GA4 tracks website sessions. It doesn’t track whether an AI answer names your business. For your WordPress site, keep mentions, source links, clicks, sessions, and conversions in separate fields.
Here’s what to record:
- Mentions and citations: Save each answer, business name, linked URL, prompt, platform, and test date.
- Traffic and conversions: Check GA4 sources, landing pages, and tracked actions. Leave missing data marked Unknown.
- Mention and citation rates: Compare repeated tests using the same prompts and settings. Keep these results separate from traffic and sales.
LovedByAI audits WordPress AI readiness and reports AI mentions and referral traffic. Pair those reports with your saved answers and GA4 records. A mention or citation alone doesn’t prove a visit or sale.
Keep Separate Records for Mentions and Citations
Once you’ve defined mentions and citations, track them in separate fields in a repeatable log. This keeps each AI answer traceable and makes prompt runs easier to compare across platforms.
Use one row per prompt execution, with its own Observation ID. Record the platform, exact prompt, timestamp, time zone, and tested location. Give every repeat a new row, and never overwrite an earlier answer. Each row records what happened at that moment, rather than a lasting result.
Use the Same Test Prompts Across AI Platforms
Run a fixed set of prompts across ChatGPT, Gemini, Perplexity, and Claude. Keep the wording unchanged. Record any platform-specific requirements, such as location settings or search mode, and don’t rewrite a prompt because you dislike the response.
Use this test template:
Illustrative test timestamp: October 15, 2026, at 10:00 a.m. EDT (UTC−04:00). This is not an actual observation; results describe only the tested sample.
Prompt: “Which plumbers serve Austin, Texas?”
Platforms: ChatGPT, Gemini, Perplexity, and Claude.
- Account state: Signed out, signed in/free, or paid.
- Model or mode: Displayed model; web/search enabled; unavailable details marked Unknown.
- Search settings: Browsing, personalization, location permissions, safe-search, and language, where available.
- Location: Austin, Texas, supplied in the prompt; platform-inferred location tracked separately.
Use the same template for every run so you can compare later traffic checks without mixing test conditions.
Log Business Names, Source Links, and Supporting Records
In the main row, record the exact business name shown, whether it appears, and its position in the answer. Use a position number only when the answer uses an ordered list. Otherwise, describe its placement.
Keep Mentioned, Cited, Clicked, Session Recorded, and Converted as separate fields. Use Yes, No, or Unknown so you don’t force uncertain outcomes into a yes-or-no answer.
Give each source link its own citation row tied to the Observation ID. Record the citation order, displayed title, cited domain, exact WordPress URL where applicable, and whether the link is clickable. Label the source type as a first-party website, directory, social profile, other third-party source, or unknown. Note redirects and whether the destination supports the claim.
Keep one observation row per prompt run, with as many linked citation rows as needed. Save the answer text, source lists, screenshots, reviewer, review date, and evidence status. Use Unknown for outcomes you haven’t verified. Keep GA4 data outside the observation row unless you can map it to that exact test.
Optimize your WordPress site for AI search
Find out what ChatGPT, Gemini, and Perplexity can and can't read on your site, and exactly what to fix.
Check AI Referral Traffic in GA4
After logging mentions and citations, follow WordPress AI SEO guides to check GA4 for sessions on the cited pages. GA4 shows traffic data. It doesn’t track mentions.
Open Reports → Acquisition → Traffic acquisition and review Session source / medium rather than First user source[12]. Track Sessions, Engaged sessions, and Key events separately[8].
Use the same date range in your site analytics, and account for consent settings, bot filtering, and event rules.
Compare Cited Pages with Session Landing Pages
Match cited URLs to landing pages to check for visits.
Open Landing page, add Session source / medium, and export Landing page + query string. Then normalize paths, check redirects, and match the exported pages to your cited URLs[9]. Keep the original URLs and campaign parameters in your log.
A matching landing page and an identifiable AI source support referral evidence. They don’t prove that a specific answer caused the visit[7].
| Signal | Evidence source | Attribution limit |
|---|---|---|
| Website citation | Saved AI answer and exact linked URL | Shows a link, not a click |
| Session tied to cited page | GA4 session source and landing page | Does not identify the specific answer responsible |
| Engaged session | GA4 engagement metrics | Shows qualifying activity, but may not be a lead |
| Key event | GA4 event reports and business records | Depends on event setup and implementation |
| Direct | GA4 source value | No usable referral or campaign data |
| Unassigned | GA4 channel-group report | Channel rules did not classify the session |
Referral, campaign, consent, bot, and browser rules can cause traffic to appear as Referral, Direct, or Unassigned. Direct and Unassigned don’t prove AI origin[6][10].
GA4 doesn’t automatically measure clicks inside AI answers[7]. Because recording has limits, a missing session doesn’t establish that nobody clicked[11].
Calculate Mention Rates and Citation Rates
After comparing citations with sessions, calculate how often your business is named or linked. Mention rate measures visibility. Citation rate measures source attribution.
Mention rate = answers containing the business name ÷ total observed answers × 100. Citation rate = answers linking to the business website ÷ total observed answers × 100.Count each prompt execution separately. For each rate, count an answer only once, even if it contains multiple links. Report the platform, prompt set, test period, execution count, and search or location settings with your results.
These rates apply only to prompt results. They aren’t traffic metrics. AI-attributed sessions ÷ observed cited answers is not click-through rate, because visits and tested answers come from different datasets. Report CTR only when you have reliable click counts and matching citation impressions.
For key-event reporting, record the date range, attribution model, lookback window, and duplicate-event controls. Citations alone don’t prove causation.
Use LovedByAI with Your Tracking Log
If you use LovedByAI in WordPress, pair its reports with your mention-and-citation log. The plugin audits AI readiness, creates AI-readable page views, and tracks AI mentions and referral traffic. Use its reports to help maintain your log, not as proof that a citation led to a visit or sale.
Review Mention Reports, Source Links, and Referrals Separately
For each mention, record the platform, date, and exact answer text. Save the destination URL for each source link, too.
Compare reported referrals with WordPress or GA4 sessions using the date and landing page. Keep AI referral traffic separate from human-session data until you have supporting business records.
If you can't connect an answer to a visit, leave the attribution unconfirmed. This holds plugin reports to the same evidence standard you use for GA4.
Compare LovedByAI Pricing, Page Limits, and Support
LovedByAI plans differ in page limits and support; see current plans on the pricing page.
Conclusion: Support Each Claim with Evidence
Once you’ve logged mentions, citations, and GA4 data, use that evidence to support each claim. A business mention shows visibility; a linked website citation shows source attribution. Track them separately. Neither proves a visit or a conversion.
Check Each Step from Name Mention to Conversion
Check attribution in the same order each time: mention → citation → click → GA4 session → conversion. Confirm a click or an attributable GA4 session. Then verify the conversion through analytics, forms, bookings, or CRM, and confirm it with business records.
Mark missing evidence as unknown. If the answer has no source link, it has no citation. Missing referral data doesn’t prove there was no visit, because some AI apps don’t pass a referrer.
Before reporting rate changes, compare repeated runs using the same prompts, platforms, and recording method. Report those changes separately from traffic and sales.
FAQs
Does a directory link count as a citation for my website?
A directory link doesn’t automatically lead to an AI citation. It helps AI models verify your business identity and learn about your business [1]. An AI citation happens when an AI engine links directly to your page as a source in its generated answer [2][3][4].
Track unlinked business name mentions separately from linked citations. A mention isn’t a recommendation, and a citation doesn’t guarantee a customer visit [3].
How often should I retest my AI visibility?
Run the same prompts on AI search systems each month to track visibility trends [1]. After technical or content fixes, rerun the same prompts on your normal monthly schedule and compare the results with your earlier runs.
Recheck your pages quarterly and after major site changes, including updates to business facts, schema, or site templates [1][2].
How can I verify AI-driven leads when referral data is missing?
AI referrals can show up as direct or unassigned traffic. Each month, test your top customer questions in ChatGPT, Gemini, and Perplexity. Track brand mentions and citation links separately, then compare them with traffic changes in Google Analytics 4. Neither mentions nor citations alone confirm customer visits.
Check your server logs for AI user agents such as PerplexityBot or OAI-SearchBot. These entries confirm crawler activity, not customer visits.
