Listed here are 5 time-tested writing frameworks that can assist each people and AI perceive your content material (alongside the analysis that exhibits they work).
Dan Petrovic wrote an important article explaining why human-friendly content material is AI-friendly content material. In a nutshell, there’s a hanging parallel between how individuals and AI fashions course of textual content info: we each attempt to glean that means from lengthy textual content with out spending a number of expensive time and vitality reviewing each phrase.
Or put merely: neither AI fashions nor individuals truly learn content material. We each skim, bounce about, and use intelligent heuristics to search out related items of textual content as effectively as potential.
Kevin Indig additionally performed nice analysis how AI fashions listen. They prefer to retrieve content material with a number of entities: particular, concrete references to manufacturers, individuals, and issues. They appear to prioritise textual content that features query/reply pairs. They search for sturdy, assured claims. They weight the primary third of a web page extra strongly than the remaining two-thirds.
The one factor lacking from the dialogue to this point is the precise writing. When rubber meets highway and digital pen meets digital paper, how can we construction the phrases on the web page to maximise our possibilities of being understood, retrieved, and cited by AI programs?
That’s a writing downside, not an information science downside, and it has options that predate LLMs by a long time. The writing frameworks beneath come from army communication, administration consulting, {and professional} journalism. They had been designed to assist busy people extract info rapidly—which seems to be precisely the identical job AI is attempting to do.
Listed here are 4 timeless writing frameworks you should utilize to earn consideration from people and AI search engines like google and yahoo alike.
BLUF—Backside Line Up Entrance—means stating your conclusion, advice, or key discovering within the first one or two sentences of any part, then supporting it.
This framework originated in army communication, the place commanders want selections quick. It turned commonplace in administration consulting, the place companions invoice by the hour and shoppers need solutions immediately.
BLUF helps readers perceive your key factors instantly, proper from the primary sentence of your introduction or paragraph. It additionally maps on to how transformer consideration works: fashions weight content material close to the highest of a piece extra closely than content material additional down.
From Kevin’s analysis, 44.2% of citations come from the primary 30% of content material.
Right here’s an instance of writing with out BLUF, typically known as “burying the lede”:
“There are various components that affect key phrase problem, together with area authority, backlink profiles, content material high quality, and search intent alignment. After analyzing 1000’s of SERPs, we discovered that backlinks stay the strongest predictor of rating success.”
Right here’s what BLUF appears to be like like in follow, with crucial concept shared at first of the paragraph:
“Backlinks stay the strongest predictor of rating success. After analyzing 1000’s of SERPs, we discovered they outweigh area authority, content material high quality, and intent alignment as rating components.”
You possibly can apply the BLUF precept all through your writing. Your introduction ought to share your key argument, takeaway, reply, or statistic. The primary sentence of every H2 ought to ship the part’s principal level.
Even your titles can (and possibly ought to) undertake BLUF. Our analysis discovered that ChatGPT “prefers” to quote pages with titles which have sturdy semantic similarity to the fan-out queries they use to seek for content material.
In different phrases, the extra clearly and straight your web page title solutions the query ChatGPT desires answered, the better your likelihood of being cited.
You possibly can take a look at your content material by studying solely the primary sentence of every part. If a reader can perceive your complete argument from these sentences alone, you’ve carried out BLUF appropriately.
Declarative statements are definitive assertions that may stand alone as solutions: “BLUF means stating your conclusion first,” “cited content material has 20.6% entity density,” “the Pyramid Precept constructions info from normal to particular.”
That is most likely the one highest-leverage change you may make for AI quotation. Kevin’s analysis discovered that quotation winners are nearly 2x extra doubtless (36.2% vs 20.2%) to include definitive language: phrases like “is outlined as,” “refers to,” and “means.”
This isn’t a brand new idea both. Instruments like Steve Toth’s SnippetBrain used the identical rules to optimise for featured snippets (bear in mind these?).
Declarative statements assist readers as a result of they’re unambiguous. There’s no work required to determine what you’re saying.
They assist AI for the same cause: AI fashions want to validate and assist the claims they make, in order that they bias in the direction of confident-sounding phrases that rapidly and straight reply the question, ideally in a single sentence.
Right here’s what a “hedged” sentence appears to be like like, filled with oblique, nonconfident language:
“It looks as if it is perhaps a good suggestion to possibly think about updating your title tags, since they may probably assist enhance your click-through charges in search outcomes.”
Right here’s what declarative writing appears to be like like in follow:
“Replace your title tags. They straight enhance click-through charges in search outcomes.”
I’ve labored with a number of English literature graduates, and educational writing trains individuals to hedge. “Could,” “would possibly,” “may probably,” “it’s potential that”—these qualifiers sign mental humility in a analysis context. However within the context of AI quotation, they sign uncertainty. If an AI mannequin is selecting between your hedged sentence and a competitor’s assured one, properly, yours most likely loses.
Not every thing ought to be declarative, in fact. Use definitive statements for definitions, established information, core ideas, and clear suggestions. Use certified language for rising analysis, contested claims, predictions, and edge circumstances.
Entity density is the ratio of named entities—manufacturers, instruments, individuals, locations, particular ideas—to complete phrases. It appears to be one of many clearest dividing traces between content material that will get cited and content material that will get ignored.
Kevin’s analysis discovered that closely cited textual content has an entity density of round 20.6%, in comparison with 5-8% in commonplace English prose. That’s 3-4x the traditional charge. Generic writing will get neglected; particular writing will get cited.
Entity density helps AI fashions “perceive” what a chunk of textual content is about. LLMs “assume” when it comes to entities, and the relationships between them (like Google’s Information Graph). The extra entities, the simpler it’s simpler to evaluate if a chunk of content material is related to a selected question.
The identical precept applies to individuals: the extra entities you utilize in writing, the extra particular and detailed you’re, and the extra helpful that means you convey in your textual content.
Right here’s an instance of low entity density, utilizing imprecise and non-specific phrases:
“There are a number of factors you are able to do to make your web site present up higher in search outcomes and get extra individuals to go to it over time.”
Right here’s an instance of excessive entity density, mentioning many particular, concrete ideas and issues:
“Optimizing meta titles, inside hyperlinks, and Core Net Vitals improves your Google rankings and will increase natural site visitors.”
The rule for writing is easy: title the instrument, cite the supply, specify the metric. Each time you catch your self writing one thing imprecise and jargony —“use a instrument,” “test a metric,” “in accordance with analysis”—exchange it with a concrete, particular instance.
Strategic repetition means putting your most necessary concepts at a number of factors in your content material, rephrased for every context.
Repetition helps readers as a result of individuals don’t eat content material linearly. They arrive through search, through inside hyperlinks, through scrolling previous your first three sections. A reader who jumps to part 5 hasn’t seen your introduction. Working reminiscence is restricted, and spaced repetition will increase retention. In case your most necessary level solely seems as soon as, most of your readers will miss it.
Repetition helps AI for a extra mechanical cause: fashions don’t learn your complete web page. As Dan Petrovic confirmed, they as an alternative retrieve snippets—brief passages that match a question.
In case your key perception seems solely in a single paragraph and the mannequin retrieves a distinct snippet, that perception is invisible. Repeating it elsewhere, with totally different phrasing, creates a number of extraction alternatives. Every rephrased model additionally matches barely totally different question fan-out formulations, growing the chances that one in every of them aligns with no matter fan-out question ChatGPT is operating.
Consider it this manner: if AI solely samples 30% of your content material, repetition will increase the chance that any given pattern accommodates your most necessary level.
It is sensible to say key concepts in your introduction…
“Inner linking is probably the most underrated lever in search engine optimization. Whereas most groups chase backlinks, the websites that persistently outrank them are quietly distributing asuthority precisely the place it’s wanted.”
…to repeat them mid-article, as a contextual reminder…
“This is the reason inside linking is so disproportionatelty efficient—the location above didn’t earn a single backlink throughout that interval.”
…and eventually reinforce your level with a abstract within the conclusion:
“Backlinks get the glory, however for greater websites, inside linking has a surprisingly huge affect on search visibility.”
Every model reframes the identical core concept, including further context with every repetition—and growing the chance that folks and machines will discover and perceive it.
The information means that these easy writing frameworks are helpful for each human readers and AI search engines like google and yahoo, trying to find content material to retrieve and cite.
However crucially, these are usually not magix fixes: your content material must rank. Our research of 1.4 million ChatGPT prompts discovered that 88% of cited URLs come from ChatGPT’s normal search index. When you’re not exhibiting up in search outcomes, on-page writing high quality is a moot level.
However as soon as you’re rating, the way you write determines whether or not you get cited or ignored—ChatGPT solely cites about half the URLs it retrieves. That hole is what these frameworks are designed to shut.
You should utilize Ahrefs Model Radar to see which AI platforms cite you—ChatGPT, Perplexity, Claude, Google AI Overviews, AI Mode, Grok—and which particular URLs get quoted. Simply enter your model and area, head to the Cited pages report, and click on the Yours tab to see your most-cited URLs (and the way their citations change over time):
