
A company blog that follows Google's rules is not about how often you publish, but why and what. Google does not state a recommended pace in its documentation. Trouble starts when articles are made mainly for search engines: lots of texts on lots of topics in the hope that something sticks, or large numbers of AI pages without added value. What works for me is one to two articles a week, written and checked two weeks ahead, and date-based publishing that runs on its own. Below I describe how I set it up on my own site and what follows from Google's rules.
- Pace: Google does not prescribe one. What matters is whether each article adds something. I stick to 1 to 2 articles a week, because I cannot check more properly.
- No batches: ten articles at once are hard to check and resemble content written "to fill a quota", which Google warns about.
- Planning: a calendar with topics based on services, a buffer of at least 14 days, a checklist before anything enters the queue.
- AI yes, but: check everything AI produces by hand, including the title, meta description and alt texts, and say how the content was made.
- Technically: an article with a future date stays hidden, and on the publication day the site rebuilds itself.
How often should a company blog publish?
On the page Creating helpful, reliable, people-first content (updated 5 October 2026) Google gives no frequency. It does say what to avoid. Warning signs include producing lots of content on many different topics in the hope that some of it performs in search, or using extensive automation to do so. On the same page Google also answers directly that it has no preferred word count.
So I derive the pace from how many articles I can write and check so that each one contains my own experience and verified facts. For me that is two a week at most, on Tuesday and Thursday. Regularity has a practical reason too: I can evaluate each article separately in Search Console after 14 and 28 days and know what worked.
Why not publish articles in batches?
I have more finished articles in the queue, yet I do not release them all at once. There are three reasons:
- Review. I cannot go through ten articles in a day as carefully as two a week. A factual error then stays live longer.
- Signal. A batch of texts on assorted topics is exactly what Google describes as content written for search engines rather than people. I do not want the site to look like that, even when the texts are good.
- Evaluation. When everything goes out at once, Search Console will not tell me which article did what.
The opposite trap is changing the date on old articles to make them look fresh. On its helpful content page Google asks directly whether you change the date of pages when the content has not substantially changed. That is why an update date only appears on my articles with a real revision, for example when a price or a tool version changes.
How I plan the editorial calendar
The calendar starts from services, not from topic ideas. My services fall into seven areas (websites, e-shops, SEO and marketing, AI, apps, education, graphics and video), and articles rotate between them. Two articles from the same area do not go out in the same week where I can avoid it. Once a month I add an overview of AI news with official sources.
Every new topic goes through two checks before I write it:
- Cannibalization: is there already a page that answers the same query? How I look for it is described in my article on keyword cannibalization.
- Own contribution: do I have my own experience, data or process for the topic that is not elsewhere on the site? If not, I do not write the article.
I keep the buffer at least two weeks ahead. When it drops, I write another article from the calendar instead of speeding up. A gap in the calendar is better than a weak article.
The checklist every article must pass
| Area | What I check |
|---|---|
| Intent | Who searches for the article and which question it answers, in one sentence. One service it supports. |
| Value | A concrete process or decision advice, an explanation of why, my own experience with a traceable record. |
| Facts | Every number, price, version or rule links to a primary source with a verification date. Nothing unverified in the text. |
| Author | A named author, first person, AI disclosed where readers expect it. |
| Form | A quick answer under the intro, headings as questions, a table, FAQ, an English version. |
| Queue | At most 2 a week, the same date in both languages, links only to articles that are already out. |
For each article I keep a source sheet: a table of claims, sources, validity dates and verification status. It is extra work, but six months later I know where every number came from and what needs updating.
How to write with AI without it being spam
Google does not ban AI in content. The page Using generative AI content (updated 1 October 2026) says generative AI can be particularly useful for research. Using it to produce many pages without adding value for users may violate the policy on scaled content abuse. I take three rules from that page:
- Check everything by hand. Not just the text, but the title, meta description, structured data and image alt texts too. AI can write a convincing falsehood.
- Say how the content was made. Google recommends explaining to readers how automation was used. I do it where readers expect it, for example with demos or images made with AI.
- The added value has to be human. Experience, real-world data and the decision on what to recommend do not come from AI. The author has to bring them.
When writing, I use AI for research, the outline, the first draft and checking facts against sources. The experience, the numbers from my Search Console and the conclusions are mine, and I am responsible for the text. Writing so that AI search engines cite you is a different topic: I cover it in my article on writing content for AI search.
Date-based publishing without a CMS
This site has no WordPress or other CMS. Articles are files in a repository and the site is built with Next.js. I set up scheduled publishing like this:
- The date in the article decides. An article with a future date stays hidden: it is not in the blog listing, the sitemap or related articles, and its URL returns 404. I count the day in Prague time so that an article does not go out an hour early or late.
- A check runs every morning. A task runs automatically every day and looks for an article dated today. If there is one, it goes through a check and only then triggers a site rebuild. On other days nothing is built, which saves hosting limits.
- Link check. A published article must not link to an article that is not out yet. The check catches that and does not publish the article in that case.
- Preview. Locally I can switch on scheduled articles and see how they will look.
For a client on WordPress, the built-in scheduling is enough: in the post settings you pick a future date under Publish (WordPress documentation). A custom solution makes sense for bespoke sites that have no CMS and where you do not want to add one just for a blog. I build such sites as part of my web development service.
When a company blog does not make sense
- You have nothing to say about your field beyond what search already offers. Then the blog only adds pages nobody reads.
- Nobody will check and update the articles. An outdated article with a wrong price does more harm than none.
- You expect results right away. I evaluate articles in Search Console after 14 and 28 days, and earlier the data tells me nothing reliable.
- List the services you want the blog to support, and for each one three questions customers ask you.
- For each question, note what experience or data of your own you have.
- Choose a pace you can keep checking, and stick to it.
- Before publishing, go through the title, description and alt texts as carefully as the text.
Frequently Asked Questions
How many articles a month does a company blog need?
There is no fixed number and Google does not prescribe one. Four honest articles a month beat twenty weak ones. I stick to one or two a week, because I cannot check more properly.
Does Google penalize articles written with AI?
Google does not ban the use of AI as such. It treats large numbers of pages without added value for users as a policy violation. What matters is what the article gives the reader and whether someone checked it before publishing, not the tool it was made with.
Should I update the date on old articles?
Only when the content has substantially changed. Google explicitly lists changing dates to make pages look fresh as a warning sign.
Does every article need a link to a service?
It does not have to, but on a company blog it makes sense. The article should help the reader first and only then show where you can help. A link to one related service is enough.
How far ahead should articles be ready?
I keep at least two weeks. The buffer gives time for review, and a lost week does not mean an empty blog.
Can you help me with content and blog setup?
I handle content strategy, the calendar and article review as part of my SEO service. I set up scheduled publishing on a bespoke site as part of web development.
