AI content strategy: build the plan around what AI actually changes

An AI content strategy is a plan designed around what AI makes cheap and what it can’t touch. Not “your old content plan, but with ChatGPT bolted on.” A real rethink of what you create, how much, and where the human effort goes.

Most teams skip that rethink. They add AI to the writing step, change nothing else, and wonder why they’re producing more content that performs worse. I wrote about what AI changed about content marketing and the data is clear: volume alone stopped working. Salesforce surveyed 4,450 marketers in 2026: 87% now use AI somewhere in their workflow. But 84% still run generic campaigns. That’s not a tool problem. That’s a strategy problem.

This is the planning layer of building an AI content engine. If you’re ready for the hands-on creation workflow, that’s in generative AI for content creation. If you want the full operational program, see AI-enhanced content marketing. And for the broader picture of AI for digital marketing beyond content, I wrote a full playbook. This post is about the thinking that comes first.

What is an AI content strategy?

The old content model goes: plan topics, write them, publish, promote. AI made the “write” step faster. That’s it. The rest didn’t change.

An AI-driven content strategy flips the question. Instead of “how can AI help me write?” you ask “what should I create now that AI changed the costs?”

Think of it like reorganizing a kitchen. You got a new dishwasher. You could keep washing everything by hand and just let the dishwasher sit there. Or you could redesign the whole kitchen flow so the dishwasher handles what it’s good at, and the chef spends time on the things a machine will never do (picking the menu, tasting, adjusting). That second option is the strategy.

The data says most teams are stuck on option one. CMI and MarketingProfs surveyed over 1,000 marketers: 87% say AI improved their productivity. But only 39% say the content actually performed better. That gap between “faster” and “better” is the whole strategy problem.

My take: If AI made you twice as fast but your content performs the same (or worse), you don’t have an AI strategy. You have a production speed-up pretending to be one. The strategy is what tells you what to speed up, what to slow down, and what to protect.

Why most teams get AI content wrong

This is what I call the bolt-on trap. A team grabs a ChatGPT subscription, starts churning out first drafts, and publishes three times more than before. For a few months, traffic goes up. Then it doesn’t.

Lily Ray tracked 220+ websites that scaled AI content aggressively. The traffic pattern looks like a mountain. Rapid growth from more pages, then a peak, then a steep drop.

Some sites lost 40 to 90% of their search traffic after Google’s updates. The playbook was simple: publish more AI content, rank fast, lose everything.

Why does it collapse? Because volume without quality is a losing bet. HubSpot’s 2026 marketing report found that 56% of marketers say the internet is now flooded with AI content. And 65% of consumers say they’re getting better at spotting and ignoring it.

The teams that don’t collapse have one thing in common. They didn’t just adopt AI tools. They redesigned how they work.

McKinsey’s State of AI report found that only 6% of organizations are genuine AI “high performers.” The single strongest predictor was whether the team redesigned its workflows around AI.

Gartner’s 2026 CMO survey puts a number on it: CMOs now spend 15.3% of their marketing budgets on AI. But only 30% say they’re actually ready to scale it. That’s a lot of spending on tools without the workflow to use them.

My take: The Salesforce CMO, Bobby Jania, said it best: “We are using the most powerful technology in history to send more one-way spam, faster.” That’s what happens when you bolt AI onto an old plan. You get faster spam.

What AI makes cheap (and what it never will)

This is the split that should drive your entire strategy. Write it on a sticky note.

AI makes cheap:

• Research and outlines
• First drafts
• Content variations (same piece in 5 formats)
• Repurposing (blog post to social posts, email, video script)
• Meta descriptions, title tags, formatting
• Social media content from long-form pieces

These tasks used to eat 60 to 70% of a content team’s time. Now they take maybe 10%, especially once you nail the prompt engineering fundamentals.

AI can’t do:

• Having a point of view
• Original examples from your actual business
• Lived experience (“I tried this and it broke”)
• Taste and editing judgment
• Real relationships with readers

A 16-month study by Digital Applied compared AI content with human content. AI content included original research in only 4% of cases. Human content: 38%. That’s a 10x gap in the thing that actually makes content different.

And there’s a ceiling on AI creativity, too. A study published in Scientific Reports (Nature) tested over 100,000 humans against GPT-4 and Claude on creative tasks. AI beat the average human. But the top 10% of humans significantly outperformed every AI model. The best human thinking is still the best thinking, period.

So stop spending human time on what AI does well. Redirect it to the part that becomes your moat.

I spent a long time getting this wrong. I’d use AI to draft a post, then barely edit it. It was fast. It was also forgettable. The posts that actually brought traffic were the ones where I spent the saved time adding my own examples, my own opinions, and data I’d actually gathered. Speed isn’t the win. What you do with the time you saved is the win.

How to build an AI content strategy (5 steps)

This is a framework you can actually follow on Monday morning. No theory, no 47-slide deck. Five steps.

Step 1: Audit where your time goes

Track every content task for one week. Write down who does it and roughly how long it takes. Research, outlining, writing, editing, formatting, publishing, promoting.

You’ll find that 60% or more of your time goes to research, outlines, and first drafts. That’s AI territory. The AI audit checklist walks you through this in detail.

Step 2: Define what only you can say

Grab a blank doc. Write down your point of view, your data, your stories. Things you’ve actually seen, done, or measured. This is your “protect list.” AI never touches these.

If you can’t fill a page, that’s a signal. Your content was probably generic before AI too. The fix isn’t more AI. It’s more experience to draw from.

A tool like an AI marketing strategy generator can give you a skeleton. But a skeleton isn’t a strategy. The strategy is the part you add on top: the angles only you would take, the data only you have, the stories only you can tell. Start with the protect list, then let AI fill in around it.

Step 3: Redesign the workflow around the split

AI does research, outlines, first drafts, and repurposing. A human does the angle, the examples, the editing, and the final publish decision.

This is what “workflow redesign” actually means. It’s not a consulting buzzword. It’s literally rearranging who does what.

BCG’s research calls this the 10/20/70 rule. AI success depends 10% on algorithms, 20% on technology and data, and 70% on people and processes. Most teams invert it, spending 80% on tools and wondering why nothing changes.

Step 4: Pick one tool per job and go deep

Don’t collect subscriptions. Gartner found that marketing teams use only 49% of the tools they pay for. Half your tools are just billing you.

One writing tool. One research tool. One repurposing tool. Commit for 90 days before you switch. The best AI tools for marketing page breaks this down by job. If SEO is a big part of your content work, the best AI SEO tools list covers the research and optimization side. If you want to automate the blog pipeline end-to-end (research through publishing), that guide walks through the five stages. And if you want to automate the SEO side, that’s a separate decision from the creation tools.

Step 5: Measure what changed

Not just “pieces published.” That’s a vanity number. Track these four things:

• Time per published piece (total hours, AI plus human)
• Organic traffic per piece (not total traffic, per piece, to catch quality drops)
• Engagement per piece (time on page, how far they read)
• Conversion per piece (leads, signups, sales)

If you’re publishing three times more but traffic per piece dropped, AI is making you mediocre faster. Cut volume, raise quality.

And if your content plan includes showing up in AI search answers, that’s a separate layer on top of traditional SEO that rewards exactly the kind of clarity this workflow produces.

If you want help mapping where AI fits in your specific workflow, I do a free 15-minute spar where we figure it out together. No pitch, just the map.

The measurement gap most teams never close

This is the finding that stopped me. A 2026 study by Digital Applied found that only 19% of content marketers have updated their measurement for AI. Eighty-one percent are flying blind.

Without measurement, you can’t tell the difference between “AI is helping” and “AI is making us produce more mediocre content.” And since volume feels like progress (more posts, more social updates, more emails), most teams convince themselves it’s working.

The minimum viable dashboard is a spreadsheet you check once a month:

The 90-day check is simple. If volume tripled but each post gets less traffic than before, you’re just producing mediocre content faster. That’s the signal to cut back and invest the saved time in fewer, better pieces. You can automate the data collection so you’re not manually pulling numbers every week. But the judgment call (what to do about the numbers) stays human.

Edelman’s trust research adds another layer: only 32% of Americans trust AI. The rest are skeptical. Your audience can feel when content is churned out rather than thought through, even when they can’t explain why.

My take: The teams winning with AI content aren’t the ones publishing the most. They’re the ones who can prove each piece earns its keep. Measurement is the unsexy part of an AI marketing strategy. It’s also the part that separates the 6% who get real results from the 84% who are just moving faster in the wrong direction.

How I can help

You just read the framework. You know where the gap is (workflow, not tools). You know what to protect (your POV, your data, your taste). And you know how to measure whether it’s working.

The hard part is doing it inside your specific business, with your team, your topics, and your goals. That’s what I help with. I’ve spent ten years in growth, and the last three rebuilding how I run content around AI. If you want someone to work through this with you, there’s a free 15-minute call where we map your AI content strategy together. No pitch, just the plan.

FAQ

What is the 10/20/70 rule for AI?

It comes from BCG research on why AI projects succeed or fail. The breakdown: 10% depends on algorithms (the AI itself), 20% on technology and data, and 70% on people and processes. Most organizations spend 80% of their AI budget on tools and neglect the 70% that actually determines success. For content teams, this means fixing how your team works matters more than which writing tool you pick.

What is the 30% rule for AI?

It’s a rough guideline that humans should keep at least 30% of the work. That 30% is the judgment, editing, and quality control layer. AI handles the other 70% (research, drafts, formatting). In content, that 30% is your point of view, the final edit, and the publish decision. It’s the part that keeps your content from sounding like everyone else’s.

Which AI is best for content strategy?

There’s no single best tool. It depends on the job. For research and analysis: ChatGPT or Perplexity. For long-form writing and strategy work: Claude. For briefs and SEO optimization: Surfer SEO or Frase. For repurposing: Claude or Jasper. The tool matters less than whether you’ve redesigned your workflow around it. Most people who are disappointed with AI results are using one tool for everything.

How is AI changing content strategy?

AI makes volume nearly free. That means volume is no longer a competitive advantage. When everyone can publish five posts a day, the question shifts from “how do I produce more?” to “how do I produce what AI can’t?” Original data, real experience, and a clear point of view are the new differentiators. That’s the shift from a production strategy to a growth marketing strategy.

Is AI content bad for SEO?

Not automatically. Google penalizes unhelpful content, not AI content specifically. But 220+ domains have lost 40 to 90% of their traffic after publishing unedited AI content at scale. The editing layer is what separates AI content that ranks from AI content that tanks. I wrote a full breakdown on whether AI content hurts SEO with the data behind it.