Editorial standards

A hands-on take on something new can be either; it depends how deep it goes.

• News: a quick first test right after a release. “I tried it on my workflow, and here's what I saw.”
• Article: a thorough evaluation weeks later. “How it changed our SEO audits: the setup and the results.”

Still unsure? If the headline needs a version number to make sense, it's news.

1. The story. Every post belongs to a story (Meta Muse, Google Search…), and the story line at the top links its other posts.
2. Why it matters for marketers. One sentence of fact: who this changes things for, and how. Without it, a post is a rewrite of the announcement, and we don’t publish those.
3. Zev’s take. What Zev thinks it means and what he’d do about it, from two sentences to a paragraph or two. It’s opinion, signed and set apart from the facts, and always in his own words: AI agents never write it. Every post has one.
4. At a glance. The key facts in a few lines: what, who it’s for, where, price, when.
5. What happened. The facts in full, in our own words, with the version and the date.
6. What to check. The next step in your own setup.
7. Questions. The two to four questions people actually search, each answered in a sentence or two.
8. Sources. Every source, the company’s own page first, with what each one backs up.

We publish news every day. Something changes in AI marketing every day, and a short, accurate post about it beats a long one that arrives late. If a day brings nothing worth a post, we say less, not more: we never pad a story to fill the slot.

A post’s date is the day it went up here. When a post covers news from earlier, it says when that news happened, and we never backdate a post.

• It starts from something we actually built, ran or tested, not from a press release.
• It shows the work: the setup, the steps, the result, and what broke along the way.
• It stays current: when a tool changes, we update it in place and show the date, instead of publishing a new one.
• About two a week, each one piece of the system that week's newsletter teaches. We don't rush one out to fill a slot.

News brings readers in. Articles are what they stay for.

• Every news post links its story, and the story’s page lists every post on it.
• When an article goes deeper, the news post links to it.
• When a story keeps mattering, we write the article and link it from the news post.
• Each week's email is a tutorial: it builds one AI marketing system, step by step, from that week's articles, then rounds up the week's news.

• Facts come from primary sources: the company’s own announcement, release notes, documentation, pricing pages, and our own runs.
• Every news post is checked against the company’s own page before it goes up. If we can’t check it there, it waits.
• A company’s claim is written as theirs (“Meta says”), and reporting by another outlet is named and marked as reporting.
• We write in our own words and don’t copy other outlets’ copy.
• Every number is checked against its source, and every news post names the version and the date.
• Every news post ends with its sources, and what each one backs up.
• When we tested something ourselves, we say what we tested it on and how.

A publication is a garden: it grows only if it's tended. This is our schedule.

• Every day: a news post, filed to its story. Its story's page lists it as soon as it's live.
• About twice a week: an article, filed to its topics.
• Every week: the tutorial email, and a review of each topic that moved: when a topic page's explainer no longer matches the news, we rewrite it.
• Every month: we look at which topics are thin and what readers search for that we haven't covered, and choose what to write next.
• Every few months: we split a topic that has grown too big, or merge two that overlap. Old links keep working.

AI may draft an update to a topic page. A person checks every fact in it against the post it comes from, and approves it before it goes up.

When we get something wrong, we fix it and note at the end what changed and when. We don't quietly rewrite. Updates that aren't corrections show an “Updated” date. Spotted a mistake? Email zev@esy.com.

We write about AI tools, and we use them: for research, first drafts and images, mostly through Esy's own workflows and Claude. A person reviews, tests and signs off every piece, and AI images say so in their captions. Zev writes AI Marketing News every day. Some posts are researched and drafted by AI agents, which find the company's own announcement, write, and check every claim against it. Every post says which, and Zev approves each one before it goes up. Agents can draft the facts; Zev's take is always his own words. News covers aren't AI images: they're cards we draw from the post's own facts. We never have AI draw a company's logo.

Esy, clip.art and SEOPage are ours. When a piece is about one of them, or uses one as the example, we say so. We don't take payment to cover a product or to rank it. The companies we cover don't sponsor, review or approve what we write. Their logos appear only as their own files, unchanged, where their brand terms allow news use; otherwise we just use their name.

• We write for readers first. A headline says what happened, with the product's name, never a tease.
• Every piece belongs somewhere: a news post to its story, an article to its topics. Those pages gather everything on them, so each is the best single place to read about its subject.
• Every piece links its sources, its story or topics, and anything we've written that goes deeper.
• Questions in a post are ones people actually ask, answered plainly.
• Dates are honest. A page's “Updated” date changes only when its words do; we never refresh a date to look current.
• We don't make pages to catch searches: no near-copies of one post, no rewrites of other outlets, nothing generated in bulk from feeds.

• Plain words. Say what a tool does and what it did for us, without hype.
• Write for the reader: what they can do after reading.
• American spelling, and one idea per sentence where we can.