06 · Measure · AI visibility monitoring

Measure what changed after you acted.

Powered by Performance

The same questions, the same AI platforms and the same sample, read again on a different date, so a before and an after can be compared.

Demo data
LumiHome  ›  06 Measure

AI crawlers

Since last actionLast 7 daysLast 30 daysCustom range

Crawlers since the guide was published

CrawlerVisitsChange
GPTBot86.2k↑ 12.4k
PerplexityBot41.3k↑ 3.1k
ClaudeBot28.9k↑ 6.6k
Googlebot19.7k↓ 1.2k
Bingbot8.2k↑ 0.4k

Crawler visits before and after publishing

184,320↑ 22% after the guide
01ProblemWithout a before, an after means nothing

A score moved, and nobody knows why

AI visibility went from 42 to 51 last month. Which action did that? Without a dated action log, the number has no cause next to it.

Different questions on different days

Ask different questions on a different platform with a different sample, and the two numbers cannot be compared. Most AI tracking is read this way.

A direction without a reference point

Your citations grew. If the competitor pages cited for the same questions grew faster, you fell behind. A change needs a benchmark, not just an arrow.

02What flowgen analyzesSix things Measure reads before and after each action

AI Visibility

How often AI names your brand across the tracked questions, by platform, scored before and after each action.

Mentions

Every time an answer mentions the brand, with or without a link, counted on the same questions each time.

Citation Share

The share of cited sources that are your pages, question by question.

Share of Voice

Your mentions against the competitors named in the same answers: for LumiHome, HomeMate, CleanPro and PureLiving.

Referral

Visits that arrive from AI answers, counted per platform where they can be identified.

Before / After

Each metric paired with the dated action it followed, so every comparison runs against the same baseline.

03Product UIOne action, measured

Before and after, side by side

Every action is logged: a page published, an FAQ added, channel versions sent. The same 24 questions and the same pages are read before and after each one, so a change in AI visibility, mentions, citations or AI referral traffic has a cause next to it.

Demo data

Before

  • AI visibility score42
  • AI mentions128
  • Citations320
  • AI referral visits1,980
Action logOct 04 · Comparison page publishedOct 11 · Product page FAQ updatedOct 18 · New citation found

After

  • AI visibility score64
  • AI mentions247
  • Citations680
  • AI referral visits2,340
Compared with competitors

Know whether you moved past the competition

After each action, Measure places your pages among the competitor pages cited for the same questions, so a score change has a reference point and not just a direction.

Set your baseline with a free GEO diagnosis
04How it worksSame questions, same platforms, same sample, different dates

Read the same thing twice, on two dates

Measure holds everything constant except the date: the tracked questions, the AI platforms and the sample size stay the same. Each action is logged with its date, so the later read can be set next to it. A change that follows an action is a clue, not proof.

Demo data
  • Same questionsThe same 24 tracked questions, every read.
  • Same AI platformsChatGPT, Perplexity, Gemini, Copilot and Claude, every time.
  • Same sampleThe same number of answers per question, so the numbers can be compared.
  • Different datesRead before the first action and again after, with each action dated in between.
64↑ 22 since the baselineAI visibility score · LumiHome and three competitors · five weekly reads
  • LumiHome
  • HomeMate
  • CleanPro
  • PureLiving
  • Oct 04Comparison page published
  • Oct 11FAQ updated
  • Oct 18New citation found

The dashed lines mark the day each action was logged. The reads after them are compared with the baseline read of Sep 27; the actions are the clues next to the change.

Measured around each action

Measure records what you did and when, then reads the same questions, pages and traffic sources afterwards, using observable AI responses and the traffic and lead data you choose to connect. Each result is handed to Optimize.

Contact sales
Step 1

Action log

Every content update, new page and channel version is recorded with its date.

Step 2

Before and after

The same questions and pages are measured before and after each action.

Step 3

AI referral visits

Visits from AI search experiences are counted where they can be identified.

Step 4

Your own data

Connect traffic and lead data to see what follows each action.

05Core capabilitiesEvery chart answers one question: what changed after we acted?

AI crawlers reading your site

Which AI systems fetch your pages, and whether they came back after a page was published or updated. Evidence that AI is reading what you shipped.

Which passages AI quotes

The lines AI actually cites from a page, and the ones it skips. The passage that is not cited yet is the next edit for 03 Create.

Page position after each action

Where each page sits among the pages cited for the same questions, and how far it moved since its last update. The lowest page is the first candidate for 07 Optimize.

From action to first citation

Each action is logged with its date, then followed until an AI crawler reads the page and an answer cites it. No submission step, just evidence of what happened next.

AI referral visits by platform

Visits that arrive from AI answers, counted per platform where they can be identified, and compared with the week before the action.

What each crawler came for

Crawler requests grouped by purpose: training, answering a live question, or indexing. A page fetched to answer a question is the earliest sign that it may be cited.

06ExampleLumiHome, from Sep 27 to Oct 25

Three actions, two reads, one comparison

LumiHome took a baseline read on Sep 27, then logged three actions. On Oct 25 the same 24 questions were read again on the same platforms with the same sample.

Illustrative example
DateEntryWhat happenedAI visibilityMentionsCitationsWhat it tells us
Sep 27ReadBaseline: 24 questions, five platforms, same sample42128320The starting point every later read is compared with.
Oct 04ActionComparison page published: robot vacuum vs cordless———Logged with its date. Nothing is read on the day of an action.
Oct 11ActionProduct page FAQ updated with runtime and filter facts———Second action in the log. Both now sit next to the next read.
Oct 18FoundNew citation found: an answer cites the comparison page———The first sign that AI is reading what was shipped.
Oct 25ReadSame 24 questions, same platforms, same sample64247680Compared with Sep 27. The three actions are clues to what moved it, not proof.

Demo data for a fictional brand. The table shows how Measure reports a comparison, not a promised result.

07MetricsWhat Measure reports

Six metrics, each reported as a before and an after for the same questions, platforms and sample. More exposure can lead to more visits and leads; we measure this flow, we do not promise it.

Demo data
AI Visibility42→64

AI visibility score across the 24 tracked questions.

Mentions128→247

Brand mentions across all answers read.

Citation Share18%→31%

Share of cited sources that are your pages.

Share of Voice21%→29%

Your mentions against the competitors named in the same answers.

Referral1,980→2,340

AI referral visits, counted where the platform can be identified.

Before / AfterSep 27 → Oct 25

Two reads, 28 days apart, with three dated actions between them.

Frequently asked questions

What does Measure compare?

The same questions and the same pages, read before and after each logged action. Connect your own traffic and lead data to see what followed, and set a baseline in 01 Discover before the first action.

Does Measure prove that an action caused the change?

No. A change that follows an action is a clue, not proof. Measure keeps the date of every action next to the numbers and holds the questions, platforms and sample constant, so the clue is as clean as it can be. Other things can move AI answers at the same time.

Can Measure show revenue?

Measure reports AI visibility, mentions, citation share, share of voice and AI referral visits, plus the traffic and lead data you connect. More exposure can lead to more visits and leads; we measure this flow. We do not claim that an action caused revenue.

How often are the questions read again?

On a schedule you set, and after each logged action. The interval and the number of tracked questions depend on your plan.

Is your name in the answer when AI describes your industry?

Start by checking your current GEO status.