Library · Blogs
Blog drafting that stays honest
Write client posts against the positioning you already agreed, beside a guide rail whose every number — counts, reading time, readability, structure checks — is measured from the draft in front of you. No invented search volume, no score out of a hundred.
Library · Blogs
Blog posts are the part of a retainer that quietly slips: the brief lives in one document, the keywords in a spreadsheet, the draft in somebody’s personal folder, and the finished piece arrives with a score out of a hundred nobody can explain. Blogs in AI Marketing Dashboard puts the writing environment where the client’s positioning, themes and keyword lists already are — a distraction-free block editor beside a guide rail whose every number is measured from the draft on the screen. There is no fabricated search volume anywhere in it, and no composite content score, because the tool has no such data and will not invent it.
A blog post that argues a different case from the client's ads is a wasted week. Creating a post asks which Theme it is for — the positioning angle, audience and message pillars your team already wrote for that client — and the answer sticks to the post. The list is then grouped by theme, so a client's content programme reads as the handful of arguments it is actually making rather than as a pile of titles.
That connection is what makes the drafting surface useful rather than generic: the writer opens the editor with the angle in front of them, and a strategist skimming the list can see at a glance that three posts exist for one pillar and none for another. A post whose theme is later removed folds into an Unlinked section and stays editable, so a strategy change never strands a draft.
The writing column is deliberately plain: one block at a time, in the seven kinds a blog post is actually made of — two heading levels, paragraphs, block quotes, bulleted and numbered items and dividers. Bold, italic, inline code, strikethrough and links are typed as marks inside the text, and a block shows its rendered form when you leave it and its source when you are in it, so nothing is hidden behind a formatting toolbar.
That is the same editing grammar the meeting notes use elsewhere in the console, which means a writer who has taken notes in a client meeting already knows how to write here. Drafts save themselves as you type on a short debounce, so there is no Save button to forget and no lost paragraph when a laptop closes mid-sentence.
The rail beside the editor counts what is on the page: words, characters, sentences, and a reading time that names the rate it used rather than presenting a mysterious minute figure. It also gives a Flesch reading ease with its reading level, labelled as computed from this draft's own sentence and word lengths — a formula you can look up, not a proprietary grade.
Everything in that rail moves because the writing moved. There is no keyword difficulty, no competitor gap, no invented score out of a hundred, because none of those can be honestly derived from a text file and the tool does not pretend otherwise. What is left is smaller and completely trustworthy, which is the trade we chose on purpose.
The guidelines cover the things that genuinely decide whether a post is readable and shareable: a title in a sensible length, a meta title inside what a search result displays, a meta description that is neither missing nor cut off, an opening that starts on prose rather than a heading, at least a couple of real sections, paragraphs that have not run to a wall of text, something that breaks the page up, at least one link, and a reading ease above the floor. Three more appear only once they apply — a word target you set, and two for the terms you are tracking.
They are reported as a count of what is met out of what is being checked, each with one plain sentence saying why it matters. A checklist that is permanently three items short trains people to ignore it, so a check you cannot satisfy yet is simply not shown — and nothing here claims to know what a ranking algorithm rewards.
A post can carry the handful of terms it is meant to be about, offered straight from that client's keyword lists so a writer is picking from the research the team already did instead of typing a phrase from memory. For each one the rail shows where it actually appears — title, headings, body, meta — which is the difference between a term you wrote about and a term you mentioned once in paragraph nine.
What it deliberately does not attach is a volume or a competition figure, because the tool has no live search data behind it. Naming the terms and counting the usage is honest; dressing that up with a number nobody measured is not, and a client who spots the difference stops trusting the rest of the report too.
Meta title, meta description and slug are fields on the post, not an afterthought bolted on at publish time, and a preview card lays them out the way a result is laid out so a writer can see where a description would be cut. The card says in plain words that it is a formatting preview and not a live search result, and it fetches nothing.
It exists because the meta fields are the part of a post that gets written last, badly, by whoever is pasting it into the CMS. Writing them beside the draft — while the argument of the post is still in your head — is how they end up describing the article rather than repeating its title.
The Assist panel drafts from the theme you attached and the words you typed: working titles across five shapes, a section plan with a note on what each section is for, an opening paragraph, a first pass at the section under a chosen heading, or search descriptions inside the length that displays. The same ask always returns the same drafts, and the panel says exactly where they came from.
It is scaffolding to rewrite in the client's voice, and it is labelled as such rather than dressed up as authorship. That honesty is the point: a starting structure removes the worst twenty minutes of writing a post, and pretending a machine wrote the finished article would only make an editor distrust every line of it.
The list is scoped to one client at a time and grouped by theme, with a theme filter, draft and ready states, search across titles and a word count on every row, so an editor can see the whole content programme for an account and where each piece has got to.
Shipping is an export, not a publish pipeline we would have to pretend integrates with every CMS on earth: copy the post as Markdown or as HTML and paste it where the site is. The HTML export escapes text and drops any link whose scheme is not a normal web one, so what leaves the tool is safe to paste into a page.
