Credible Roots / Editorial standards
Editorial standards
How this is written, and who stands behind it
We argue on this site that generated video should be disclosed and that unsourced claims should be distrusted. It would be inconsistent not to apply both to ourselves.
Every field note here is researched against primary sources first, drafted with AI assistance, then read, corrected and approved by the person named on the byline before it publishes. We disclose the AI assistance because a reader might reasonably want to know, and because arguing elsewhere that avatar video should be disclosed while hiding our own method would be incoherent. What we do not publish is anything written to a word count, anything that restates what the first page of results already says, or any precise figure we cannot trace to a source. Where we get something wrong, we correct it on the page and say that we did.
Why these notes exist
Two reasons, and it is worth being straight about both.
The first is that this market is full of claims that do not survive checking. Guaranteed encyclopedia entries, knowledge panels for sale, precise numbers attributed to research nobody has read. Publishing the accurate version is the clearest way to demonstrate that we know the difference, and it is more persuasive than asserting it.
The second is commercial and we are not going to pretend otherwise. People who read this material and find it useful sometimes hire us. That is the business model. What it does not change is the standard: if a piece would only exist to attract search traffic, it does not get written.
How a field note is made
Primary sources first
Before drafting, the claims are checked against the source itself. Wikipedia's own guidelines rather than a summary of them, the Wikimedia Terms of Use rather than a blog post about them, Google's published documentation rather than an agency's interpretation. Where a widely repeated claim turns out to have no traceable source, we say so.
Drafted with AI assistance
Drafting is AI-assisted. We are disclosing that because it is the sort of thing a reader might reasonably want to know, and because arguing elsewhere on this site that generated video should be disclosed while hiding our own method would be incoherent.
Reviewed and approved by the named author
Nothing publishes unread. The person on the byline checks the claims, corrects what is wrong, cuts what is padding, and stands behind the result. A byline on this site means that review happened, not that a person typed every word.
Sources listed
Where a piece rests on documentation, the documentation is named at the bottom. Where a claim is a practitioner report rather than research, it is labelled as one.
What we will not do
- Pieces written to a word count. Google states there is no preferred length, and padding is obvious to read.
- Restatements of what the first page of search results already says.
- Precise figures we cannot trace to a source, however convenient they are for us.
- Content on topics we have no business writing about, chosen because they attract traffic.
- Date changes to make old pieces look fresh when nothing substantive changed.
- Claims about outcomes we cannot demonstrate.
Corrections
If something here is wrong, tell us and we will fix it and note the correction on the page rather than editing silently. Two examples of the kind of thing we would rather catch than defend: a widely cited marketing rule attributed to Google that Google appears never to have published, and a LinkedIn reach figure repeated across the industry that we could not trace to any published study. Both are addressed in the relevant notes and logged on our corrections page.
Corrections can go through the contact page.
Who writes this
More than five years in personal branding. Runs atharvsankpal.com and works with founders and businesses in the US on their digital presence. Previously produced AI video for brands.
Published 2026-08-14
Researched against primary sources, drafted with AI assistance, then reviewed and approved by the named author. How we write · About the author
Every field note carries a byline for a reason. If you cannot tell who stands behind a claim, you have no way to weigh it, which is the same argument this site makes about everything else.
Questions people actually ask
Is this content written by AI?
It is drafted with AI assistance and it is not published unreviewed. Research is done against primary sources first, drafting is assisted, and the person named on the byline reads, corrects and approves every piece before it goes out. We are telling you this because a site that argues for disclosing AI avatar video would be inconsistent not to disclose its own method.
Why disclose it at all?
Because readers reasonably want to know how something was made, and because our whole position is that credibility is built on things that can be checked. Concealing the method while selling honesty would be the wrong trade.
How do you decide what to publish?
One test: does the piece contain something you could not get from the first page of search results. If it is a restatement of what everyone already says, we do not publish it, regardless of how well it would rank.
What do you do when you get something wrong?
Correct it and say so on the page. Any factual correction after publication is noted rather than quietly edited, because a silent fix is indistinguishable from having been right all along.
Do you use word counts or keyword targets?
No. Google states plainly that there is no preferred word count and that keyword stuffing is a spam policy violation. Pieces here run as long as the argument takes and no longer.
Related
Sources
- Google Search Central. Creating helpful, reliable, people-first content — including the statement that there is no preferred word count.
Where a claim rests on documentation, the documentation is linked above rather than summarized. Corrections go on our corrections page.
Found something wrong?
Tell us and we will correct it on the page. That is a lower bar than most of this industry sets, and it is the whole standard.
Get in touch