Quality & safety

Automated content is only useful if it is good, consistent, and safe to publish under your name. Several systems run on every single piece theStacc generates to make sure of that. None of them are settings you have to babysit - most work quietly in the background, and the few you can tune are explained below.

Brand voice#

Upload a few of your existing articles or posts and theStacc learns how you write - tone, vocabulary, sentence rhythm, and the words you never use - then applies that voice to everything it generates. Your automated content sounds like you, not like generic AI.

You can also keep a short words to avoid list. Anything on it is treated as off-limits and will not appear in your content.

Grounding in your sources#

Upload documents (PDF, Word, or text) or paste in material, and the engine treats it as ground truth while writing. This keeps your facts, product details, and claims accurate instead of invented. When you give theStacc real source material, it leans on that material rather than guessing.

Anti-repetition#

Publishing on a schedule only works if the content stays fresh. theStacc actively steers each new post away from everything it has already written for you:

• Keyword rotation - the content calendar will not target the same keyword twice in a rolling window, so you never get two posts chasing the same term back-to-back.
• Title and topic de-duplication - the writer is shown your full title history (up to 200 prior posts). If a new topic overlaps an existing one, it differentiates the angle: a different sub-topic, audience, stage of the buyer journey, or framing. Two posts in your library should never read as alternates of each other.
• Opening and hook diversity - the writer is also shown the actual opening lines of your most recent posts and told to avoid reusing the same hook. If your last few posts all opened with a year stamp ("In 2026...") or the same statistic, the next one deliberately opens a different way - a specific scenario, a counter-intuitive claim, a direct question, or a fresh number.

SEO scoring#

Every blog post is scored 0-100 so you can see at a glance how search-ready it is, with a category-by-category breakdown and specific fixes on the blog's SEO analysis panel.

The headline number is a blend of two independent checks:

• A mechanical score - a deterministic check of the things that can be measured exactly: keyword placement in the title and headings, content length, heading structure (H2 to H3 hierarchy, a subheading roughly every 200-300 words), image count and alt text, and internal and external links.
• A semantic score - a judgment of the things only a careful read can catch: topical depth (did the post actually cover the subject at the breadth a knowledgeable reader expects?), whether the structure matches what the searcher wanted, and how genuinely helpful it is.

The two measure different things on purpose, and the score you see is weighted toward the semantic read because that is what modern search engines reward.

The score is a signal and a quality bar - it tells you (and you can tell at a glance) which posts are strongest and exactly what to improve. A few hard problems trigger an automatic rewrite before the draft is ever saved (see the minimum-quality gates below); everything else is surfaced as a score and suggestions so you stay in control of what publishes.

Topics to avoid#

Set a per-project "do not write about" list and theStacc steers clear of those subjects entirely - it will not discuss, reference, or promote them anywhere in your content. This is useful for competitors you would rather not name, sensitive subjects, or off-brand themes. You can list up to 50 topics per project.

Content-integrity guardrails (always on, never optional)#

Beyond your settings, every blog runs through a set of integrity rules that protect your business from content that looks fine but could create real legal or reputational risk. The writer is instructed never to invent these things, and a separate automated check scans the finished draft before it is saved. If it finds a serious problem, the post is sent back to be rewritten - it does not reach you with the problem in it.

theStacc will never fabricate:

• Promo codes, coupon codes, or discount codes (no invented "SAVE20" or "WELCOME10").
• Specific discount percentages or dollar amounts (no "20% off" or "$50 off" unless you provided it).
• Specific prices (no "$99/month" unless it is in your business details).
• Phone numbers or email addresses in place of a real contact path.
• Statistics, studies, or research data without a real, citable source.
• Testimonials, reviews, customer quotes, or star ratings - never an invented "They were amazing! - Jane D."
• Awards, certifications, or rankings you did not actually earn.
• Partnerships or endorsements that imply an affiliation you do not have.
• Leftover bracket placeholders or template text like [your product], [company name], or [competitor] - every mention uses your real business name and details.

Instead of any of the above, the writer falls back to safe, general language: "contact us for current pricing," "ask about our latest offers," or a pointer to your website for specifics.

These rules cannot be overridden#

The integrity rules are inviolable. They sit above your own per-post instructions. You can steer almost everything about a post - its angle, what to emphasize, what to skip, who it speaks to - but you cannot direct theStacc to fabricate the things above.

If you ask theStacc to "include promo code SAVE20" or "mention our 30% off sale," it reads that as a request to talk about promotions in general. The post will invite readers to ask about your current offers, but it will not print a specific code, percentage, or price unless that exact detail is in your verified business information. The same is true for fabricated reviews and quotes - those can never be generated, for any business, for any reason, because inventing them creates liability for you and for the platform.

Minimum-quality gates#

A handful of checks act as hard floors. They run automatically on every draft:

• Minimum length - a post that comes back too short (under roughly 500 characters) is treated as truncated and rewritten rather than saved.
• External links recommended - the writer is told to include at least two links to authoritative outside sources (industry sites, government data, studies). A post with fewer external links scores lower on the SEO check, which surfaces it to you as something to improve.
• AI-cliche and "robot" pattern detection - the mechanical scorer subtracts points when it spots the patterns that make content read as machine-written: stock AI opener phrases ("in today's fast-paced world," "when it comes to," "let's dive in"), overused scaffolding transitions ("Additionally," "Furthermore," "Moreover" stacked up), several paragraphs in a row that all start with the same word, and em-dash or en-dash overuse. Each pattern costs the post points, so cleaner writing scores higher.
• Keyword-stuffing detection - keyword density is not a ranking signal, so theStacc does not reward padding. If a keyword is repeated to an unnatural degree, the score is reduced. The goal is natural prose, not a target density.

Together these keep weak or padded posts from quietly slipping through while still leaving the final publish decision with you.

Compliance (for regulated businesses)#

Businesses in regulated fields - legal, healthcare, financial, real estate, and similar - can turn on an optional compliance layer. It is off by default, and the vast majority of projects never need it. When you enable it, theStacc does two things:

• At writing time, it injects the disclosures and cautious language your field expects into the content as it is generated.
• Before publishing, it routes each post through a review gate that checks the content against your compliance profile and holds anything that needs a human look.

Compliance is a suggestion and a gate, not a guarantee. It helps you catch issues and enforces a review step, but it does not certify that any individual post meets your professional obligations. The licensed professional stays responsible for what is published under their name.

If theStacc detects that a project looks like a regulated business, it may suggest turning the layer on - but it never enables it for you and never blocks your publishing on a hunch. Setup is opt-in. See Compliance Profiles for how to turn it on and configure it.

Built-in platform guardrails#

Beyond your own settings, theStacc enforces the rules of each publishing destination automatically:

• Google Business Profile posts are kept within Google's content policy - the kind of promotional content Google prohibits (like phone numbers and sales-y promo codes in the post body) is handled so your posts are not rejected.
• Social captions are kept within each network's limits. The platform knows, for example, that a standard X (Twitter) post caps at 280 characters while other networks allow much longer captions, and it sizes each caption accordingly.
• Fabricated reviews, testimonials, statistics, and client names are blocked on social posts too - the same integrity standard applies across blogs, local posts, and social content.

Learn more#

• AI Content Safety & Integrity Guardrails - the full list of what theStacc will and will not generate, and why these rules cannot be overridden.
• Compliance Profiles - opt-in setup for regulated businesses, the writing-time disclosures, and the pre-publish review gate.
• The content engine - where quality and safety fit into the full generation pipeline.
• Publishing - how review queues and publishing modes work.