How to Write Content for SEO & AI That People Love (5-Step Guide)

To perform well in search engines, content has to be valuable and engaging. But how does that translate to performance in LLMs? 

As it turns out, writing for SEO, AI, and human readers doesn't require three separate strategies. In fact, many fundamentals remain the same: Identify a genuine need, provide an original, trustworthy answer, and make it easy to find and understand.

In this guide, we combine recent AI research with real marketing principles for a modern take on content that ranks in search, earns citations and mentions in AI, and keeps human readers engaged.

What Do LLMs Actually Look for?

There’s no universal formula for earning LLM citations. ChatGPT, Google AI Overviews, Perplexity, and other platforms use different retrieval systems, and their citations frequently change across prompts and searches. 

Still, it helps to understand the retrieval process. When prompted, an AI search tool may:

1. Decide whether the prompt requires a web search.
2. Translate the prompt into one or more searches.
3. Retrieve and rank relevant pages or passages.
4. Use some of that material to generate an answer.
5. Cite select sources.

At the very least, this process suggests that a page must be accessible to the platform’s search system, relevant enough to enter its source pool, and useful enough to inform its answer.  Beyond that, optimisation becomes more platform-specific—and reliable evidence about factors for citation and mentions remains scarce.

Google says existing SEO best practices remain relevant to AI Overviews and AI Mode; no special optimisation required. Public guidance from other platforms focuses primarily on technical access: OpenAI says sites that block OAI-SearchBot won't appear as sources in ChatGPT search responses. At the same time, Perplexity uses PerplexityBot to surface and link websites in search results.

Third-party studies, meanwhile, suggest there’s no single content “hack” that can guarantee visibility. A 2026 review of 45 GEO studies, for instance, found topical relevance produced the most consistent results, but no technique demonstrated a stable, long-term effect on organic visibility across platforms.

Our practical takeaway is to focus on the following:

• Accessibility: Make important information crawlable and easy to understand.
• Relevance: Address a clearly defined question or task.
• Originality: Contribute information and perspectives that are unavailable elsewhere.
• Credibility: Support claims with expertise, evidence, and transparent sourcing.
• Authority: Earn independent mentions and links that reinforce why your source is worth using.

Step 1: Find a Real Audience Need

Strong articles start with a genuine audience need. To identify these, you have a number of worthwhile tools at your disposal. 

Keyword research, competitor analysis, and Google Search Console queries offer one way to identify what people search for, how they describe their needs, and which topics attract demand. 

Still, you’ll want to look beyond search data.

First, over-reliance can encourage content creation for rankings rather than readers, which Google classifies as “search manipulation” and may penalise with lower rankings, deindexing, and manual actions.

More importantly, though, conventional SEO tools won’t tell you everything you need to know about your audience. 

To uncover more specific and emerging reader questions, supplement SEO research by exploring:

• Community discussions (such as Reddit), industry forums, and trade media
• Social media posts and comments
• Subject-matter experts
• Customer interviews and surveys
• Transcripts from sales calls and product demonstrations
• Customer service tickets and onsite search data
• First-party chatbot or AI-assistant logs, where available and appropriate

For example, a content strategist at a cybersecurity company could ask customer support for recurring questions from Zendesk tickets, review anonymised sales-call transcripts in Gong, and collect relevant discussions from IT forums like Spiceworks or Reddit’s r/sysadmin. They could save the questions, excerpts, and source links in a shared Notion page or research document for later analysis.

Turn audience questions into an outline

Once you’ve collected audience questions:

1. Enter each question in a spreadsheet.
2. Group questions by shared problem, theme, or decision.
3. Record where each question appeared and how often.
4. Prioritise themes based on frequency, business relevance, and your expertise.
5. Choose one primary question to define the article’s focus.
6. Use related questions to shape the H2s, examples, or FAQs.

‍

For instance, say you’re creating a content plan for your HR software company. Your curated research could look like this:

In this example, customers most commonly ask about integrations, so you might prioritise an article titled “How to Switch HR Software Without Disrupting Payroll.” You could address related implementation questions, such as “Will switching platforms disrupt payroll?” and “How long does HR software implementation take?”, in H2s or FAQs.

Bonus tip: Use your audience’s language, not industry jargon. For example, write “cost per employee” if that’s how customers describe pricing, rather than “per-seat licensing.”

Choose the right format and angle

Search the primary question in Google and several LLMs, like ChatGPT, Claude, and Gemini. Use the existing results to determine what the user likely wants to do: learn a process, compare options, solve a problem, make a purchase, etc. Note recurring subtopics, frequently cited sources, and gaps in existing coverage.

Use those findings to decide whether the article should be a guide, comparison, list, video, calculator, or another content format. Then identify how your content will improve on existing results through stronger evidence, a missing perspective, greater specificity, or a more useful presentation.

For example, if results for “how to tie shoes for kids” favor visual demonstrations, create a concise guide with original step-by-step photos or a video rather than a long, text-only article.

Step 2: Give People and Algorithms Content Worth Sharing

LLMs can produce versions of the same generic article for countless websites. Readers have little reason to choose another summary of widely available information, while traditional and AI Search have little reason to retrieve or cite a page that adds nothing new. 

To resonate with readers and gain visibility, your content needs a meaningful differentiator. That might come from the information itself, the people behind it, the audience it serves, or the way it helps readers act.

This chart outlines ways to differentiate an article for readers, search engines, and LLMs.

Once you’ve identified a promising differentiator, turn it into a concrete editorial plan using these steps.

Choose the differentiator before assigning the article

Select at least one differentiator and state it explicitly in the brief. “Write the best guide” is too vague; specify the new information or utility the writer must deliver.

For example, a brief for an article about AI adoption in construction might require an original survey comparing usage across small, midsize, and enterprise companies. That requirement determines the research, sources, and structure before drafting begins.

Put the differentiator front and center

Introduce the key finding, tool, or perspective early, then build the article around it. Use descriptive headings that clearly state what each section covers—for example, “How Much Does Project Management Software Cost Per User?” rather than “Pricing Considerations.”

Include charts, callouts, and downloadable assets, where appropriate. They can make key insights easier to understand and share.

For example, if your survey finds that small construction companies are adopting AI faster than enterprises, contradicting the conventional assumption, highlight that in a callout, chart adoption by company size, and offer the complete results as a downloadable report.

Apply SEO fundamentals

Many traditional SEO best practices still play an essential role in getting your content to rank on Google and appear in LLMs. And those without direct ranking influence can encourage readers to click beyond an AI summary. 

We recommend treating these and other familiar SEO "rules" as practical guidelines.

• Write an accurate title tag: Clearly describe the page, include the primary keyword, and aim for roughly 50–60 characters when practical.
• Create a useful meta description: Summarise the article's value and give the target reader a reason to click.
• Use keywords naturally: Include relevant terms in the title, H1, introduction, URL, and appropriate headings without forcing repetition.
• Optimise URLs and links: Use a short, descriptive URL, add relevant internal links, and support factual claims with credible external sources.
• Optimise useful images: Compress large files and add concise alt text that communicates the image's meaning or purpose.

Step 3: Demonstrate Expertise and Earn Trust

A deluge of AI-generated content has also put added emphasis on Experience, Expertise, Authoritativeness, and Trustworthiness (E-E-A-T), Google’s quality framework that has significantly influenced organic content strategies since its V1 (E-A-T, which didn’t yet focus on first-hand experience) launched in 2014.

E-E-A-T isn’t a single ranking factor. Instead, it describes qualities associated with helpful, reliable content, including relevant expertise, firsthand experience, accurate sourcing, and transparency.

These qualities also matter in AI search for a practical reason: An AI-generated answer is only as dependable as the information it retrieves. Clear sourcing, relevant credentials, and firsthand evidence help readers verify claims even when they arrive through an AI summary.

A strong differentiator can strengthen an article's E-E-A-T, as can these strategies.

Provide evidence

Match the strength of your evidence to a claim’s importance. Whenever possible, trace facts back to primary sources, then use qualified experts and credible secondary analysis to provide additional support or context.

Make expertise visible

Research loses credibility when readers can’t tell who created it, how it was produced, or why they should trust it. Make those signals explicit by:

• Assigning writers with relevant subject-matter knowledge
• Naming qualified expert reviewers for technical or high-stakes content
• Displaying relevant credentials in bylines and author bios
• Explaining testing, survey, research, and scoring methodologies
• Placing citations next to the claims they support
• Distinguishing facts from estimates, opinions, and inferences
• Disclosing affiliate relationships, sponsorships, and potential conflicts
• Providing clear editorial, corrections, and contact information
• Adding review or update dates when freshness matters

Step 4: Write With Distribution in Mind

Content quality influences whether journalists, experts, partners, and communities will cover, cite, share, or link to a piece. 

Original findings, quotable expert insights, useful tools, and compelling visuals encourage distribution. Placements put the work in front of relevant readers while building the third-party authority that may support visibility in traditional and AI search.

A 2026 Muck Rack analysis of more than 1 million AI-cited links found that about 25% of citations came from journalistic sources, with trade and niche publications contributing a high share. Eleven's own correlational analysis also found that referring domains and backlinks were among the factors most strongly associated with LLM citations.

These steps can help make distribution part of the article’s ideation and development process.

Identify the hook

Start with the differentiator selected in Step 2, then determine why someone outside your company might share it. The strongest hook could be a:

• Surprising original finding
• Timely response to an industry development
• Quotable expert viewpoint
• Practical tool or template
• Proprietary chart or visual
• Useful resource for an underserved audience

Make that hook prominent in the headline, introduction, and article copy. For example, a survey finding that independent medical practices spend more time on prior authorisations than hospital-owned clinics is a clearer, more compelling pitch than a general article about administrative burdens in healthcare.

Involve people who can extend its reach

Interview experts, customers, or industry organisations who can strengthen the article and introduce it to relevant audiences. Choose contributors for the value of their insight while also considering their following size.

Before publication, confirm names, titles, and quotations. Once the article is live, send contributors the link, key findings, and any visuals they may want to share. Don’t require or imply that participation depends on promotion, however.

Match the channel to the audience

Different channels reward different aspects of an article. Decide where the work should appear while it is still being developed so you can create any necessary supporting assets.

Create a distribution plan before drafting

For every priority article, identify the:

• Shareable asset: What finding, quote, tool, or visual is worth passing along?
• Primary audience: Who directly benefits from the article?
• Potential amplifiers: Which experts, partners, journalists, or publishers already reach that audience?
• Relevant channels: Where does the audience discover and discuss this subject?
• Repurposing plan: Which sections can become an email, video, social post, webinar, or sales asset?
• Success signal: Are you aiming for links, mentions, qualified visits, signups, or assisted conversions?

These steps won’t guarantee coverage, but they will help you produce articles people have a clear reason to read, use, and share, which should boost AI and traditional search performance in the long term.

Step 5. Measure, Refresh, and Improve

Even the best-laid SEO or AEO/GEO content plans can land with little more than a whimper due to changing algorithms, fragmented discovery across platforms, and increasingly crowded search results.

That’s why “test and iterate” remains an important part of the process. To do so effectively, you’ll need to monitor performance across several layers

Keep in mind that AI citation data is currently noisy. Use the same prompts, platforms, settings, and testing cadence when comparing results. Run prompts more than once and treat small changes as directional signals rather than proof that a single edit worked.

At Eleven, we strongly believe in using a multi-touch, multivariate attribution model to measure content performance. Most people don’t purchase a product or sign up for a service after reading a single article. They may visit several pages before converting, so it’s important to understand how your entire content library works together to influence that decision.

To improve, turn measurement into an update queue

Review priority content quarterly or when a meaningful change occurs. Flag pages when:

• Impressions remain strong, but clicks fall.
• Rankings or AI citations decline across repeated measurements.
• Products, laws, statistics, or best practices change.
• Competitors introduce stronger evidence or a more useful format.
• The page earns traffic, but few readers take the next step.
• A successful article creates unanswered follow-up questions.

Try to diagnose the issue ahead of a rewrite. For instance, a falling click-through rate may call for a better title and description, while lost rankings may reflect stale information or stronger competing pages.

Record the change, publication date, and expected outcome, then measure again. That turns refreshing from a vague maintenance task into a repeatable editorial experiment.

Final Take 

Content stands the best chance of reaching readers through traditional and AI search when it addresses a real need and makes a useful answer easy to discover, understand, and trust.

The strongest modern strategy goes beyond basic optimisation. Create something original, make genuine expertise visible, and actively distribute the work across channels and reputable third-party sources. Then repeatedly measure how your content library contributes to discovery and business results.

Remember, no one step in this guide works in isolation. For instance, content is more likely to earn mentions, links, and press when it's genuinely useful, original, and supported by credible evidence or firsthand expertise.

Expert-led content, brand authority, and a multichannel presence form the foundation of "Expertise, Everywhere," Eleven's bespoke approach to content marketing. 

Download our free eBook, "How scaling B2B brands win in AI Search," to learn more about this strategy, then get in touch to see how we can help your brand.