How we write the Placton blog

01Who writes the blog

Articles are signed by Raúl Díaz, founder of Placton, who is responsible for what the blog publishes. Placton is our product: when an article mentions it, it says that we make it.

02How we choose each topic

We start from real search data (monthly searches and difficulty, from Semrush and DataForSEO) and only take topics close to what Placton measures: how AI assistants answer, which brands they recommend and which sources they cite. We leave out searches about hiring agencies or services.

Each search gets one page. Before writing, we compare the current Google top 10 for the search with the top 10 of our existing pages; if three or more results match, both pages would answer the same thing, so we do not write a new one: the search stays with the page we already have. We also compare the planned title and sections with our other pages, and drop the article if it would repeat one.

03What we read before writing

Before writing, we read the pages that rank for the search at that moment (their headings, data and dates) and the questions Google shows under “People also ask”. The aim is to answer what they answer, more clearly, and to cover what they leave out.

04Our own data

Most articles include a block with our own data. We ask ChatGPT two to four real questions about the topic (on chatgpt.com, logged out and with web search, through DataForSEO); if that fails, we ask Claude through its API. We record which brands the answers name, in what order, and which websites they cite.

The block always says which assistant we asked, from which country, on what date and how many answers we got. It is a snapshot of a few answers, not a study, and the article says so.

05Drafting with AI assistance

The first draft is written with Claude, a model made by Anthropic, from a brief built with the steps above. A second pass edits it to remove filler and stock phrases.

The rules are strict. Every figure, date or claim about another company has to come from a page we read or from our own data. What we say about Placton comes from a fact sheet generated from Placton’s own code, including what it does not do. We do not invent statistics, quotes, customer stories or experience we do not have.

06Checks before publishing

• A second model reads the draft against the sources and flags any claim without support.
• Automatic checks of structure, length, headings, links and readability, and of style: no stock phrases and no keyword stuffing.
• Links to other websites only go to official documentation, studies and specialist industry press from a list of reference sites, and we check that they load.
• We compare the text with our other pages so we do not publish near duplicates.
• An article that fails stays as a draft and is never published automatically.

07Human review

Raúl Díaz approves every article before it is published, and edits it first when needed. Those articles say “Reviewed by Raúl Díaz” under the byline. The update date only changes when the text itself changes.

08Images and charts

Cover images are generated automatically from the article title. Charts are drawn from our own data, and their caption says where and when the data was collected.

09Corrections and updates

Each article shows when it was published and, if it changed later, when it was last updated. If you spot a mistake, write to contact@placton.tech and we will fix it.

10Links and independence

Articles are not sponsored. Links to other websites are chosen as sources and are never sold. When we compare tools, Placton is one of them, and we say so.