less effective overall. More content, less impact.

The real win isn’t “AI writes the blog post.” It’s repurposing. One long piece turns into 10 social posts, an email, a video script, and a summary thread. That’s using generative AI for content creation the right way. The drafting is the cheap part. The human edit is what makes it work. For a concrete example of what this looks like end to end, the guide on AI for affiliate marketing walks through the full workflow from keyword research to published review.

If you want the full system for this, I broke it down in my guide to AI-enhanced content marketing. Short version: AI handles the 80% that’s grunt work. You protect the 20% that sounds like you. And when the volume of AI-generated content starts piling up, you’ll need a system for managing digital assets with AI so nothing gets lost or duplicated.

Personalization at scale

Writing one email is easy. Writing 50 versions of it for 50 different segments used to take a week. GenAI makes it an afternoon.

Bain found that AI-powered ad campaigns drive 10-25% higher return on ad spend (the money you get back per dollar you spend on ads). JPMorgan Chase tested AI-written ad copy through Persado and saw click rates jump up to 450% in pilots. In one mortgage test, human-written copy brought in 25 applications per week. The AI version brought in 47.

But personalization has a bottleneck, and it’s not the AI. Salesforce’s 2026 report found that 98% of marketers hit obstacles with personalization. The problem is almost always data. You need your customer data organized and accessible before the AI can do anything useful with it. If your data lives in five different spreadsheets, no AI tool will fix that.

Ad creative and testing

This is where gen AI marketing picks up speed. Generating 30 ad variants used to take a design team two weeks. Now it takes a morning.

Bain reported that campaign time-to-market dropped by 50% in companies using genAI for creative. That’s real. You test faster, learn faster, and kill bad ads sooner.

The catch? Brand consistency. Without clear guardrails (a set of rules about what your brand looks and sounds like), AI-generated creative starts to drift. One week your ads look cohesive. The next, they look like three different companies.

If you’re exploring AI tools for social media marketing, this is where the biggest time savings show up. The same principle applies to using AI for influencer campaigns: AI handles the creative variants and audience matching, but someone needs to watch the output.

Marketing operations and reporting

The least glamorous use case. Often the highest ROI.

Summarizing a weekly dashboard. Drafting campaign briefs from a template (or using campaign generators to build the whole skeleton). Building a content calendar. Writing meeting recaps. None of this is exciting. All of it eats hours.

I started using AI for ops before I used it for content, and it saved me more time. The trick is that ops tasks are repetitive and structured, which is exactly what AI handles best. There’s less creativity required, so there’s less risk of the output being wrong or off-brand.

If you’re running a small team, this is where I’d start. Not content. Not ads. Ops and workflows. The payback is fastest because the tasks are the most repetitive. And once you’ve automated the boring stuff, you’ll know when it makes sense to move beyond prompts into agentic AI vs generative AI territory.

Where generative AI is overrated in marketing

I’ve been positive so far. Time to get honest about the limits, because the marketing world has a real problem with treating generative AI like it solves everything. For a structured look at the trade-offs of AI in marketing, I wrote a dedicated breakdown. Here’s the highlight reel.

The content flood problem

The paradox: everyone got the same tool, everyone created more, and nobody reads more.

HubSpot’s data again: 53% of marketers now struggle to make content stand out. That’s a direct consequence of everyone using AI to produce more. The floor rose. The bar didn’t.

Bobby Jania, CMO at Salesforce, put it bluntly: “We are using the most powerful technology in history to send more one-way spam, faster.” That one stung because he’s right.

If you’re worried about whether AI-produced content will hurt your search rankings, I looked into whether AI content is bad for SEO. Short answer: it’s not the AI that’s the problem. It’s publishing without a human edit.

The clicks-vs-leads trap

This stat from a 2025 study in the Journal of the Academy of Marketing Science is one of the most interesting things I found:

AI-generated ads produce 3x higher click-through rates than human-created ads. But human-created ads generate 9.5x more leads.

Read that again. AI gets clicks. Humans get customers.

AI is great at the surface layer. Catchy headlines, clean copy, tested formulas. But the depth layer (the thing that makes someone actually fill out a form or pick up the phone) still comes from understanding the customer in a way AI can’t replicate yet.

The consumer trust collision

Almost nobody talks about this one, and it’s the most important.

Gartner surveyed 1,539 consumers and found that 50% prefer brands that avoid using genAI in customer-facing content. Only 7% trust a brand more when AI is visible (Klaviyo/eMarketer). Thirty-one percent trust it less.

Meanwhile, CMOs are allocating 15.3% of their marketing budgets to AI. The money is going in. The consumer trust is going out.

Emily Weiss, Senior Principal Analyst at Gartner, said it well: “Marketers should treat GenAI as a trust decision as much as a technology decision.”

My take: Use AI behind the scenes. Let it do the research, draft the brief, test the subject line. But the customer-facing layer? That should still feel human. Because your audience can tell. And research confirms they’ll walk away if it doesn’t. There’s a flip side to this trust question too: AI search engines are already telling people about your brand, and AI reputation management is how you make sure what they say is accurate.

The 95% failure rate

This last one is the cold shower. Deep Marketing’s analysis found that 95% of genAI marketing projects produce zero measurable impact on profit. Not negative ROI. Zero. As in: nothing changed.

The reason isn’t that AI doesn’t work. It’s that most teams automate a task without redesigning the workflow around it. They make one step faster and don’t change anything else. The bottleneck just moves.

How generative AI fits into marketing strategy

Knowing where AI helps and where it’s overrated is step one. Figuring out how to actually implement it is step two. Here’s how I’d approach it if I were starting today.

Start with one high-repetition task

Don’t build an “AI strategy.” Pick the task that eats the most time and involves the most repetition. For most small marketing teams, that’s either reporting, email drafts, or social repurposing.

Run it for two weeks. Measure the hours saved. If it works, systematize it (write the process down so anyone on the team can follow it). Then pick the next task.

If you need a structured approach, I wrote a full guide on implementing AI step by step and a genAI-specific guide on implementing generative AI step by step. Or if you’d rather talk it through, that’s what my 15-minute spars are for. The short version: one task, one tool, two weeks, then decide.

Fix your data first

Gartner’s 2026 CMO Spend Survey found that only 30% of marketing organizations are ready to scale AI. The bottleneck? Data. Not budget, not tools. Data.

Salesforce found that only 51-58% of marketing teams have complete data access across their own functions. Think about that. Half of marketing teams can’t even see all of their own customer data in one place.

Before you buy another AI tool, ask: can I pull my customer data into one place? If the answer is no, that’s your first project. Not AI adoption. Data cleanup. For help figuring out the right platform, check out my guide on choosing the right AI platform. And if you want to see how all the pieces fit together layer by layer, the gen AI stack guide breaks it down to five jobs and one tool per job.

Three stages, not a big bang

I’ve seen teams try to “go AI-first” overnight. It doesn’t work. The barriers to AI adoption are real, and most of them are organizational, not technical.

Stage 1: Experiment. One person, one tool, one task. No committee. No strategy deck. Just try it and measure.

Stage 2: Systematize. Write the workflow down. Add quality checks. Train one more person. Now it’s a process, not a person.

Stage 3: Scale. Connect tools. Measure across teams. This is where you start thinking about building a generative AI workflow that runs across the whole marketing function.

Most teams should spend 2-3 months in Stage 1 before even thinking about Stage 3. The companies that skip to Stage 3 are the ones in the 95% that see zero impact.

The mindset shift

The 6 hours a week that genAI saves you shouldn’t go to creating more content. They should go to the work AI can’t do: talking to customers, building the strategy, writing the thing only you can write.

Andrew Fried, CMO of Mint Mobile, said it perfectly: “AI isn’t the strategy. AI is a means to an end, or a way to move faster, more efficiently, or broaden creativity.” The strategy is still your job.

65% of CMOs expect AI to dramatically change their role in the next two years. But only 32% think they need to learn significantly new skills. That’s a dangerous gap. You can’t expect the job to change while assuming your current skills are enough.

Gartner predicts that by 2027, lack of AI literacy will be a top-three reason CMOs get replaced at large companies. The good news: you don’t need to become a developer. You need to understand what AI can and can’t do, how to write a good prompt, and how to tell whether the output is good enough. If you want a head start, this cheat sheet for AI prompts covers the six patterns that handle most daily work. And for structured AI prompts for marketing specifically, that guide covers the 4-part structure that stops generic output. That’s a skill set you can build in weeks, not years.

The tools that matter (and how to pick)

I’m not going to give you a roundup of 30 tools here. I wrote that in the best AI tools for marketing, and the best AI tools for marketers narrows it to six picks with real costs. What matters more is the decision framework.

Match the tool to the job:

The cost reality: Deloitte’s Q4 2024 enterprise survey found that 73% of organizations meet or exceed ROI expectations from AI initiatives. Median payback is about 4 months. Content-heavy teams often see payback in under 3 months.

The decision rule: Don’t adopt a tool until you can name the specific task it replaces. “We need an AI tool” is not a reason. “We need to cut our email drafting time from 4 hours to 1” is.

If you’re looking for free AI tools for digital marketing, start there. Once you’ve outgrown the free tier, you’ll know exactly which paid features you actually need. And if you want help figuring out which tools fit your specific setup, I built an AI audit checklist that walks you through it.

For agencies specifically, the tool decisions are a bit different because you’re managing multiple client brands. The guardrails and brand consistency piece matters more.

How I can help

So you know where genAI helps in marketing and where it doesn’t. The honest next question is: where does it fit for your team?

That depends on your current workflow, your data situation, your budget, and what you’re actually trying to grow. The relationship between generative AI and marketing is different for every team. Those are the questions I help founders and marketing leads work through. If you want a quick read on where to start, I do 15-minute spars, no pitch. Just an honest conversation about what’s worth trying and what’s not.

FAQ

How is generative AI used in marketing?

Marketing generative AI is used for five main jobs: research and competitive analysis, content creation and repurposing, personalization (writing email variants, ad copy for different audiences), ad creative testing (generating visual and copy variants fast), and marketing operations (summarizing reports, drafting briefs, building calendars). Most teams start with content, but research and ops often deliver faster payback. For 15 specific AI marketing examples with real tools and outcomes, I put together a separate list.

Will generative AI replace marketers?

No. It replaces tasks, not roles. The operators who learn to use it well become more valuable, not less. McKinsey’s framing is right: this is augmentation, not replacement. The things AI handles well (repetitive drafting, data summarization, variant testing) are the things most marketers don’t love doing anyway. The parts that need a human (strategy, brand judgment, customer empathy) are exactly the parts that make marketing interesting. Andrea Brimmer, Chief Marketing Officer at Ally Bank, put it this way: she uses AI for media and efficiency, but holds back on the creative and strategic work. That’s the pattern playing out across the industry.

What’s the ROI of generative AI in marketing?

It depends on how you use it. Deloitte found that 73% of organizations meet or exceed ROI expectations, with median payback around 4 months. But McKinsey’s data says only about 5.5% of organizations see a real impact on profit at scale. The gap is workflow redesign: teams that just add AI to an existing process see small gains. Teams that rebuild the workflow around AI see large ones. Start with one task, measure the hours saved, and expand from there.

What are the best generative AI tools for marketing?

It depends on the job. For writing and research: ChatGPT or Claude. For images: Midjourney or DALL-E. For email personalization: Klaviyo or HubSpot’s AI features. For automation: Make or n8n. I wrote a full breakdown with pricing and use cases in my guide to the best AI tools for marketing.

How do you start using generative AI in marketing?

Pick one task that eats a lot of your time and involves repetition. Common starting points: email drafts, social media repurposing, or weekly reporting. Run it with one AI tool for two weeks. Measure the time saved. If it works, write the process down and expand. If it doesn’t, try a different task. Don’t start with a strategy deck. Start with the boring stuff. For the full playbook, read my guide on implementing AI step by step.