We publish to help marketing leaders make better decisions about search, AI answers, and content. Most of our readers are deciding where to spend real budget. So, every post has to be accurate, useful the day you read it, and honest about where the information came from.

This page explains who writes our content, how we research it, how we use AI, and what we do when we get something wrong. We hold our clients' content to the same standards.

Who writes our content

Every post carries the name of the person who wrote it. Most posts on our blog are written by Matthew Proctor, Narrative Bent's founder. He started as a copywriter after graduating from Florida State University in 2007 and has worked in SEO since 2009.

Matthew managed SEO for Procter & Gamble brands at Catalyst Search Marketing, then ran professional services engagements at BrightEdge for brands including Nike, 3M, McKesson, and Facebook. He joined HomeLight as employee #27 and led content and SEO for six years, building a team of about 30 and growing organic search from zero to over one million monthly visits. He founded Narrative Bent in 2023.

Our editors and team leads also write posts in their areas, and they keep their bylines on the work they write. We sometimes publish outside contributors. They are practitioners with a track record in search or AI, and they get their own byline and bio.

Every author is a real person, and each bio links to more about them.

How we research

We start every post with what our readers ask us. The questions come from sales calls, client meetings, and the follow-up questions people search for, so the headings match the decisions you're trying to make.

Then we read what already ranks in Google and what AI assistants like ChatGPT, Perplexity, and Google's AI Overviews cite for the same question. If a post can't add something those pages don't already say, we don't publish it.

What we add comes from 4 places:

• Our client work: what we've tested, measured, and learned on live programs.
• Primary sources: documentation and statements from Google, OpenAI, Anthropic, Perplexity, and Microsoft, quoted exactly and linked.
• Studies: research with a stated method and sample, credited to the people who ran it.
• Our own experience and opinions: labeled as opinion, never presented as a finding.

How we use AI

We use AI, mainly Claude, because it makes our research deeper and our drafts more thorough. We'd rather tell you exactly how than leave you guessing.

Every claim is checked against its source before publishing, and nothing goes live until a person has read it and approved it. That includes the parts search engines see, like titles, meta descriptions, and image alt text. The author named on the post is responsible for what it says.

Fact-checking and sources

Every factual claim in a post is backed by our own client work, a primary source, or a published study. If we can't back it, we cut it. We never make up a statistic, a quote, or a client detail.

• We link to the original source (the documentation page, the study, the person's own post), and the anchor text names it.
• We say when something is a leak, a rumor, or a single study, and we note the sample size when it matters.
• Quotes from people are exact, and we link to where they said it.
• Before publishing, we open every link and confirm the number or quote is on that page.
• Titles describe what the post delivers. We don't write clickbait (or try not to, at least).

Client stories and confidentiality

Our best lessons come from client work, and our clients trust us with their data. So, we share the lesson and protect the client.

Unless a client has given us written permission to name them, we describe them by category (for example, "a proptech SaaS company"). We also remove details that could identify them, like product names, the exact keywords they rank for, and the publications that covered them. The numbers stay, and every number we share has been checked against the client's own reporting.

We report results by what the business cares about, like leads, signups, pipeline, and revenue, and we're honest about what didn't work.

Independence and disclosures

Narrative Bent is an agency, and our blog is one way people find us. Some posts mention our services or end with an offer to look at your site. When a post describes how we sell our own work, we say so and explain the tradeoffs.

• We don't take payment, free products, or favors in exchange for coverage, mentions, or links in our content.
• We don't use affiliate links.
• When we compare tools, vendors, or agencies, we state the criteria we used, say where the alternatives are the better choice, and disclose any relationship we have with the companies named.
• We name and link the people whose ideas we build on, including when we disagree with them.

Corrections and updates

Search and AI change fast, and sometimes we'll get something wrong or a post will go out of date. When that happens, we fix it.

• Errors: we correct factual mistakes as soon as we confirm them. If the correction changes the meaning of the post, we add a note at the bottom saying what changed and when.
• Updates: when we substantially revise a post, we show a "last updated" date. We don't change dates to make old posts look fresh.
• Outdated posts: if a platform changes its rules and a post's advice no longer holds, we update it or say clearly at the top that it's out of date.

Contact us

If you spot an error or want to know how we made something, tell us here. A person reads every message, and we'll reply with what we changed or why we didn't.