2 Forms of Content material AI Can’t Create (And Why They Win)

0
233
2 Forms of Content material AI Can’t Create (And Why They Win)


There’s a lot speak about what AI can do. However what can’t it do?

Years in the past, my buddy Steve Rayson printed a report that analyzed the efficiency of one million articles to see what will get hyperlinks and shares. His conclusion modified my content material technique perpetually…


Steve Rayson, Founding father of BuzzSumo

“If you wish to create content material that achieves a excessive stage of each shares and hyperlinks, then it is best to think about opinion-forming, authoritative content material on present subjects or well-researched and evidenced content material.”


He’s telling us that two varieties of content material persistently outperform: sturdy opinion and unique analysis. This was seven years earlier than the launch of ChatGPT, however it’s extra related at the moment than ever.

Why? As a result of AI can’t make this type of content material. Solely people can.

Let’s discover some of these content material and rethink our personal content material methods. Let’s future-proof our content material for the AI-writes-lots-of-everything period. And let’s think about a couple of AI prompts to assist us make these sorts of articles. As a result of it could actually.

Why can’t AI do that?

All the things AI does relies on what it has already learn. It summarizes, restructures, rephrases, and averages. We typically joke that AI ought to stand for “Common Info.” Sure, averaging the web is genuinely helpful. And sure, most content material will get higher while you use AI.

However two sorts of enter are usually not within the AI coaching information:

  1. The ideas in your head: your expertise, your perspective and the opinions you’ve shaped over years of doing the work
  2. The information you create by analysis: Each new analysis and the info you could have already got in what you are promoting, that are numbers solely your organization can see

These are the 2 classes of enter that come from exterior the machine, out of attain from AI and bodily unattainable for it to generate.

Let’s take a look at every and (satirically?) search for methods to make use of AI to floor novel concepts for the content material strategist.

Sort 1: Your individual opinions (thought management)

We people have experiences that form our views and opinions. However AI has no experiences, no perspective and no opinions. AI fashions are skilled to search out the center, the place individuals agree. AI by no means takes a stand.

Ask AI for a contrarian take and also you’ll get a assured sentence, however it’s in all probability one thing that nobody would disagree with. However keep in mind, if you happen to can’t disagree with it, it isn’t thought management. Most content material advertising doesn’t meet the definition of thought management.

Right here’s the take a look at that separates an actual place from a faux one: might a sensible, knowledgeable particular person in your trade sincerely argue the other? This, from legit advertising thought chief Seth Godin:


A bald man with glasses wearing a suit and tie, photographed against a plain background.A bald man with glasses wearing a suit and tie, photographed against a plain background.
Seth Godin, Advertising Legend

“Thought management all the time creates pressure. It’s about making assertions. It’s important to be keen to be flawed. You might be sure that some individuals will disagree.”


However that doesn’t imply you must be controversial. No want to the touch the third rail. There are numerous comparatively mundane subjects that also qualify as thought management. Listed below are a couple of ideas in my head:

  • “Delete TikTok and Instagram. The extra boring your cellphone, the extra productive you’ll be.”
  • “Entrepreneurs ought to by no means use themselves as examples in conferences. You’re a dataset of 1.”
  • “I’d by no means put on clothes with logos, until it’s my model or a model that I truly assist.”

Did you’re feeling one thing while you learn these? These statements aren’t actually controversial, however they create a little bit of pressure, don’t they? And guess what? None of these sentences might have been written by AI.

These assertions don’t actually slot in my content material technique. I used to be simply being provocative. To be a match, the counter-narrative must be protected sufficient to say on a podcast or in entrance of purchasers, however spicy sufficient that somebody desires to answer.

Sturdy opinion content material drives engagement in social media.
The opinion drives responses from different customers; these responses set off the algorithm; the algorithm reveals it to many extra individuals. That’s why opinion-based content material is catnip for social streams.

You too can differentiate your content material from AI simply by including a bit extra of your self in your content material. Share somewhat story. Add some humanity. Be your self. Simply look again on the first 4 phrases of this text. AI can’t write that.

AI can’t write opinion-based content material. However it could actually assist set off concepts. Right here’s a immediate that permits you to use AI to search out the one factor AI can’t do. It’s a enjoyable one, I believe…

Thought Management (provocative, not controversial) Matters Immediate

My function / trade: [job title + industry]
Positions I truly maintain (non-obligatory): [any contrarian beliefs, pet peeves, or things you think the industry gets wrong]

You might be an skilled content material strategist who finds thought management angles that get quoted. Your job is to floor positions within the Sound-Chunk Zone: contrarian sufficient to make somebody cease scrolling and wish to reply, however protected sufficient that an actual govt would fortunately say them on a podcast and put their title to them. You aren’t on the lookout for controversy, sizzling takes that might harm a fame, or something political, private, or inflammatory. The goal is “spicy to say, protected to personal.”

Work by these 5 veins of provocative-but-safe opinion for this function:

  1. Consensus is flawed — a broadly accepted finest observe, frequent piece of recommendation, or perception a few trending subject that this particular person might credibly argue is mistaken and even dangerous.
  2. The unstated — a secular or unglamorous subject individuals on this function maintain surprisingly sturdy opinions about, or a query they’re quietly afraid to reply out loud.
  3. The buried lede — an vital subject that trade blogs and thought leaders persistently ignore or under-cover.
  4. The outsider lesson — a seemingly unrelated subject or self-discipline that holds a vital, transferable lesson for this trade.
  5. The reframe / forecast — a typical worry this function thinks is overblown, a counterintuitive method that quietly works, or a prediction most friends would name unlikely.

Step 1: Generate two candidate positions per vein (ten whole).

Step 2: Commit. Rating every on provocativeness 0–5 (does it cease you and make you wish to argue?) and flag reputational danger as inexperienced (protected to say publicly), amber (wants cautious framing), or purple (genuinely controversial — reduce or closely caveat). Rank all candidates and drop something under 3 on provocativeness or flagged purple. Winners are high-provocativeness, inexperienced or amber.

Step 3: For the highest [3–5] survivors, write the precise sound chunk: one quotable line, 15 phrases or fewer, {that a} journalist or podcast host might pull verbatim. A subject isn’t completed till it has a line.

Current the outcomes as a desk: the place, which vein it got here from, its sound chunk, provocativeness rating, danger flag, and a one-line observe on why it lands with out crossing into controversy. Lead with the strongest.

Sort 2: Your inside information or stats you produce (unique analysis)

New analysis is the opposite class of content material that might be perpetually differentiated from AI content material. There are two apparent causes:

  • AI doesn’t spontaneously generate new major analysis. Irrespective of how good the AIs get, they’ll by no means do that.
  • AI doesn’t practice on the info that you’ve got inside what you are promoting. Irrespective of how good the AIs get, they’ll by no means have this. They haven’t any entry.

The advantages are massive, however completely different from the advantages of thought management. Authentic analysis drives search rankings. Analysis attracts hyperlinks. Hyperlinks drive Area Authority, which is the rating potential of each URL in your web site.

Analysis can be extra more likely to get clicks from citations in AI responses, possibly as a result of individuals wish to test AIs sources. Should you’ve ever printed analysis, you possibly can affirm this your self by utilizing GA4 to test which of your pages will get visitors from AI fashions.

Should you want an instance of analysis in motion, take a look at how I opened this text: I cited Steve’s analysis. Why? As a result of it helps my thesis. His examine has additionally been linked to from 1,400 different web sites. I checked.

Listed below are another current examples of analysis from the Orbit Media weblog:

And naturally, analysis can grow to be charts, and these visuals make for very compelling content material. Right here’s one from our annual blogger survey. Right here you possibly can see how repeating your analysis yearly makes it extra invaluable over time.

Bar chart showing the average hours spent creating an article from 2014 to 2023, peaking in 2021 at 4:10, with a note: “Research makes for compelling visuals!”.Bar chart showing the average hours spent creating an article from 2014 to 2023, peaking in 2021 at 4:10, with a note: “Research makes for compelling visuals!”.

AI could make charts, however it could actually’t produce the info behind the charts. AI doesn’t conduct new analysis. It simply summarizes and cites information it finds.

So unique analysis works. However what sort of analysis? You’re on the lookout for a “lacking stat” which is an information level that doesn’t but exist on the web. It fills an information hole for a frequently-asserted-but-rarely-supported assertion individuals are already making.

The web just isn’t completed but.

One of the best analysis research produce information that’s helpful and that’s proprietary to your model. It makes your web site the first supply for brand spanking new data.

A 2x2 matrix categorizing content by “Who owns the data?” and “Useful to readers?” The quadrants are labeled: Citable corner, Already covered, Boring stats, and Vanity dataset.2 Forms of Content material AI Can’t Create (And Why They Win)

Add your personal sizzling take to the report (or fill it with quotes from subject material consultants and influencers) and also you’ve added a splash of thought management to your analysis. As soon as full, stand again and take a look at it. It will likely be apparent that AI by no means might have made any of it.

However superb information isn’t sufficient to drive efficiency. One of the best content material doesn’t win. One of the best promoted content material wins. So earlier than you publish a brand new report, scan by our Authentic Analysis Promotion Information. It might provide you with a couple of concepts for higher attain.

Hassle pondering of analysis subjects? Right here’s a immediate which will set off a couple of concepts. It begins by asking you concerning the information you could have already got…

Authentic Analysis Matter Immediate (Proprietary Information Version)

My firm: [name + industry]
Information I do know we now have (non-obligatory): [product usage, sales/CRM, support tickets, transactions, survey list, etc.]

You assist me discover unique analysis subjects that earn citations from journalists, editors, and bloggers. One of the best subjects sit the place two issues overlap: a declare my trade repeats consistently however has by no means measured, and proprietary information we already maintain — or might cheaply acquire — that might settle it. Nobody else can publish our information, which is precisely why it will get cited.

Step 1: Earlier than producing something, ask me 3–4 inquiries to floor information property I could have ignored: what we seize by the product, what prospects hold asking that I might reply with numbers, a stat I cite internally however have by no means printed, or what a fast survey to our record might reveal. Then anticipate my solutions.

Step 2: After I reply: record the claims my trade asserts however hardly ever helps with information, then filter to those our information might credibly show or disprove.

Step 3: Give me the [3–5] strongest subjects as a ranked desk. For every, embrace: the trade declare it settles; the proprietary information that settles it; analysis method (survey / evaluation / experiment); why we’re uniquely credible to publish it; the headline a journalist may write; a observe if a comparable examine already exists; and a 0–5 rating for newsworthiness and 0–5 for feasibility, every with a one-line rationale.

Rank by newsworthiness × feasibility, and lead with the one I might realistically ship soonest that also makes information.

Flag any subject counting on delicate information (buyer information, CRM, assist content material) so I can plan for aggregation or anonymization. Deal with this change as confidential.

10 content material codecs that AI can’t make

Past the 2 classes of content material, there are some content material codecs that AI can’t produce. I can consider at the least ten. Do content material in any of those codecs, and also you’ve executed one thing AI can’t do.

  1. Movies. Sure, AI makes video, however not as a named, human character being themselves.
  2. Convention displays. A human out of your model, dwell on stage
  3. Stay webinars. Certain, the slides may very well be generated by AI, however a human handles the speaking and Q&A.
  4. Panels and roundtables. People having an actual dialog. Good showcase for these spicy subjects.
  5. Podcast visitor appearances. A human host asks questions. You, the human, reply in actual time.
  6. Workshops and coaching. A room stuffed with people, with somebody out of your model main the dialogue.
  7. Stay teardowns and audits. People critiquing issues in actual time (like our month-to-month LinkedIn Stay periods).
  8. Authentic images and video of your crew. People on digicam in a bodily workplace, holding it actual.
  9. Contributor quotes and skilled roundups. Precise people, saying one thing they’d defend on the document.
  10. Group areas. Slack teams, boards, pleased hours the place people collect to be with different people.

The distinction is apparent. A few of these aren’t even on-line. Some are older than the web itself. These are essentially the most human codecs for content material advertising they usually aren’t going wherever. They’re resistant to tech disruption.

Each content material technique ought to have at the least a couple of of those within the combine. At Orbit, we do all ten of those regularly. However if you happen to might do only one? Our buddy Jay Acunzo has a robust opinion on this…


A man with dark hair and a beard, wearing a gray suit and white shirt, smiles at the camera against a light pink background.A man with dark hair and a beard, wearing a gray suit and white shirt, smiles at the camera against a light pink background.
Jay Acunzo – Speaker, Advisor, Coach

“I imagine essentially the most defensible, irreplaceable, but uncommon ability is your means to talk effectively and ship distinctive displays, on-line and in-person.”


It’s straightforward to distinguish your content material from AI

In truth, it’s pure. Many of the sentences we communicate in conversations each day might by no means be written by AI. They’re our personal private factors of view. But most content material packages hardly ever publish sturdy opinion content material.

And few content material packages ever publish new unique information. Most web sites are usually not the first supply for any information in any respect.

As a substitute, entrepreneurs in all places use AI to write down after which use AI once more to cover the truth that AI wrote it.

Sadly, one of the crucial upvoted “expertise” within the AI expertise directories is the Cease Slop Ability, which simply tries to make writing sound extra human (“Take away em dashes. Break formulaic constructions. Keep away from binary contrasts, dramatic fragmentation, rhetorical setups. No inanimate objects performing human actions. Range rhythm…”)

That is absurd.

Let’s shut with my favourite quote about AI from one among my favourite thought leaders.


A person with glasses and an open mouth expression in front of a brick wall.A person with glasses and an open mouth expression in front of a brick wall.
Jay Baer – Creator, Speaker, Advisor

““The issue just isn’t that AI is doing what entrepreneurs can do. The issue is that entrepreneurs hold doing what might be executed by an AI.”