Resources and working guides
An index of RankRift protocols, worksheets, checklists, and specifications. Each entry states its purpose, audience, when to use it, and its format.
Use the manual visibility snapshot · Browse the published methodology
Every item below is an on-page guide, worksheet, or reference. None requires a download, an account, or an email address.
In short
This page indexes manual measurement instructions, a sample-size calculator, prompt-set design rules, a scorecard layout and reporting outline, crawler and readiness checks, referral-detection guidance, a competitor worksheet, an incident playbook, and a research-data specification. Each is usable without a subscription or an account.
Who this is for
- In-house SEO leads assembling a measurement programme from scratch with no tooling budget
- Agency strategists who need reusable artifacts they can run across several clients
- Analytics and RevOps owners instrumenting AI referral traffic and agent crawls
- Technical SEOs and engineers auditing crawler access, rendering, and structure
- Founders and marketing leads who need one defensible artifact for a specific meeting
When to use this
- You know what you want to do and need the artifact rather than the article
- You are standardising a process across clients or across a team
- You have a report, audit, or teardown due and need a starting structure
- You want the specification a citable study must meet
Measurement tools
These two pages help you produce numbers from observations you record yourself. Neither submits prompts or calls a model on your behalf.
- Manual visibility snapshot — a spreadsheet-friendly protocol for choosing buyer prompts, sampling authorised interfaces, recording raw answers, and calculating a range. Format: complete on-page procedure; no automated result or email gate.
- Sample-size calculator — an interactive calculator for sizing repeated runs, with its formula, inputs, and assumptions shown beside the result. Format: on-page calculator that runs in your browser.
Both keep uncertainty visible rather than producing a bare figure. If a tool in this category shows you a number with no sample size, that is worth questioning.
Prompt sets and competitive research
A measurement is only as honest as the questions behind it. These assets standardise the input so results are comparable between runs, between clients, and between people arguing about them.
- Prompt-set patterns — buyer-stage prompt patterns for four buying-pattern families (SaaS, DTC and ecommerce, B2B services, local services), with the design rules that keep a set honest. For in-house leads and agencies who do not want to invent sixty prompts from a blank page; use it at the start of a measurement programme and re-version your set whenever your category framing changes. Format: on-page patterns and rules.
- Competitor teardown worksheet — a step-by-step method for running a category teardown in an afternoon: which prompts to run, what to record per answer, how to log the sources cited, and how to summarise who holds the default recommendation. For agency strategists and in-house leads preparing a competitive review; use it once per quarter rather than continuously. Format: on-page method and run-log columns in the linked article.
The design rules ban brand baiting. A prompt that names your brand measures the model's willingness to discuss a name you supplied, not whether buyers would find you.
Scorecards, checklists, and reference tables
These are the artifacts people forward to colleagues. They are deliberately low-tech, because a spreadsheet that someone actually maintains beats a dashboard nobody opens.
- Spreadsheet scorecard layout — a two-tab sheet with one row per run (prompt-set version, model version, run time, raw answer, brands mentioned, sources cited) and a formulas-only metrics tab for mention rate, share of voice, and citation share. For anyone running the spreadsheet method before buying tooling; use it from your first run onward. Format: on-page column specification in the linked article.
- AI-readiness guide — crawler access and verification, rendering tests, extraction-friendly structure, schema, llms.txt, and log checks, each with how to verify it rather than just an assertion that it matters. For technical SEOs, engineers, and the person who inherited the robots.txt file; use it once as a baseline audit and again after any platform migration. Format: on-page guide.
- AI crawler reference — a table of AI crawler and agent user agents with the owner, stated purpose, whether the fetch is for training or live answering, the documented control mechanism, a source link, and a last-verified date per row. For engineers and analysts deciding what to allow, block, or look for in logs; use it before you change any crawl directive. Format: dated on-page table.
Reporting and attribution
The point at which most AI visibility programmes fall over is not measurement, it is the meeting. These assets exist to keep the language defensible when the number reaches someone who controls budget.
- Monthly report outline — six blocks covering metric, method, result and confidence, delta, actions taken, and what cannot be claimed causally, so overclaiming is hard to do by accident. For in-house leads and agencies presenting to executives or clients; use it from your second measurement cycle, once you have a delta to report. Format: on-page reporting outline.
- AI referral detection guide — configuration steps for identifying AI-assistant referral traffic in analytics, plus a way to validate it against server logs, with the limits stated plainly including dark traffic and missing referrers. For analytics and RevOps owners; use it before you promise anyone an attribution model. Format: on-page configuration and log-validation guide.
Neither asset will help you claim causality you cannot support. Both are written to make the defensible version of a claim the easiest one to write.
Risk and incident response
Occasionally the problem is not that assistants ignore you but that they say something untrue about you. That is a different job with a different clock.
This playbook is general information, not legal advice. Anything touching defamation, takedowns, or regulatory obligations should go to qualified counsel.
Research-data specification
The research specification requires a qualifying study to publish data with the prompt set, model identifiers, run dates, sample size, variance, prior-run changelog, explicit licence, and citation string.
If a reader cannot re-run a study and disagree with it, the result is marketing rather than evidence. RankRift has not published a study or dataset; the specification is the standard to hold any study to.
- Research study and dataset specification
- The method a conforming dataset uses
- Correction and re-run protocol for a disputed result
Availability and format
Every entry above has an on-page destination, and that page is the complete resource.
Use them in this order: measure, record, present, then handle risk.
Related pages
Frequently asked questions
Are these resources free, and what do you want in exchange?
Yes. Every item above has a usable destination with no email gate, account, or charge.
Which asset should I start with?
If you have never measured anything, start with the manual snapshot, then the scorecard layout to record it in a form you can compare against later. If you already have a number you distrust, start with the prompt-set patterns, because bad prompts are the most common reason a score is wrong. If your problem is a meeting rather than a measurement, start with the report outline.
Can I use these with clients or inside a paid engagement?
Yes. The worksheets, checklists, and outlines can be used commercially, including by agencies, and you do not need to credit RankRift on internal deliverables. RankRift publishes no research dataset; the reusable assets here are methods and layouts rather than measured data.
How often do these get updated?
It varies by asset and by how quickly the underlying facts move. The crawler reference needs the most frequent re-verification because user agents and control mechanisms change without notice, which is why each row carries its own last-verified date. Methods and layouts change when the measurement method changes, which triggers a version bump.
Why are some resources pages rather than downloads?
The useful part is the method, not the file extension. Each linked page contains the full checklist, worksheet, or sequence, so this index treats that page as the resource rather than offering a separate download.
Do I need a paid tracker to use any of this?
No. Every resource on this page works without an owned product, and the spreadsheet method is complete. A tracker becomes worth considering only when the number of brands, categories, models, and reporting cycles makes manual runs unmanageable.
Take one artifact into your next meeting
The manual snapshot procedure and the prompt-set design rules give you the inputs for a dated result you ran yourself.
Use the manual snapshot · Browse the prompt-set patterns · Read the methodology