AI in Marketing: The "Whether" Question Is Already Settled

A few days ago, a friend of mine who runs an e-commerce business complained to me.

He said his team had hired a fresh graduate who uses AI every day to write copy, create images, and edit videos — one person doing the work that used to take three. The veteran staffers got nervous and came asking: should we be learning this too?

He said, it's not that I don't want to learn. I just don't know where to start. There are too many tools. A new one launches today, another tomorrow. I can't keep up.

When I heard that, I thought of an AI trends report HubSpot published in 2025. They surveyed more than 1,500 marketers worldwide, and one number stuck with me:

66% of marketers use AI every single day. In the US, that number is 74%.

Let that sink in.

Two-thirds of them use it daily. This isn't "experimenting" anymore. This is mainstream. It's like when smartphones first came out — some people insisted, "I'll stick with my flip phone." Two years later, you couldn't find a flip phone anywhere.

So today, using that report as our map, I want to walk you through where AI marketing actually stands. Seven shifts, one by one.

1. AI Has Gone from "Experiment" to "Infrastructure"

What counts as infrastructure?

Electricity, water, gas. You never say "our company is piloting electricity this year." You just plug in and go.

Inside marketing teams, AI is becoming electricity.

The report says teams that use AI well ship faster, publish more, and convert better. Why? Because they've stopped "testing tools." They're rebuilding their workflows around AI.

So what should ordinary teams do? Two moves.

First, inventory the parts of your work with the most repetitive labor. Weekly reports, competitor summaries, asset organizing — that kind of work should go to the machine first.

Second, honestly score yourself. The HubSpot report includes an AI maturity self-assessment. Once you finish it, you'll know exactly what level your organization is at.

Most people have "touched" AI. Touching it and weaving it into how you work are two different things.

2. Writing Is AI's Number One Use Case

Another number from the report: more than half of marketers use AI to write.

Write what? Emails, social posts, ad scripts, product descriptions. Everything.

Why does writing rank first? Because writing is flexible. One article can be broken down into ten social posts, three emails, one video script. AI turned it from "heavy lifting" into "a starting point."

But here I need to offer a warning.

An AI draft is a 70. Your job is to push it from 70 to 90 — add your own stories, your own judgment, your own voice.

If you publish 70-level material straight out, readers can tell. Machine-flavored writing looks fake at a glance.

So my advice: let AI produce the draft, and you supply the soul. Then test. Take one topic, have AI generate five headlines and three openings, and see which one your readers actually click. The data will tell you the answer.

There's also a lazy trick I use myself: just tell ChatGPT, "Summarize the latest AI marketing news for me every week." It genuinely works. Give it a try.

3. AI Doesn't Just Write Anymore — It Can "See" and "Listen"

This next shift is arriving faster than most people expected.

What does multimodal mean? It means AI no longer handles text only — it can work with text, images, audio, and video at the same time.

One example and it'll click.

Gemini can watch an entire YouTube video and then tell you: your next script could cut in from these angles. Descript is even wilder — when editing video, you don't drag the timeline. You just delete a word in the transcript, and the corresponding segment disappears from the video.

Like what? Like editing an article: delete a typo, and the formatting fixes itself automatically.

Design, too. Canva's Magic features and DALL·E produce thumbnails and product mockups in seconds. Can't design? Doesn't matter. You just need to know how to use it.

And there's one trick I love: drop a blog post into Synthesia or HubSpot's Clip Creator, and it directly generates a fully voiced video.

One content idea becomes an entire campaign — no handoffs, no scheduling in between. That's what multimodal is doing right now.

4. Personalization Has Shifted from "Segments" to "Individuals"

The first three shifts were about how AI helps you "do more." This one is about how it helps you "do it more precisely."

What did personalization used to be? Divide users into a few segments, send each segment one version of an email. At the end of the day, you were still talking to "a crowd."

Now? AI can talk to "one person."

HubSpot's own demand generation team did something: they used AI to scrape client companies' websites, judge company size, analyze content preferences — and then, for each individual user, generate content made only for them.

The result?

Conversion rates up 82%. Open rates up 30%. Click-through rates up 50%.

82%! I double-checked that number. One single move, and conversions nearly doubled.

Why such a huge difference? Because audiences changed long ago. Cookie-cutter content gets swiped away instantly. Show people what they actually want to see, and they'll stop.

How should small teams do this? Look at behavior, not just demographics. Someone who downloaded the starter guide and someone who booked a product demo should receive completely different content next. Let AI handle this segmentation and follow-up.

5. More Tools Isn't Better — They Need to "Grow Together"

The report has another uncomfortable number: marketers use five or more AI tools on average.

Five or more.

But think about it — do more tools mean better results? Not necessarily. The more tools, the higher the switching cost, and the more exhausted people get.

So how do people choose tools? 48% look at peer reviews, 47% try the free version first. Very pragmatic. Nobody picks tools from a launch keynote — people want real feedback.

Which one is used most? ChatGPT, by a wide margin. Fast, flexible, easy to plug into existing workflows.

HubSpot also spotted a pattern: the more seamlessly AI embeds into your existing tools, the more the team actually uses it. They eventually integrated OpenAI outright, connecting ChatGPT directly to their CRM — look up customers, write outreach, build campaigns, all in one interface.

What's the lesson here?

What marketers have never lacked is tools — it's tools that fit naturally into the flow. There's only one criterion for choosing: can it grow into your current workflow? If it can't, no matter how dazzling, it's useless.

6. If You Don't Measure It, AI Is Just a "Feeling of Saving Time"

You've been using AI for months. Your boss asks: how much did we save? How much did efficiency improve?

Plenty of marketers can't answer.

The report says a large number of teams are using AI but not tracking what AI actually delivers. And that's a real problem. When year-end budget season comes, what are you going to bring to the table?

Teams that do track run faster and get budgets approved more easily. The logic is simple: numbers give you confidence.

How to track? Three steps.

Pick a task you've already automated — say, writing blogs or emails — and compare before and after: time spent, open rates, lead quality.

Turn the time saved into a concrete number. "AI saves time" is an empty phrase. "37 hours saved per month" is a bargaining chip.

Finally, tag your AI content and track its performance separately in your CRM. Over time, which types of content AI actually helped with becomes obvious at a glance.

7. The Rules of Search Have Been Rewritten by AI

The last shift may be the most lethal one.

If your SEO still revolves around keywords and Google rankings, I'd urge you to pause and reconsider.

Because the way people look for information has changed.

I'm an example myself. A while back I wanted to check which plants are toxic to cats. I didn't open Google — I asked ChatGPT directly. It didn't give me ten links. It gave me one clear answer.

Ten links became one answer.

Google itself has launched AI search overviews. What sits at the top is no longer web pages, but AI-generated content. If your content doesn't get "cited" into the answer, users may never see you at all.

Invisible means nonexistent.

What to do? Three moves.

Type the questions your customers actually ask into ChatGPT and Gemini, and see whose content shows up in the answers. That's your roadmap.

When writing content, aim for "answers," not "clicks." Clear headlines, bullet points, evidence-backed conclusions — AI loves to cite things that read like they were written by an expert.

And stop churning out ten shallow articles. Go deep in one field and build structured content clusters. AI favors sources that feel like genuine insiders.

Finally, Tying the Seven Threads Together

I've said a lot, but implementation really comes down to seven steps:

Use AI every day — start with small tasks and build the habit first. Let AI carry the heavy lifting of content production. Market to the "person," not the "list." Choose tools that plug into your existing workflow; if you don't know which, start with ChatGPT. Track results, beginning with "how much time saved." Write content that answers real questions, to win AI search. And then, keep learning.

See? None of this is rocket science. It's all grunt work.

But business battles have always been won by grunt work. More than 1,500 marketers have already voted on these trends — with their actual, billable hours. The evidence is sitting right there.

The people who weave AI in one day earlier, and the people still watching from the sidelines — a year from now, they'll be two different species.

That's not a scare tactic. It's what actually happened in the years when electricity, water, and gas first rolled out.