Content policy

01Why this page exists

We produce text with the help of language models and publish it to customer websites. That is a position of some responsibility, and the honest thing to do is write down the rules we operate under rather than leave you to find out where the edges are.

This policy binds us and it binds you. We apply it to what our system generates, and you agree to it when you publish through us.

02Human review is not optional

Nothing publishes without a person approving it. Not on any plan, not with any setting, not for any customer who asks nicely. This is the single design decision the rest of the product is built around.

The reason is practical rather than sentimental. Search engines have drawn their line at oversight rather than authorship, and content published at volume with nobody accountable for it is the exact pattern that gets demoted. There is a longer version of this argument on our blog.

Approving in bulk is possible and sometimes sensible. Approving without reading is your call to make and your risk to carry.

03What we will not write

Some of these are legal lines and some are ours. We do not distinguish between them in practice.

• Specific medical, legal or financial advice presented as professional guidance. We will write about a regulated industry; we will not write the advice itself.
• Claims about health outcomes, earnings or investment returns that the customer cannot substantiate.
• Content that names a real person or company in order to disparage them, or that presents invented quotes as real.
• Political persuasion, election content and material designed to influence public opinion on a live vote.
• Adult content, gambling promotion aimed at minors, and anything promoting self-harm, extremism or illegal goods.
• Copies of a competitor's page. Our similarity gates reject drafts that read too closely to an existing source, ours or anyone else's.
• Pages generated per location, per product variant or per keyword permutation with no substantive difference between them.

04Accuracy and sources

Every factual claim in a draft has to trace back to one of two places: your own connected business data, or a source the system retrieved and recorded. Citations are checked to confirm the page exists and actually says what the draft claims it says. Drafts with unverifiable claims are rejected before they reach your queue.

That process is good and it is not perfect. Models still produce confident errors, particularly with numbers, dates and attributions. This is the specific thing your review step is for, and it is why we highlight claims and citations separately in the queue rather than presenting a wall of finished prose.

05Originality

• Every draft is checked against your existing pages, so we do not quietly cannibalise something you already rank with.
• Every draft is checked against the other drafts in your plan, because forty articles that say the same thing is the classic failure mode of automated content.
• Passages that closely match a retrieved source are flagged and either rewritten or quoted and attributed.
• Where a draft has nothing to add to what is already ranking, the better outcome is to not publish it. The system will say so.

06Volume limits

New sites are capped in their first 60 days regardless of plan. A domain that goes from four pages a month to two hundred looks like abuse whether or not the writing is good, and the pattern alone is enough to cause problems. Customers ask us to lift this and we do not.

07Disclosure

You are not required to label content as model-assisted, and search engines do not require it either. We think a byline naming the person who reviewed and approved the piece is worth more than a disclaimer about the tooling, and the review queue records exactly that.

If your industry, your regulator or your own editorial standards require disclosure, that is your call and we will not argue with it.

08Your side of this

• Read what you approve, or accept the consequences of not doing so. You are the publisher.
• Make sure the business information you connect is accurate and current. Wrong prices in your catalogue become wrong prices in your articles.
• Do not edit an approved draft to insert claims we rejected. If it comes back to us, that is a breach of these terms.
• Tell us if something we produced turned out to be wrong. It helps us fix the gate that let it through.

09When something goes wrong

If we find content that breaches this policy, we will contact you and ask you to remove or correct it. Where the breach is serious or repeated, we may unpublish it through the connection we already have, and we will tell you when we do.

Accounts producing content deliberately designed to break these rules get suspended. We would rather lose the revenue than be the supply chain for it.

10Reporting a problem

If you have found content published through ScaleRank that breaches this policy, send the URL and a short description to legal@scalerank.io. We investigate every report and will tell you what we did. Complaints about our crawler belong in the bot and crawler policy instead.