Standards
Ethical AI reputation management: no fake reviews, no fake sources and no hiding facts.
These principles govern every matter we accept.
We work from the primary record.
Your published pages and documents are the reference for every judgment we make about what AI says.
Confidentiality by default.
A brief, its data and its findings serve that client alone. An NDA is available before you share a brief. Client material appears in public examples only with written permission, and client data is deleted at the end of an engagement unless you ask us to keep it.
No suppression, astroturfing or fake reviews.
We do not try to remove information from AI models or suppress search results, and we do not create reviews, sources or coverage. Clients decide what to publish.
We say when the evidence does not settle a question.
An unresolved point is reported as unresolved, together with what would resolve it.
We take on work only where it is useful.
If the answers already reflect your record, we tell you during scoping.
We act to state the record accurately.
We do not act to obscure an established fact, and we may decline a matter after checks.
How a finding is written
Every finding keeps four parts separate, so each colleague can check the part that concerns them.
- Observed statement.
- What the model said, with its hedges and caveats intact.
- Primary record.
- The passage in your own material the statement should match.
- Interpretation.
- Our reading of the gap and why it matters to the brief.
- Proposed work.
- The change we recommend, the page it belongs on and its priority.
The Narrella Standard
We judge an account on four qualities at once: whether it appears in the questions buyers ask, whether it is right in the detail, whether it reflects the company as it stands today, and whether it uses the company’s own wording. Work that improves one quality at the expense of another is unfinished.