There's an irony in marketing right now: the more content we produce with AI, the less anyone reads. Not because AI is bad at writing. But because we've turned it into a volume machine instead of a value machine.
A recent article on Search Engine Journal raises exactly this problem and proposes a five-pillar framework. At difrnt., we see things similarly, but with some nuances from our direct experience working with clients in Romania and across the region.
The problem isn't AI, it's the absence of strategy
Most marketing teams use AI as a shortcut: "write me an article about X." And they get exactly what they deserve: grammatically correct, inoffensive text that nobody remembers. The reason? Missing context.
A good AI brief looks very different from a quick prompt. It includes the specific audience, desired tone, emotional objective, brand constraints, and reference examples. Essentially, everything you'd put in a brief for a senior copywriter. The difference is that AI won't guess what you forgot to specify.
Data backs this up: according to a 2025 Salesforce study, 76% of consumers say they can detect AI-generated content. And 52% of them become less likely to trust the brand after detecting it. It's not about hiding AI, it's about using it with a plan.
Visceral storytelling makes the difference
Safe content is invisible content. It's one of the lessons we see confirmed constantly in the campaigns we manage. When a piece starts with "In today's digital landscape..." or "It is well known that...", the audience has already left.
Storytelling that works is anchored in sensory details and real experiences. It doesn't need to be dramatic. It needs to be specific. One client told us last year: "We increased email CTR by 40% just by changing the first sentence of every newsletter from product description to personal question." That's storytelling that convinces, because it's real and verifiable.
AI can generate good storytelling, but only if you give it the raw material: data from real campaigns, client feedback, market observations. Without specific human input, it produces generic stories that sound fine but say nothing.
Every platform demands something different
One of the most expensive mistakes we see from brands: they take a blog post, chop it into pieces, and post it identically on LinkedIn, Instagram, and their newsletter. Then wonder why engagement drops.
Each platform serves a different emotional intention. On LinkedIn, people seek professional validation and fresh perspectives. On Instagram, they look for visual inspiration and authenticity. In their inbox, they want direct utility. The same message needs to be adapted to each channel's native dialect.
AI excels at this adaptation when given clear platform context. A prompt that says "adapt for LinkedIn, professional but personal tone, max 1300 characters, hook in the first sentence" produces far better results than "rewrite for social media."
The metrics that matter aren't the ones you're tracking
Likes and impressions are vanity metrics we report because they're easy to measure, not because they're relevant. The metrics that actually show whether your AI content works are different: read time, scroll depth, completion rate, secondary clicks.
These retention signals do two important things. First, they tell you whether people actually consume what you produce. Second, they trigger algorithmic amplification across all major platforms. Google, LinkedIn, Meta all prioritize content that holds attention.
In our practice, when we optimize content based on watch time and scroll depth instead of impressions, we consistently see 25-35% increases in downstream conversions. It's not magic, it's just aligning with what actually matters.
Transparency is not a vulnerability
Many brands fear admitting they use AI. That's a mistake. Sophisticated audiences appreciate honesty and penalize pretense. A simple "AI-Assisted" label or a mention of AI in the process shows maturity, not weakness.
Of course, transparency without human editorial control is insufficient. Fact-checking, cultural sensitivity, brand voice alignment are all responsibilities that remain human. AI is a tool, not an author.
At difrnt., we use AI in our editorial process. But every article goes through at least one round of human editing that checks accuracy, tone, and relevance for the Romanian market. We don't publish anything we wouldn't sign our name to.
AI-generated content isn't going away. If anything, it will become the norm. The question is no longer whether to use AI, but how to use it so the result is worth reading. Clear strategy, specific storytelling, platform adaptation, relevant metrics, and transparency aren't optional. They're the minimum price of entry.
