Quattr leads AEO, SEO, and content rankings on G2 Spring 2026. View our G2 badges →
Request demo
Request demo

The Quattr Method

AI visibility optimization: the Quattr Method

Five questions that locate you in AI search and classic search. Eight levers that move them. One measurement substrate that makes both comparable week to week.

This is the map underneath everything Quattr ships for AI visibility optimization, whether our team runs it for you, your own agent applies it, or your team runs it themselves.

First time here

What Quattr is

Quattr is the AI search growth platform that does the work. It warehouses your first-party search data, models your demand daily, and ships the content, internal links, landing pages and technical fixes that earn citations and rankings, through GIGA, its AI agent, then proves the impact in revenue.

What the method is

The method is how those decisions get made. Not a framework for a deck, the actual ordering Quattr uses to decide what to diagnose, what to fix first, and when a change counts as a result. It is published in full because a method you cannot audit is a claim.

Who it is for

  • Search owners under an AI mandate, accountable for citations and rankings, and for saying which one moved.
  • Teams running SEO, AEO and GEO as three programs with three vocabularies and three scoreboards.
  • Anyone connecting an AI assistant to their own search data who needs it to reason in a fixed order rather than improvise one.

What it is not. It is not a maturity model and there is no certification. The rungs are diagnostic states you can be in and out of within the same quarter.

Three ways to run it

The method is the constant. What changes is whose hands it is in, our team, your agent, or your own people. All three read your first-party data through the same Quattr search intelligence infrastructure, so the diagnosis does not change when the hands do.

  1. Run for you

    Quattr's growth concierge team

    Our team runs the method on your program, on top of the Quattr MCP and the search intelligence infrastructure underneath it. You get the diagnosis and the sequenced work rather than a tool to learn.

  2. Run by your agent

    The Quattr plugin

    Your own AI assistant applies the method for you, descending the ladder before it answers, naming the lever and the rung that lever moves. Saved expert workflows, in private beta, inside the client you already use.

  3. Run by your team

    Your in-house team or an agency

    Once you are set up on the Quattr MCP and the infrastructure behind it, your own analysts or your agency run the method themselves. Same rungs, same levers, same evidence standard, applied by the people who already own the program.

Bring your own data to a 45-minute working session → Our AI analyzes your top pages first, so you evaluate Quattr on your site rather than a canned deck. One report, then it's your call.

Know

The five R's, where you stand

Five diagnostic questions in a fixed order. Each is a gate: failing it makes every rung above it unreachable, not merely weaker.

R1 Reachable Can crawlers and AI bots fetch you at all? R2 Retrieved Do you enter the consideration set? R3 Referenced Are you the source answers are built from? specimen page → R4 Represented Does AI describe you the way you would? partially instrumented R5 Rewarded Does any of it turn into traffic, conversions, revenue?

THE SUBSTRATE A taxonomy and intent model expressed in your business language, one data lake stitched at URL and keyword grain, and the weekly reporting spine on top of it. Every number on the ladder comes from here, which is why the ladder can be compared week to week at all.

Diagnosis descends the ladder, the gates are physical: you cannot be cited for an answer you were never crawled for. Work runs in parallel through the levers.

  1. R1

    Reachable

    Can crawlers and AI bots fetch you at all?

    Nothing above this rung is possible if this one fails. A page no crawler fetched cannot be retrieved, referenced, represented or rewarded, the gate is physical, not a matter of degree.

    Measured by Server logs for what was actually fetched and with which status, site crawl for what is reachable and indexable, and the two crossed to separate "could not" from "did not".

    R1 · Reachable, can search engines and AI bots fetch, render, and index your site? →

  2. R2

    Retrieved

    Do you enter the consideration set?

    Being fetchable is not being considered. This rung asks whether you appear in the pool a ranking system or an answer engine draws from at all.

    Measured by Impressions and ranked-keyword coverage on the search side; prompt-level presence in answer-engine results on the AI side. Absence here reads very differently from a bad position.

    R2 · Retrieved, does your site enter the consideration set for the demand that matters? →

  3. R3

    Referenced

    Are you the source answers are built from?

    The rung where AI search stops resembling classical SEO. You can rank well and be cited never, and the two are measured on different surfaces.

    Measured by Citation rate and cited URLs per answer engine, share of voice against tracked rivals, and prompt coverage gaps, kept strictly separate from Google's on-SERP AI Overviews, which is a different surface.

    R3 · Referenced, is your site the source AI answers use? → The specimen rung page →

  4. R4

    Represented

    Does AI describe you the way you would?

    Being cited says nothing about being described correctly. A confident wrong summary of your product is worse than no mention.

    Measured by Sentiment and framing in answer text where you are mentioned, read as a distribution over prompts rather than as a single score with an absolute threshold.

    Partially instrumented Instrumented for mentions and sentiment; the positioning-fidelity half is not fully measured yet. The rung is shown as partial rather than quietly scored as complete.

    R4 · Represented, does AI describe your brand per your positioning? →

  5. R5

    Rewarded

    Does any of it turn into traffic, conversions, revenue?

    The rung the business actually asks about, and the one that cannot be inferred from the four below it.

    Measured by Clicks and sessions joined to your own named goals and revenue, with movement significance-gated before anything is called a gain.

    R5 · Rewarded, does search turn into traffic, conversions, revenue? →

Act

The eight levers, what to do about it

P marks the rung a lever primarily moves; s a secondary one. The grid commits cell by cell, so you can disagree with any single cell, which is why this is a grid and not a paragraph. The capabilities hardest to assemble anywhere else sit on the crossings: crawl logs against citations, organic against AI answers.

Lever R1ReachableR2RetrievedR3ReferencedR4RepresentedR5Rewarded
L1Technical & crawl health Technical & crawl health primarily moves R1 Technical & crawl health secondarily moves R2
L2Demand modeling Demand modeling primarily moves R2 Demand modeling secondarily moves R3 Demand modeling secondarily moves R5
L3Refresh & content quality Refresh & content quality primarily moves R2 Refresh & content quality secondarily moves R3
L4Internal linking & architecture Internal linking & architecture secondarily moves R1 Internal linking & architecture primarily moves R2 Internal linking & architecture secondarily moves R3
L5Net-new content Net-new content primarily moves R2 Net-new content primarily moves R3
L6Authority & off-site Authority & off-site primarily moves R3 Authority & off-site secondarily moves R4
L7AI-answer visibility AI-answer visibility primarily moves R3 AI-answer visibility secondarily moves R4
L8Paid/organic interplay Paid/organic interplay secondarily moves R2 Paid/organic interplay primarily moves R5

Prove

The substrate, how you know

One taxonomy, one data lake, one weekly spine. Every number on the ladder comes from here, which is the only reason two weeks can be compared at all.

A taxonomy and intent model expressed in your business language, one data lake stitched at URL and keyword grain, and the weekly reporting spine on top of it. Every number on the ladder comes from here, which is why the ladder can be compared week to week at all.

The enterprise AI-visibility evidence index →

Sequencing

Fastest proof first: substrate, then an L1 baseline, then the wedge, internal linking and refresh, where the work is already-published pages and the feedback loop is weeks not quarters, then demand-guided content, then AI-answer visibility, with paid interplay and off-site authority running continuously. Prove the levers that need no headcount before scaling the ones that do.

Your assistant, on this method

Install the Quattr plugin and your AI assistant applies this method instead of improvising an order. Ask why citations fell and it descends the ladder, reachability before consideration set, consideration set before citations, rather than pattern-matching a plausible answer. Ask what to do next and it names the lever, the rung that lever moves, and the evidence for both. The plugin carries the method; the Quattr MCP gives it your first-party data to run on.

That is what makes an assistant's answer something you can audit: you can disagree with a rung or a cell, and there is something specific to disagree with.

Where to start

For the buyer

Bring your own data to a 45-minute working session →

Our AI analyzes your top pages first, so you evaluate Quattr on your site rather than a canned deck. One report, then it's your call.

For the practitioner

Connect in about two minutes ↗

Your assistant, your accounts, this method. Included with a Quattr subscription, no per-call metering.

In practice

Every use case diagnoses somewhere on this ladder. These are the moments the rungs turn into.

Five things this method assumes

Each one is a distinction the ladder is built to keep, and each is a place programs lose quarters.

  1. A scoreboard is not a strategy

    The default

    Most AI-visibility tooling reports a number and stops there. Citation rate fell four points. Now what, the crawler, the consideration set, the engines, or the quarter? A dashboard that cannot be decomposed turns every movement into a meeting.

    What the method does

    Every reading belongs to a rung, and every rung names the levers that move it. A drop arrives with a place to be diagnosed rather than a place to be worried about, and the next action is a lever with an owner, not a hypothesis.

  2. Ranking is not being cited

    The default

    Rank and citation get collapsed into one panel because both feel like visibility. They are separate surfaces. So are Google's on-SERP AI Overviews and the answer engines, ChatGPT, Perplexity, Gemini, which are routinely added together into a single number that means nothing.

    What the method does

    The ladder keeps them apart by construction. Retrieved asks whether you enter the consideration set. Referenced asks whether answers are built from you. Each surface carries its own metric and its own name, and neither inherits the other's caveat.

  3. You cannot be cited for an answer you were never crawled for

    The default

    Citation tracking gets bought before anyone has checked whether AI crawlers can fetch the pages meant to be cited. The report comes back thin, reads as a content problem, and a quarter of writing goes at a gate that was closed the whole time.

    What the method does

    The gates are physical, not a matter of degree. Diagnosis descends the ladder, Reachable first, every time, and crosses server logs against site crawl to separate "could not" from "did not". Work then runs in parallel through the levers.

  4. Doing everything at once is not sequencing

    The default

    The default plan runs every lever in parallel and reports on all of them. Nothing gets attributed, and the levers that need headcount get funded on the same evidence as the ones that do not.

    What the method does

    Fastest proof first: the substrate, a baseline, then the wedge, internal linking and refresh, where the pages already exist and the loop closes in weeks rather than quarters. Levers that need no headcount prove themselves before the expensive ones are scaled.

  5. A number without a gate is not a result

    The default

    Week-over-week movement gets read as cause. A good week becomes a case study, a bad week becomes a fire drill, and both were inside the noise.

    What the method does

    Movement is significance-gated before anything is called a gain. Every card carries its scope, its freshness date and a link back to the source it came from, and when a source is missing, the answer says so and names it instead of estimating.

Questions this page anticipates

What is the Quattr Method?
A diagnostic ladder of five rungs, eight levers that move those rungs, and the measurement substrate underneath both. It is the order Quattr uses to decide what to fix first and when a change counts as a result. The method stands on its own: run it with our growth concierge team, have your own agent apply it through the Quattr plugin, or run it in-house once you are set up on the Quattr MCP.
Is this SEO, AEO or GEO?
All of them, measured on one ladder rather than three vocabularies. AI search is the umbrella term; answer engine optimization, generative engine optimization and agentic search optimization (ASO) are the category co-terms. The rungs do not change when the acronym does.
Do I need Quattr to use the method?
No. The ladder and the levers are published here in full and shared with the Quattr Academy curriculum. What Quattr adds is the substrate that makes the rungs measurable week to week, and the work that moves them.
Why is Represented marked partially instrumented?
Mentions and sentiment are measured. The positioning-fidelity half, whether an answer describes you the way you would describe yourself, is not fully instrumented yet. The rung is shown as partial rather than quietly scored as complete.

Method structure ratified 2026-07-08 · shared with the Quattr Academy curriculum.