There is a version of AI marketing that is embarrassing. You've seen it: blog posts that read like they were produced by someone who has never met a human, email sequences that open with "As an AI language model, I...", and ad copy so generic it could belong to literally any brand. This is not what we're talking about today.
There is also a version of AI marketing that quietly produces real, measurable business results. It saves time, surfaces ideas that wouldn't have emerged otherwise, and scales efforts that were previously bottlenecked by hours in the day. That gap between the two approaches is enormous — and almost entirely about how you use the tools, not which tools you use.
After months of testing AI across real client campaigns, here is an honest assessment of what's working and what isn't.
The AI Marketing Landscape in 2025
The AI marketing tool landscape has matured significantly. There are now category-specific tools for almost every marketing function: content generation, email optimisation, ad copy testing, personalisation engines, social scheduling, and analytics interpretation. The paradox is that the abundance of tools has made it harder, not easier, to know where to start.
The businesses seeing real results from AI marketing are not the ones using the most tools. They're the ones who have identified two or three high-leverage applications, built repeatable processes around them, and integrated human judgment at the critical decision points. That's the model worth copying.
Content at Scale: Done Right
The most common misuse of AI in content marketing is asking it to write finished articles from a one-line brief. The output is technically coherent but strategically hollow — it has no specific insight, no first-hand experience, no point of view worth reading.
The approach that actually works looks like this: use AI for research and structure, then let a human write the article.
The research phase
AI tools are excellent at rapidly surfacing what's already been said about a topic, identifying the angles that are underrepresented, and generating a range of possible structures. Feed an AI your target keyword and ask it to identify the five most common questions people have, the most contested points in the current conversation, and what a thorough answer would need to cover. That output becomes your editorial brief.
The writing phase
A human writes the article. Not because AI can't produce words, but because readers can tell the difference between writing that comes from genuine experience and writing that doesn't. Your competitive advantage in content isn't volume — it's perspective. Preserve it.
The rule: AI shapes the brief and the structure. A human brings the insight, the specific examples, and the voice. This combination produces content that ranks and content that people actually share.
Personalisation That Works
Personalisation has been a marketing buzzword for a decade, but most implementations amount to "Hi [First Name]," in an email subject line. AI-powered personalisation goes considerably further — but it has to be grounded in real behavioural data to matter.
Two areas where AI personalisation is producing genuine results right now:
- Dynamic email content: Tools like Klaviyo and ActiveCampaign use behavioural signals — pages visited, products viewed, time since last purchase — to vary the content blocks within an email. The AI determines which content variant each recipient sees. Open rates and conversion rates both improve when content is actually relevant.
- Website personalisation: Tools like Mutiny and RB2B allow you to show different homepage headlines, CTAs, or social proof based on the visitor's company, industry, or referral source. A visitor arriving from a healthcare trade publication sees different copy than one arriving from a tech newsletter. Small change, measurable lift.
What doesn't work is fake personalisation: using someone's first name seventeen times in an email, or showing "content picked just for you" that is clearly random. The bar for what feels genuinely personal has risen. If it wouldn't feel relevant in a conversation, it won't feel relevant in a channel.
AI for Ad Creative
Ad creative testing has historically been slow and expensive. You write five variants, run them for two weeks, wait for statistical significance, pick the winner, and repeat. AI has changed two things about this process meaningfully.
First, generating copy variants is now near-instant. You can brief an AI on your audience, the product, the platform, and the goal, and get twenty distinct copy angles in minutes. Not all of them will be good, but surfacing the range of possibilities quickly is genuinely useful. You'll often find an angle you wouldn't have thought of independently.
Second, AI analysis tools (including Meta's Advantage+ and Google's Performance Max) can optimise creative delivery in real time based on performance signals. The machine is better than a human at deciding which creative to show to which person at what time. The human's job is to give it good raw material to work with.
The creative brief is still the most important document in your ad process. AI can execute from a brief faster than any human team. It cannot write the brief for you.
Email Sequences with AI
Email is one of the highest-ROI applications of AI in marketing — and one of the most misunderstood. The mistake is asking AI to write your final emails. The right application is using AI to draft sequences that you then heavily edit for voice and specificity.
A practical workflow: give Claude or GPT-4o your product description, your ideal customer profile, the specific problem you solve, and two or three examples of emails that have performed well in the past. Ask for a five-email welcome sequence. What you get back is structurally sound — logical flow, appropriate length, clear CTAs. What it lacks is your specific language, your unique examples, and the voice that makes your brand recognisable.
Plan for a thorough edit pass on every AI-drafted email. The time saving is still substantial — you're editing rather than starting from a blank page — but the final product needs your fingerprints on it. Readers notice when they get a sequence that sounds like every other sequence in their inbox, and they unsubscribe.
Measuring AI Marketing ROI
One of the places AI marketing efforts fall apart is measurement. Teams add AI tools to their stack, produce more content, run more tests, and then struggle to attribute improvements to specific changes. Before you add any AI tool to your process, define what success looks like in measurable terms.
- Content production: track time-to-publish per article, not just volume
- Email: measure click-to-open rate, not just open rate (it's a better signal of content quality)
- Ad creative: cost-per-result per variant, with a minimum spend threshold before drawing conclusions
- Personalisation: run A/B tests against a control group before scaling
The goal is not to use more AI. The goal is to produce better marketing outcomes with the same resources. Keep that distinction clear and your measurement will naturally point you toward the applications that matter.
What to Avoid
Not everything AI can do in marketing is worth doing. Three things to actively avoid:
Fully automated "AI slop" content. Publishing AI-generated articles without meaningful human editing is not a content strategy — it's noise generation. Google's helpful content systems are increasingly capable of identifying low-value AI content, and more importantly, real readers can tell. Content that exists to fill a publishing schedule rather than genuinely help a reader is a liability, not an asset.
Fake personalisation. Inserting a person's first name into a subject line while sending completely irrelevant content is worse than sending no personalisation at all. It signals that you're performing relevance rather than delivering it. Use personalisation only when you have the data to make it genuinely useful.
Over-reliance on AI for strategy. AI can analyse patterns, surface options, and accelerate execution. It cannot tell you what your brand stands for, what market position you should own, or which customers you should prioritise. Strategic decisions require human judgment because they require values, not just optimisation.
The businesses winning with AI marketing in 2025 are not the ones who have automated the most. They're the ones who have identified where human effort was previously spent on low-leverage, repeatable work — and redirected that effort toward strategy, voice, and genuine creative thinking. That's the right use of the technology. The shortcut-seekers will produce forgettable work at scale. The strategists will produce less content and better results.