AI content marketing in 2026: Why teams aren't getting results they expect

In February 2024, Klarna's CEO Sebastian Siemiatkowski announced that AI was doing the work of 700 customer service agents. Marketing followed. Per Klarna's own press release, AI handled 80% of copywriting, image production dropped from 6 weeks to 7 days, and the company saved roughly $10 million in annual marketing costs. The headcount halved, from around 200 to 100. Siemiatkowski was on every fintech podcast saying AI could already do all the jobs humans do.

By May 2025, he was reversing course. "We focused too much on efficiency and cost," he told Bloomberg. "The result was lower quality, and that's not sustainable." Klarna started rehiring. The numbers had been real. The strategy had not.

This is the story of AI content marketing in 2026, in miniature. The tools work. The savings are achievable. The output is faster. And yet most teams running aggressive AI strategies are quietly walking parts of them back, because the math on the spreadsheet missed something the customer noticed immediately.

Per McKinsey's State of AI 2025 report, 88% of organizations now use AI in at least one business function. Only about 6% qualify as "AI high performers" who attribute meaningful EBIT impact to their AI use. The 88% have the same tools the 6% have. They've been using them for years. The gap isn't tool access. It's something else.

This piece is about what that something else actually is.

What does AI content marketing mean?

AI in content marketing is the use of artificial intelligence across the content lifecycle: research, ideation, drafting, personalization, distribution, and performance analysis. It's not one tool. It's a category of capability. Applied well, it lets a small team produce the volume and personalization that previously required a large one.

McKinsey's research finds that the largest economic value pools for AI sit in customer operations and marketing & sales, contributing to an estimated $2.6 to $4.4 trillion in annual impact potential. But that potential is concentrated. A small group of organizations capture most of it. Almost everyone else is running pilots without scaling them.

This wide-adoption-narrow-value pattern is the most important fact about AI content marketing right now. It explains why teams that have spent two years using AI tools often feel like they're working harder, not smarter. The work has compressed. The strategy has not. Before any tool decision, the program needs grounding in fundamentals: a real go-to-market strategy, clear buyer personas, and sharp product positioning that AI can actually amplify.

How AI content marketing differs from traditional content marketing

Three differences matter; the rest are implementation details.

1. Speed of iteration, not speed of output

The obvious headline is that AI lets a marketer ship more content. The more important shift is that AI lets a marketer test more variations of the same content faster. Headlines, hooks, framings, formats: the cost of producing version B has collapsed. Per Harvard Business Review, the gains from generative AI in creative work come less from automating finished output and more from compressing the cycle between idea and validated experiment.

2. Predictive, not reactive, planning

Traditional content marketing asks what worked last quarter. AI content marketing asks what is likely to work next quarter, given current search behavior, audience signals, and competitive gaps. This shifts the planning conversation from retrospective reporting to forward-looking modeling, which is why so many AI content programs underperform when they're plugged into a reporting cadence built for the old model.

3. Personalization as default, not feature

Generic content is now a competitive disadvantage. McKinsey's research on personalization finds that personalization typically drives a 10 to 15% revenue lift, with company-specific lift spanning 5 to 25%, driven by sector and execution capability. With AI handling the variation cost, there's no longer a defensible reason to send the same email, show the same hero copy, or surface the same case study to every visitor.

Why most AI content marketing programs underperform

Back to Klarna for a moment. The original strategy wasn't wrong because the AI couldn't write copy or generate images. It could. The strategy was wrong because Klarna treated AI as a labor substitute inside an unchanged workflow. The customer service agent disappeared; the customer service expectations didn't. The marketing copywriter disappeared; the marketing standards didn't. When the gap became visible, the system broke.

McKinsey's State of AI 2025 surfaces the single most useful finding for marketing teams trying to avoid the same trap: among the practices that distinguish AI high performers, the strongest predictor of value capture is fundamental workflow redesign. High performers are nearly 3 times more likely than peers to have rebuilt their workflows around AI rather than bolting AI onto existing ones.

This is the entire game. Most marketing teams have done the opposite. They've kept their existing process (brief, draft, edit, review, publish) and inserted ChatGPT into the "draft" step. The result is a workflow that's marginally faster and noticeably worse. Output increases. Quality declines. Because the underlying process wasn't designed for AI, the team ends up spending the time savings on cleanup.

The teams getting results have rebuilt the workflow from scratch. They've moved upstream, letting AI handle research, gap analysis, and structure, then letting humans focus on the parts AI cannot do: judgment, taste, original insight, lived expertise. The publish step looks the same. Everything before it is different.

Old workflow vs. redesigned workflow

What actually works: a workflow redesign, not a tool stack

Here's what the rebuilt workflow looks like, broken into the 4 stages that genuinely benefit from AI.

1. Research and gap analysis

AI is excellent at parsing what already exists in your category and identifying what doesn't. Use it to map competitor content, identify topic gaps, surface emerging questions in your niche, and synthesize findings from primary research you've already done: interviews, support tickets, sales call transcripts. This is where the leverage lives, and it's where most teams underuse AI most badly. Anchor this stage in your market segmentation and first principles thinking so the gaps you find are the ones that matter for your business, not just the ones that look interesting in a spreadsheet.

2. Structure and framing

Once you know what to write about, AI can generate multiple structural options for the piece: outline variations, framing alternatives, narrative arcs. The marketer's job here is to pick. Picking is a skill. It's the part of the work that compounds, because it forces clarity about what the piece is actually for. Teams that skip this stage and let AI both decide and draft end up with content that has no point of view.

3. Drafting

AI drafts well when given good inputs: audience, goal, tone, examples of your existing voice, and the structure chosen in stage 2. It drafts badly when given a one-line prompt and asked to figure the rest out. Most of the "AI content sounds robotic" complaints trace back to this. The system was asked to do a job it wasn't briefed for. Treat the draft as a draft. Not a deliverable.

4. Editorial judgment and original insight

This is where humans earn their keep. AI cannot fabricate the things that make content actually valuable: the unexpected insight from a recent customer call, the contrarian take rooted in real experience, the sentence that lands because the writer has been thinking about this exact problem for 3 years. The editor's job is to add what the model could not. If you're not adding anything in this stage, the content isn't worth publishing. AI didn't cause that problem; the workflow did.

Where AI content marketing works best, by format

Not every format benefits equally. Here's where AI delivers real leverage, where it requires guardrails, and where it should be used sparingly.

Long-form and SEO content

Highest leverage. AI shortens research from days to hours, generates structural options, and drafts cleanly when briefed properly. The workflow above maps directly. This is also where the ranking math has shifted. AI search engines like Perplexity, Google AI Overviews, and ChatGPT Search increasingly answer queries directly rather than sending traffic, which means content has to be cited, not just clicked. That's a different optimization problem, closer to brand awareness than to traditional SEO.

Email and lifecycle marketing

Strong fit. Subject line variants, personalization based on behavior data, and milestone-triggered sequences all benefit from AI's variation cost being near zero. The compound effect on lead nurturing is significant. Nurture sequences become genuinely personalized rather than "first name" templated.

Social content

Mixed. AI handles platform adaptation well (turning one piece into LinkedIn, X, and Instagram variants). It handles original perspective badly. The teams winning on social are using AI for distribution mechanics like repackaging, scheduling, and formatting, while keeping a human firmly in the strategy seat. Social is where audiences sniff out generic AI output fastest.

Demand generation assets

Good fit when paired with real positioning. AI is useful for drafting landing pages, ad variants, and one-pagers, but only against a clear narrative. Without that narrative, you get well-written marketing about nothing. See demand generation and the marketing funnel for the strategic prerequisites.

Technical or regulated content

Caution. AI hallucinates confidently in technical domains and cannot reliably cite primary sources. For technical documentation, regulated content (financial, medical, legal), or anything where factual precision is the product, AI should be treated as a research assistant, not a drafter. Human review is non-negotiable.

How to measure AI content marketing ROI

Engagement metrics like opens, clicks, and time on page were always lagging indicators. They're even weaker now, because AI-generated content can perform well on those metrics while contributing nothing to the business. The 4 metrics that matter:

• Pipeline influence: how much qualified pipeline can be traced back to content created or distributed with AI assistance, compared to baseline.
• Cost per published asset: total cost (tools, labor, editorial review) divided by assets shipped. A useful efficiency proxy when measured against the same quality bar.
• Cycle time: time from brief to published asset. AI's clearest, most defensible benefit. Track this even before tracking ROI.
• Quality drift: a quarterly review of a sample of published content against a quality rubric. This catches the slow degradation that often shows up 6 months into an AI program.

None of these are vanity metrics. Each ties to a real business question. McKinsey's State of AI work makes this point repeatedly: AI value comes from rewiring how companies run, not from the tools themselves, and the teams that measure rewiring outcomes are the ones that capture value.

The 4 most common pitfalls

These show up across every AI content marketing program that underperforms. Klarna hit at least 3 of them.

1. Treating AI as a draft button

If your AI workflow is "prompt, paste, edit lightly, publish," you're using the most expensive part of the model (generation) and skipping the most valuable parts (research, synthesis, variation). The output will be generic. So will the results.

2. Confusing volume with progress

Klarna generated over 1,000 images in Q1 2024 alone. The volume was real. The strategic question of whether all those images were doing useful work was harder to answer. Publishing 4x more content is only a win if the additional content is doing work: ranking, converting, building category authority. Most teams use the AI dividend to publish more of what wasn't working before. Volume is not strategy.

3. Skipping the brand voice work

AI defaults to a generic register: confident, mid-formal, slightly American, light on opinion. If that's not your brand, you have to do the work to teach the system what your brand sounds like through examples, voice documentation, and consistent editorial review. Teams that skip this end up with content that all sounds like the same anonymous LinkedIn post.

4. No fact-checking process

AI hallucinates. It will confidently misattribute statistics, invent quotes, and cite studies that don't exist. Every AI-assisted piece needs a fact-check pass: every statistic verified to its primary source, every quote verified, every claim about a competitor or third party checked.

This is the single most important editorial discipline in an AI program. Per Deloitte's research on generative AI in the enterprise, trust and verification are emerging as the dominant constraints on AI value capture, far more than capability or cost.

What changes in 2026 and beyond

The shift from AI tools to AI agents is the most consequential change underway. Per McKinsey, 62% of organizations are at least experimenting with AI agents, and 23% report scaling them somewhere in the enterprise. The difference matters. Tools require prompts. Agents pursue goals. For content marketing, this means the unit of work moves from "generate a draft" to "plan, research, draft, distribute, measure, iterate," handled in one continuous loop.

AI search rewrites the SEO playbook

Traditional SEO optimized for clicks. AI search optimizes for citations. When a buyer asks Perplexity, ChatGPT, or Google AI Overviews about your category, you either get cited or you don't. Getting cited requires structured, source-quality content, clear answers up front, and authoritative entity signals. Content programs built for the click economy will need to be rebuilt for the citation economy.

The lean team becomes the default

The most underdiscussed implication of agentic AI is what it does to team structure. McKinsey's data shows that workflow redesign, not headcount, is what separates high performers. The 5-person marketing team running on agentic infrastructure will routinely outperform the 20-person team running on legacy workflows. This compounds with sustained brand equity and a clear unique selling proposition. AI amplifies a real brand. It doesn't replace one.

Trust becomes the moat

As AI-generated content saturates every channel, the brands that compound trust through original research, distinctive voice, and genuine expertise will pull ahead structurally. Trust is the one thing AI cannot generate. It can only amplify or erode what's already there. Klarna's reversal is the early evidence. There will be more.

Where Tenet comes in

Most AI tools sit in one box of the workflow. A drafting assistant. A scheduler. A research summarizer. Stitching them together is its own job, and it's the job that keeps marketing teams stuck in the 94%.

Tenet is built for the workflow redesign McKinsey describes.

It's an AI marketing agent that runs the work end to end across content, SEO and AEO, product marketing, demand generation, social, and design, anchored in your real brand voice, positioning, and knowledge base, not generic AI output. The point is a rebuilt workflow where one marketer ships what a 5-person team would.

If that's the shift you're trying to make, see how Tenet works.

Frequently asked questions