Pre-publish scan for Substack, Ghost, WordPress, Webflow, and Wix posts before they reach your readers or Google's helpful-content classifier. Sentence-level highlights show which lines drift into AI-template phrasing so you can rewrite specific sentences instead of the whole article. Built for bloggers, niche publishers, and Substack-adjacent writers. Free tier covers a typical post. No card.
Independent bloggers running a personal site, hobbyist writers publishing a few posts a month, and monetised creators with AdSense or sponsorships share the same pre-publish need: a quick scan that catches AI-template phrasing before readers, advertisers, or Google see it.
Blogging in 2026 is a different risk model from agency content work. Independent bloggers are not delivering to a single client; they are publishing to an audience that chose to follow them. The stakes are not a single underperforming page; they are reader trust and a monetisation stack that took years to assemble and can erode in three flagged posts.
Personal site, one writer, niche audience. Posts shipped on a weekly or biweekly cadence. The free tier ships three scans a day at 5,000 characters per scan, which covers a typical 800-word post in one pass or a 1,500-word post in two. For independent bloggers shipping less than once a week, free is genuinely enough to maintain a consistent pre-publish scan without ever upgrading.
Substack, Medium, or Ghost newsletter. Writing for love rather than income. Pre-scanning still matters because readers can spot AI cadence faster than most hobbyist writers expect, and the unsubscribe is silent. The free tier covers monthly cadence; Starter at $9.99 a month covers the one-to-three posts a week zone where the daily scan limit starts to bite.
AdSense, sponsorships, affiliate placements, or paid Substack subscriptions. Voice is the monetisable asset, and one flagged post can hit AdSense compliance or void a sponsor deliverable. Pro at $19.99 a month standard, $14.99 a month on yearly, fits this segment: unlimited scans, 50,000 AI rewriter words a month, 10,000 characters per scan, and 90-day scan history covering a quarter of publishing.
Three shifts in 2024 and 2025 turned pre-publish scanning from a power-user habit into a baseline workflow for any blogger who cares about traffic, ad revenue, or sponsor relationships.
Google's March 2024 core update and the 2025 scaled-content-abuse policy specifically target sites publishing AI-only content without editorial layer. The penalty distribution looks like: ranking erosion at first, manual action at scale. Pure ChatGPT output on a blog risks an algorithmic demotion that wipes organic traffic for a quarter. The defensive workflow is AI-assisted drafting plus a human editorial pass plus a pre-publish AI scan to confirm the prose reads human enough to clear both Google's classifiers and reader perception.
AdSense's 2025 content policy update allows account suspension on sites publishing scaled AI content without editorial value. Direct sponsorship contracts moved faster: most 2026 sponsorship agreements now include an AI-disclosure clause and a clean-detector-scan deliverable as standard. A flagged sponsored post can void the placement fee and end the relationship. The cheapest insurance against both is a pre-publish scan.
Engaged readers have learned to recognise AI prose the same way they learned to recognise stock photography a decade ago. Substack and Ghost analytics both surface unsubscribe upticks within 48 hours of posts that score low on AI detectors. The reader-side risk is the bigger long-term concern because the audience is small enough that every subscriber matters and the unsubscribe is permanent.
Free tier handles a typical post and covers most personal blogs without ever upgrading. Pro at $19.99 a month standard, $14.99 a month on yearly, is the right fit for monetised blogs shipping three or more posts a week. Full details on the pricing page.
Billed $89.88/year — Save $30
Billed $179.88/year — Save $60
Billed $359.88/year — Save $120
Yearly billing saves 25%. View full pricing →
Most bloggers draft in Notion or Google Docs and paste into the CMS at publish time. The scan slots into that flow at the paste step. Thirty seconds for a clean draft, a few minutes when there is residue to rewrite.
Most bloggers keep a running ideas file. AI tools are fine here; nobody scans an outline. The detector enters the workflow only at the publish step. Keeping the AI part of the workflow as upstream as possible is fine for readers and for Google: it is the finished prose that gets evaluated, not the brainstorming.
Write the prose yourself, in your voice, across a few sittings. The temptation to paste an AI-drafted section grows on deadline days, especially for Substack writers shipping on a Tuesday or Thursday cadence. This is the leakage point. The discipline is to use AI for stuck moments and then immediately rewrite the assist into your own register.
Paste the full draft into TextSight before you copy it into Substack, Ghost, WordPress, Webflow, or Wix. Read the Authenticity Score. Above 85 means your readers and the platform classifiers will read it as yours. 70 to 85 means look at the highlighted sentences and rewrite the ones that flag. Below 70 means the post needs a real voice-pass; better to hold it a day than ship a flagged piece to your list.
Once the scan reads clean, paste into your platform of choice and ship. The whole pre-publish step takes 30 to 60 seconds for a clean draft. After the first month, the scan becomes a reflex the same way running spellcheck did a decade ago, and the rewrites get faster because you internalise which phrasings your voice does not actually use.
A flagged blog post is not just a ranking problem. AdSense compliance, direct sponsorship contracts, and affiliate placements all now treat AI-content as a material risk. Pre-scanning is the cheapest insurance.
Google updated AdSense's content policy in 2025 to allow account suspension on sites publishing scaled AI content without editorial value. The enforcement looks gentle at first (warning, demonetisation of specific URLs) and escalates to full account suspension on sites that ignore the warnings. For monetised bloggers, an AdSense suspension wipes the income stack overnight. Pre-publish scanning catches the pattern before AdSense does.
Most 2026 sponsorship agreements include an AI-disclosure clause and a clean-detector-scan deliverable as standard. A flagged sponsored post can void the placement fee and, more importantly, end the relationship with the sponsor. Attaching a TextSight scan to the delivery email pre-empts the conversation: most sponsors stop running their own check after one or two clean delivery cycles.
Affiliate networks like ShareASale, Impact, and Awin started flagging AI-heavy content in 2025 because the conversion rates on flagged pages run materially below human-written pages. Some networks now demote affiliate links on flagged URLs, reducing payouts even if the placement stays live. The fix is the same as for sponsorship: clean pre-publish scan, attach the screenshot to the placement record if the network asks.
Substack does not police AI use directly, but the unsubscribe pattern does the policing for them. Paid subs at $5 to $10 a month churn fast when posts read AI-template. The math is straightforward: at 100 paid subscribers at $5 a month, one reader who stays because a post sounded like you covers the Starter plan for the entire year.
Engaged readers spot AI cadence faster than most bloggers expect. They rarely comment about it. They unsubscribe, stop sharing, or quietly drop out of the comments. The audience erosion compounds across posts.
Engaged readers have internalised your sentence rhythm, your favourite words, and your tics. When a post slides into AI-template phrasing, they notice on the first paragraph. They might not comment. They will quietly unsubscribe. Substack ESPs and Ghost analytics started surfacing this pattern in 2025: posts with elevated AI signal density correlate with measurable unsubscribe upticks within 48 hours of send.
The early warning sign is not the unsubscribe number; it is the comments section. Engaged readers stop commenting on posts that read AI before they unsubscribe. Watching the comments-per-post number across your last ten posts is the leading indicator. A sudden drop usually means a piece (or several) read AI to the audience in ways the writer did not catch.
Posts that read AI get fewer shares, fewer pull-quotes on social, and fewer "this resonated" replies in the email response. The drop is visible inside two or three weeks of the publishing pattern shifting. Pre-scanning before publishing is the single workflow change that prevents the share-pattern erosion before it starts.
SEO writers can recover a flagged page by editing it. Bloggers cannot recover a reader who left because last Tuesday's post sounded like ChatGPT. The cost of one AI-template post is not the post; it is the relationship. Pre-scanning protects the asset (your voice) that readers actually follow you for.
Listicles, how-to guides, reviews, opinion pieces, tutorials, and evergreen guides each have their own register and their own false-positive risk. Read the score in context of the format rather than chasing a single number across every type of post.
Numbered or bulleted format where every item gets a short paragraph of supporting prose. Listicles tend to score lower because the per-item paragraphs follow the same structural rhythm. Healthy scores run 65 to 80. Sentence-level highlights matter more than the headline here because the issue is usually two or three flat items, not the whole article.
Step-by-step content with explicit numbered steps and standard transitional phrasing. Reads templated for the same reason a methods chapter does: a sequence of identical structural moves. Scores tend toward 60 to 75. The defence is varied step-intros and concrete first-person examples woven through, both of which read human to the classifier and read useful to the reader.
Reviews of tools, books, products, or services. Common structure is intro, pros, cons, verdict. Scores volatile because the structure is uniform across the category by convention. Specific concrete vocabulary and a real point of view lift these pieces faster than rewriting the whole structure. Reviews that read like personal experience score the highest.
Personal essays, hot takes, and commentary. These typically score the highest of any blog format (often 85 to 95) because a strong point of view varies structurally from the templated AI default. The pieces that flag are usually the ones where the writer drafted in a neutral register to push through a deadline. Catching that drift on a scan is the diagnostic.
Long-form reference content meant to rank for years. Healthy scores run 75 to 90 on prose written from a real outline. The intro and the closer are the highest-risk paragraphs because both default to stock phrasing when drafting moves fast. Scan the full piece, then re-scan the intro and closer alone if the headline number is borderline.
A single percentage is not a fix path. The TextSight result panel shows which sentences reacted and why, with paragraph-level rollups for longer posts, so you can edit specific lines instead of rewriting the whole article.
Every sentence is colour-coded by its own AI-likeness score. Red sentences clustered in one paragraph are a stronger signal than scattered yellows. Scattered yellows in otherwise structured prose often just mean a stock transitional phrase. You read the pattern, not just the headline number, and rewrite the lines that actually flagged.
Longer posts get paragraph-level rollups so you can see which paragraph is dragging the headline score. The intro and the closer are the usual suspects on opinion pieces and evergreen guides, while the per-item paragraphs are the usual suspects on listicles. Targeting the lowest paragraph first is the fastest way to lift the article.
Perplexity is how predictable your word choices are to a language model. Low perplexity reads AI-like. The score is shown per-sentence on Pro, which is the diagnostic context you need to decide whether a flag is real AI residue or a particularly well-rehearsed turn of phrase you have used in a dozen prior posts.
Burstiness is how much your sentence length and structure vary across the post. ChatGPT defaults to uniform medium-length sentences. Real human writing has bursty rhythm: one short sentence, one long, one fragment. Low burstiness across an entire post is the classic AI fingerprint and the one readers learn to spot first.
More for bloggers.
Full content-writer landing page with the delivery-attached scan workflow.
For writers →CMS automation and the ranking angle for content built to rank.
For SEO writers →Light, Balanced, and Maximum modes for fixing flagged passages without losing voice.
Read the guide →Free, Starter, Pro, Business. Yearly billing saves 25%. Hobby to monetised tiers.
See pricing →Free tier covers a typical post. No card. Pro at $14.99 a month on yearly for monetised blogs shipping three or more posts a week.