Content Writing Principle
AI Overviews (Google)
Gemini (Google)
ChatGPT (OpenAI)
Perplexity
Answer Questions Directly and Comprehensively
Scans content for the most direct and complete answer to a user’s query. It synthesises information from multiple top sources to create a single, comprehensive summary at the top of the search results.
Uses its direct integration with Google Search to find a factual basis for its conversational answers. It seeks a definitive answer to ground the rest of its explanatory or creative response.
Rephrases and elaborates on direct answers found in its training data or through live web browsing. It uses the core answer as a starting point for building a more complete, human-readable paragraph.
This is its core function. It actively seeks out and collates multiple direct answers to a question, presenting them to the user with direct citations for each point, prioritising verifiability.
Demonstrate Experience & Expertise (E-E-A-T)
Heavily relies on Google’s established E-E-A-T signals. It prioritises content from sources that are already considered authoritative and trustworthy within the traditional search index for inclusion in the overview.
Inherits the E-E-A-T bias from its connection to Google Search. In its responses, it may prefer or explicitly mention information that originates from domains with high authority on a given topic.
Implicitly trusts and patterns its responses on the style and information of authoritative sources that were prominent in its training data. When browsing, it also tends to favour well-known, credible websites.
Empowers the user to assess expertise by clearly showing the source domains for every piece of information. While it doesn’t have an internal E-E-A-T score, it allows for immediate, user-driven verification of the source’s credibility.
Use a Clear, Logical Structure
Parses structured content (using H2/H3 headings, lists, tables) to easily extract key points, steps, or features. This makes it easier to pull your content into a formatted list or summary within the overview.
Leverages a clear structure to quickly understand the hierarchy of an article. This allows it to accurately summarise the content, reformat it into a different structure (like a table), or answer specific questions about a particular section.
Uses the document’s structure as a “map” to navigate the information. A logical flow helps it generate more coherent summaries and correctly interpret the relationship between different concepts in the text.
Relies on clear headings and sections to pinpoint the exact sentence or data point that answers a user’s question. This allows it to create highly specific and accurate citations for its claims.
Address a Specific Search Intent
The entire response is generated to match the presumed intent of the search query (e.g., informational, commercial, navigational). Content that doesn’t align with the primary intent is unlikely to be included.
Filters and interprets information from its knowledge base or web search through the lens of the specific intent of the user’s prompt. It tailors its response to be creative, informational, or analytical as needed.
Is highly sensitive to the user’s prompt intent. It will generate a response that directly serves the user’s goal, whether that’s to learn, create, or summarise, using web-sourced content as evidence or inspiration.
Focuses almost exclusively on satisfying informational and research-based user intent. It is optimised for answering “what,” “why,” and “how” questions with factual, cited evidence.
Write with Factual, Unambiguous Language
Clear, factual statements are easier to parse and are seen as more trustworthy. Vague or overly promotional language is often ignored, as the AI seeks verifiable facts to present to the user.
Unambiguous language reduces the risk of misinterpretation or “hallucination.” It can process and repeat factual statements with higher confidence than subjective or unclear ones.
Factual language present in its training data helps ground its responses. When providing explanations, it relies on clear, definitive statements to build its narrative and appear more authoritative.
Prioritises objective, factual statements that it can directly attribute to a source. Ambiguous language is difficult to verify and is therefore less likely to be used in a Perplexity answer.
Implement Structured Data (Schema Markup)
Directly ingests schema markup (for FAQs, How-Tos, Products, etc.) to understand content in a machine-readable format. This can directly influence inclusion in specialised AI-powered results.
Can interpret structured data found via Google Search to provide more detailed and formatted answers, such as presenting steps from a HowTo schema or details from a Product schema.
Does not rely on schema as a primary signal when browsing. However, it benefits indirectly as pages with good schema are often well-organised, making the content easier for it to parse and understand.
Similar to ChatGPT, it does not directly prioritise schema. Its focus is on the visible text on the page, but the clarity that schema provides to a webpage’s structure can indirectly aid its parsing process.
