ChatGPT writes a text on instruction; a blog system runs the blog. The difference isn't in the writing, but in everything around it: topics from real search queries, a fixed weekly rhythm, publishing in the CMS, internal links, images and an approval that runs without you. If you use ChatGPT, you take on these steps yourself and usually need one to two hours per post. ChatGPT is enough if you need only a few posts and have the discipline for the rest.
ChatGPT writes a blog text on "What does a heat pump cost" in a minute. That's impressive, and for many business owners it's the moment they think: why bother with a system at all? The question is fair, and the answer isn't "because the system writes better". The difference lies elsewhere.
What ChatGPT is good at
ChatGPT and similar assistants are excellent writing tools. Given a clear brief, they deliver a structured draft, write subheadings, suggest FAQs, shorten, expand and change the tone on request. If you can brief well, you get good drafts; the briefing template for AI texts shows what belongs in such a brief. A step-by-step guide is in the post Writing blog posts with AI in seven steps.
So for the writing itself, the assistant isn't a drawback. But the text is only one of eight steps a blog post needs before it's online and found.
What ChatGPT alone doesn't do
| Step | With ChatGPT | With a blog system |
|---|---|---|
| Choosing a topic from real search queries | You research yourself or guess | The system tracks search queries for your industry and region |
| Keeping to a rhythm | Your discipline, week after week | Fixed rhythm, the post arrives without prompting |
| Swiss Standard German and form of address | Set it in the prompt every time | Set up once |
| Checking facts | You, for every post | You, at approval |
| Internal links to existing posts | You, by hand in the CMS | The system knows your own posts |
| Image | You search for or generate one | Included, depending on the system |
| Publishing in the CMS | Copying, formatting, setting title and description | Directly via the connection |
| Approval | Not applicable, you're both author and reviewer | By email or click, with the option to stop |
Two of these points carry more weight than the others. The first is the topic: ChatGPT doesn't know your customers' search queries. It suggests what's generally plausible, not what's actually being searched for in your canton right now. A blog whose topics don't come from search data often misses the demand. The second is the rhythm: ChatGPT won't remind you on Tuesday that a post is due. The discipline stays with you, and that's why most self-run blogs grind to a halt after a few weeks. The page Write it yourself or use a system describes this pattern in more detail.
The sum: time versus subscription
The example calculation below assumes one post per week. The times are experience based values for someone who's comfortable using ChatGPT but doesn't work with it every day.
| Step with ChatGPT | Time per post |
|---|---|
| Choosing a topic and writing a brief | 20 minutes |
| Generating a draft, two to three rounds of revisions | 30 minutes |
| Checking facts, figures and Swiss spelling | 25 minutes |
| Sourcing an image | 15 minutes |
| Setting it up in the CMS: title, description, formatting | 20 minutes |
| Adding internal links | 10 minutes |
| Total | around 2 hours |
At four posts a month, that's eight hours. At an assumed hourly rate of CHF 100, that's CHF 800 a month in your own time, plus the small subscription fee for the assistant.
With a blog system, what's needed per post is the approval, that is reading it and one click, roughly 10 to 15 minutes. For four posts, that's around an hour a month, so CHF 100 in your own time, plus the subscription. For comparison: BlogPilot's Starter plan costs CHF 199 a month for one post a week (excl. VAT, as of October 2026). The sum then reads CHF 800 against around CHF 300.
This calculation shifts if your hourly rate is lower, if you only need one post a month, or if you enjoy writing with the assistant. It shifts the other way if you want to publish weekly and can't actually find the eight hours in practice; then ChatGPT doesn't cost CHF 800, it costs you a blog that doesn't appear.
When ChatGPT is enough
There are good reasons to stick with the assistant:
- You need one or two posts a month, not per week.
- You already have a list of topics, for example from customer questions or from Search Console.
- You enjoy working with the tool and keep to deadlines even without external pressure.
- Your posts need a very personal voice that you can capture better in dialogue with the assistant than with a system that has fixed settings.
- You run the blog as a side activity and don't want to be tied into a contract.
In these cases, ChatGPT isn't the cheap option, it's the right one. You can cut the effort further with templates: a saved brief with company, region, tone of address and spelling conventions, a fixed process for checking and publishing, and a calendar entry for writing day.
When the system is the better route
- You want to publish weekly or more often, over several months.
- The topics should come from real search queries, not guesswork.
- The blog should keep running even when the business is at full stretch.
- Publishing should happen directly in the CMS, without copying and formatting.
- One person in-house should be able to handle approval in 15 minutes per post.
Both routes lead to texts that Google, by its own statement, treats the same way: what counts is whether the post is helpful, not what it was written with. The difference lies in who handles the seven steps around the writing, and whether they're still being handled in week 30.
Frequently asked questions
Can I just ask ChatGPT to write a post every week?
ChatGPT writes when you ask it to, and forgets what happened last week between sessions, unless you use a project function or templates. You have to organise the weekly rhythm, the topic list and the publishing in the CMS yourself. This organisation is exactly the part where most self-run blogs fail.
How does ChatGPT know what my customers are googling?
It doesn't. ChatGPT can suggest plausible topics, but it has no access to the search queries from your industry and region. For a topic list based on real search data, you need Search Console, a keyword tool, or a system that analyses this data.
Are texts from ChatGPT a problem for Google?
In 2023, Google stated that it doesn't penalise content based on how it was created, but assesses its helpfulness. AI texts become a problem when they're published in bulk without review or contain no first-hand experience. A reviewed, specific post from ChatGPT is a normal post as far as Google is concerned.
What does ChatGPT cost compared to a blog system?
The subscription fee for ChatGPT is small; the real cost is your time for briefing, review, images, publishing and linking. A blog system costs a fixed monthly fee and only needs approval per post. Which side is cheaper depends on your hourly rate and the number of posts.