Contact RankRift
RankRift does not run a contact form, inbox, or phone line. This page points each kind of question to the page that answers it, and lists what a dispute or methodology challenge needs to contain so it can be settled by a re-run rather than an argument.
Read the correction protocol · Read the methodology
RankRift has no contact inbox, accounts, or email list, and it collects nothing from the manual snapshot. The privacy policy covers the data the site itself handles.
In short
Most questions about RankRift are already answered on a public page. Disputed results follow the correction protocol, methodology questions are answered by the versioned method, scope questions by the about and disclosure pages, and site-data questions by the privacy policy. If you are recording a dispute about a figure published anywhere, the fields below are what make it reproducible.
Who this is for
- Brand and marketing owners evaluating a disputed figure about their company
- SEO and analytics leads with a challenge to a definition, a sampling choice, or a stated limit
- Journalists and analysts checking the provenance a study needs before they cite it
- Readers with questions about the tracker evaluation brief
- Anyone looking for a way to reach RankRift directly
When to use this
- You read the relevant page and it still does not answer your question
- You need to record a disputed figure and want the fields that make it reproducible
- You want to check what a citable study has to publish
- You want to know what RankRift collects about readers
Where each question is answered
Each kind of question has a page that answers it. Measurement questions in particular are answered by the public method rather than by correspondence, because a method anyone can read settles more than a private reply would.
- General questions about the study specification
- A disputed or outdated measured result, or a factual correction
- How measurement, sampling, and variance reporting work
- What RankRift is and what it does not claim
- The tracker evaluation brief
- Privacy and site data
Disputing or correcting a measured result
When a published figure names a brand and is disputed, settle it with a re-run rather than an email debate. The dispute record needs enough information to reproduce both the published figure and the alternative observation.
- The exact prompt wording you used, copied rather than paraphrased
- Which assistant you used, and the model or mode if it was shown to you
- The date, plus the approximate time and region if you have them
- How many separate runs you observed it in, and how many showed the disputed result
- Which published page and figure you are disputing
A single answer from a single session is not a measurement. Two people asking the same question in the same hour can get different brands, so run counts matter more than screenshots.
Methodology questions and challenges
A useful methodology challenge is specific: it names the method version, the disputed definition or sampling choice, and the reasoning behind the objection. A ruler nobody can check is only another confident number.
Resolving a challenge means a recorded decision and, where the objection holds, a dated version bump that says which figures it affects. Quiet edits are not a correction process.
- Read the current method version and changelog first
- See how metric definitions and their failure modes are framed
Citing a study
Before citing a study, verify that it publishes a licence, citation string, method version, prompt set, model identifiers, run dates, sample size, and variance. RankRift publishes a study specification rather than a dataset, so cite the study you are relying on rather than this site.
Tracker evaluation questions
The product page is a vendor-neutral evaluation brief; RankRift does not sell, support, or demonstrate a tracker. Questions about a specific tracker belong with its vendor, and no answer is worth sharing passwords, API keys, card details, or confidential prompt sets.
- The tracker evaluation brief
- How to evaluate measurement pricing
- Independence rules for tool coverage
Privacy and site data
RankRift runs no accounts, forms, or email list, and the manual snapshot is run in your own worksheet, so the site holds no snapshot inputs or results about you. The privacy policy covers the data the site itself handles.
What we cannot help with
Nothing on this site can make an assistant say something about a brand. What the site offers is a documented measurement method, its uncertainty, and general source-correction guidance, without an outcome promise.
- We do not accept payment to change, remove, or reorder a measured result, in any form
- We do not give legal advice; defamation, takedowns, and compliance need qualified counsel
- We do not offer placement or traffic services, and we do not partner with anyone who does
- We do not call a page a review without recorded hands-on use
- We do not help with tactics designed to mislead an assistant or a review platform
General information here is not legal advice. For anything with legal consequences, talk to a qualified lawyer in your jurisdiction.
Related pages
- Corrections and disputes
- Methodology
- Editorial policy and independence
- Disclosure and conflicts
- Privacy
- Terms of use
Frequently asked questions
How do I reach RankRift directly?
You cannot: RankRift does not operate a contact form, inbox, or phone line, so none is listed here. Every route on this page leads to a public page instead: the correction protocol for disputes, the methodology for challenges, and the privacy policy for site data.
My brand appears in a study that follows this method and I want it removed. What should happen?
Removal on preference alone is not a correction. A sound process distinguishes factual error, staleness, legal rights, and dislike of an accurate result, and settles a disputed figure with a re-run record that carries its dates, model identifiers, sample size, variance, and outcome.
Can I pay to change how my brand appears in a study?
No. There is no price, sponsorship, or advertising arrangement that changes a measured result, an ordering, or an inclusion decision under this method. This is the hardest line the site has, because the moment it bends, every number measured under the method becomes worthless. If someone offers you this elsewhere, treat it as information about their data.
I think the method is wrong. What then?
Then the method is doing its job. Publishing it is what makes the numbers checkable, so a specific objection to a metric definition, a sampling choice, a prompt set, or a stated limit is exactly what it is exposed for. Pin the objection to the version and section you are challenging and test it with a re-run; an objection that holds should become a dated method change with a changelog entry that says what moved.
Can you tell me whether my agency's AI visibility report is trustworthy?
We cannot audit a specific vendor's report for you, and this publication does not operate a competing tracker. What we can give you are the questions that expose a weak report on your own: which prompts, how many runs, which model versions, on what dates, and what the variance was. If those answers are missing, the number is not checkable regardless of who produced it.
Settle a dispute with a re-run
Record the prompt, the assistant or model identifier, the date, the run count, and the alleged error, then follow the correction protocol.