Measure useful AI search inquiries by connecting identifiable discovery evidence to real requests, qualification, meetings, and customer outcomes. Keep observed referrals, customer-reported influence, and unknown origins separate so your report explains what the evidence supports.

In his guidance for this article, Mike Mangione, founder of The Mangione Group, recommends:

Attribution has limits: analytics can classify traffic as direct when a clear referral source is missing. We shouldn’t relabel that traffic as AI without evidence.

A useful inquiry is a genuine request from someone whose need, circumstances, and expectations fit what your business can deliver. An AI citation is a different thing: a reference to your business or content in an AI-generated answer. Both can matter, but they answer different questions. A mention can indicate visibility without establishing that anyone contacted you, understood your offer, or became a suitable customer.

In my written guidance for this article, I recommend bringing together search visibility, identifiable referrals, inquiry records, sales outcomes, and what customers say led them to the business. That is a proposed measurement standard, not a claim that a particular client achieved a measured AI-search result. It gives the business a practical way to decide what deserves attention next.

Seven key facts

  • A citation, a visit, an inquiry, and a customer are separate observations.
  • Source evidence should stay attached to the record that produced it.
  • A customer can report AI influence even when the recorded visit came from another source.
  • Unknown origin should remain unknown until evidence supports a narrower description.
  • Qualification needs a definition your marketing and sales teams both understand.
  • A booking click does not establish a completed or attended meeting.
  • A useful report connects its findings to a specific decision and states its limitations.

Mike Mangione, founder of The Mangione Group, brings 18 years of marketing experience to the relationship between discovery and business outcomes. For this measurement problem, his recommendation connects visibility with conversations and sales decisions while preserving the gaps in the evidence.

What business outcome should you measure first?

Start with the outcome that makes additional discovery useful to your business: suitable inquiries that progress toward work you can deliver profitably. Define the inquiry, qualification, meeting, and customer stages before comparing sources. Otherwise, one team may count every form submission while another counts only conversations with a realistic buying need.

Write the definition in ordinary language. For a hypothetical regional consulting firm, a suitable inquiry might require a relevant project, an organization the firm can serve, and a person willing to discuss the work. That does not mean the prospect must already have every detail settled. Someone researching on behalf of the decision-maker can still be valuable. Someone requesting an unavailable service is a different situation, however enthusiastic the message sounds.

Then decide who can confirm each stage. Marketing can often verify that a request arrived and identify the page involved. Sales may need to confirm the need and decision process. Operations may need to confirm capacity or delivery fit. Finance may be the dependable source for received revenue and delivery costs. Asking one dashboard to supply all those judgments encourages guesses at precisely the point where the report should become more useful.

Keep the definitions narrow enough to apply consistently. If qualified means something different for every salesperson, the source comparison will reflect those differences. Record the reason when a request is unsuitable, such as outside service area, incompatible scope, duplicate request, or no genuine commercial inquiry. A short explanation is more actionable than a single rejected label.

The goal is not to disqualify people quickly. It is to understand whether visibility introduces the business to people it can help. Our approach to connecting marketing with business decisions starts with that question. A digital marketing agency should be able to explain why its selected measurements matter to your actual offer.

Which evidence belongs in separate measurement layers?

Keep discovery, visits, inquiries, and commercial outcomes in separate layers, then connect individual records where a dependable link exists. This prevents a visibility trend from becoming an unsupported sales claim. The layers can inform one another without having matching totals, identical reporting periods, or a complete record of every step a buyer took.

What each measurement layer can establish
LayerUseful evidenceWhat remains unproven
DiscoveryDated answer capture or reported search impressionThat the exposed person visited or bought
VisitRecorded source, landing page, and sessionThat the visitor was a suitable buyer
InquiryReceived request and contact recordThat the need fits or the meeting happened
Qualified conversationConfirmed need, fit, and next stepThat the opportunity will close
Customer outcomeAccepted work and verified financial recordThat one channel caused the entire outcome

For discovery, maintain a small, repeatable set of questions that resemble real buying situations. Record the exact wording, engine, date, location or account context when relevant, and whether the answer named the business or cited a page. An answer observed once is an observation of that answer, not a representative estimate of every buyer's experience. Keep those captures separate from the numbers a platform reports about actual appearances.

For visits, preserve the source and landing-page evidence rather than only the channel label. Google's traffic-source documentation distinguishes dimensions used for acquisition and attribution. Choose the scope that matches your question. A first-user source answers how the user was first acquired in the available data; a session source describes the session. Switching between them can change the story even when the underlying activity has not changed.

For inquiries and outcomes, use identifiers that can be reconciled with the business record. Count a person or opportunity according to a written rule, not once for every form, email, and call. If a buyer submits twice because the first confirmation was unclear, the business has one buying situation and a possible experience problem. Three tracked actions do not automatically create three leads.

This separation also makes missing data useful. You might have strong evidence that relevant pages appear in answers but weak evidence about what happens after a booking handoff. That is a measurement repair task. It is not evidence that the discovery work succeeded or failed commercially. The next investigation should follow the missing connection.

What can current search and analytics reports actually tell you?

Use each platform's reports for the events and populations it explicitly measures, then verify the reporting definition before drawing a conclusion. Search visibility reports describe appearances in search experiences. Analytics reports describe recorded site activity and configured outcomes. Neither replaces the sales record that establishes whether an inquiry became suitable work for your business.

Google announced dedicated Search Generative AI performance reports on June 3, 2026, with a note stating worldwide rollout by August 31, 2026. The announcement describes impressions, pages, countries, devices for Search, and dates. It also says this visibility remains in the overall performance report. That means the dedicated view should not be added to the overall total as if it were a separate audience. See Google's reporting announcement. These are visibility measures, not a list of qualified leads.

Google's ordinary Search results performance report explains clicks, impressions, click-through rate, and position. A page can perform differently across those measures. A business should inspect which pages and questions are involved before concluding that a change in an overall total means buyers have become more or less interested in its offer.

As of the November 13, 2026 documentation check, Google Analytics defines an AI Assistant channel for recognized assistant sources, while excluding Google's AI Overviews and AI Mode. Those Google experiences are included in Organic Search. The default-channel documentation also distinguishes first-user, session, and event scope. Keep the selected scope and grouping rules attached to any comparison, because a classification change can alter the report without changing the underlying inquiries.

Google also provides a lead acquisition report. Its availability does not make your qualification process automatic. The events, identifiers, and business definitions still need to be configured correctly. A platform can summarize a qualify_lead event without independently determining that the underlying prospect meets your service criteria.

Google's AI features guidance describes the continuing relevance of standard search fundamentals. Measurement should therefore connect to useful page improvements, not become an excuse to chase a proprietary visibility score. For the complementary content question, read our guide to making service-business expertise easier to verify.

A woman writes in a notebook while speaking with a visitor across a reception counter.

How do you investigate a gap between search visibility, recorded visits, and actual inquiries?

Compare the events each system measures before trying to reconcile its total. Search impressions, clicks, recorded sessions, and received inquiries are different observations. Establish the reporting period, URL, source definition, and collection conditions for each, then trace a controlled contact journey. A difference between totals is a question to investigate, not proof that one system lost customers.

On November 12, 2026, Google added documentation explaining how website owners can identify Google Search traffic in analytics and server logs. The official update log confirms that date. This is newly published guidance, rather than a claim that every visitor’s browser or Google’s measurement behavior changed that day.

Google explains that Search generally sends an origin-only HTTP Referer value, without the referring path or search query. It also documents situations in which browsers omit that header. This gives owners a useful diagnostic starting point while limiting what a referral observation can establish.

Which observation supports which conclusion?

Use this matrix as an evidence boundary. It is an editorial review method, not an attribution model. Each row separates an observation from the stronger commercial claim that still requires evidence.

Keep visibility, arrival, and business outcomes separate.
ObservationSupported descriptionStill unverified
AI impressions in Search ConsoleLinks appeared in supported Google AI features.A visit, inquiry, or customer.
Google origin in a landing requestA Google-origin referral signal arrived.The specific search experience.
Recorded AI Assistant sessionAnalytics applied its assistant-channel rules.A matched inquiry and commercial outcome.
Direct session and customer testimonyOrigin is unclear; the customer reports AI influence.A confirmed assistant referral click.
Matched qualified inquirySource evidence connects to a suitable request.A sale or incremental impact.

Because the documented Google referrer values identify an origin, that value alone does not establish which search result format the person used. Preserve that boundary instead of assigning every Google visit to AI Mode or an AI Overview. Google’s channel definitions also separate recognized assistant visits from Google’s AI search experiences, which belong within Organic Search.

What should you check before calling the gap a tracking failure?

First align the population and reporting period. Are you comparing one page or the whole property? Are you using a canonical URL in one report and a redirected address in another? Do the dates use the same time zone? Google’s generative AI report documentation says page data is generally assigned to canonical URLs and dates use Pacific Time. A CRM operating in Eastern Time can place an event near midnight on a different calendar day.

Then inspect what collection actually recorded. Check the contact form, booking integration, analytics configuration, and receiving record separately. A working component does not establish that the connections between components work. Record any difference between the expected journey and the observed result, together with the conditions under which it occurred.

Use a short investigation record: the conflicting totals, their definitions, the affected period, the suspected connection, and the person responsible for checking it. Avoid rewriting historical source fields while the cause remains uncertain. Preserving the original observation makes a later correction easier to explain.

How can a controlled journey make the investigation useful?

Write down the expected path before testing: arrival, contact request, receiving record, owner assignment, and qualification. Use a clearly labeled test inquiry through a route the business controls. Record device, browser, consent selection, redirects, and any separate booking domain. Test the supported experience without overriding privacy choices to make a dashboard look complete.

Compare what each system recorded. Did analytics collect the intended event? Did the contact system receive the inquiry once? Did available source information remain attached to that record? Did the responsible person receive enough context to act? A confirmation screen answers only part of that chain.

Keep the conclusion bounded. A controlled test can establish that a particular journey worked under the tested conditions. It cannot reconstruct every historical inquiry or prove that AI caused a sale. If the test reveals a missing handoff, repair it and repeat the affected check. Document collection limits instead of inventing missing history.

What should change in the business report afterward?

Add a measurement note beside the result: what was checked, what changed, when, and which comparisons are affected. Keep unresolved origins visible. Google’s direct-traffic guidance defines direct traffic as lacking a clear referral source. It should not become an allowance for otherwise unobserved AI activity.

Finish with the decision. A repaired form may justify resuming a broken contact path. A source-label correction may justify restating a channel report. More suitable inquiries may justify reviewing capacity or testing investment. These conclusions require different evidence. The investigation is useful when it produces a defensible next action and states what remains uncertain.

How do you connect a visit to an inquiry without inventing a journey?

Connect a visit to an inquiry only through a documented record match, and keep the original evidence alongside the resulting classification. Establish a stable inquiry identifier, record the request received, and preserve available source information. When the buyer moves to a booking service or another device, acknowledge any break instead of filling it with inference.

Start by testing the actual contact path. Submit an authorized test request with an unmistakable test label, confirm that it reaches the correct destination, and check whether the recorded source survives. Test the mobile journey as well as the desktop journey. A visually successful confirmation is only one part of the check; the receiving team needs the request and enough context to act on it.

Google defines generate_lead as a recommended event for a submitted form or request for information in its recommended events guidance. That is a useful starting vocabulary. It does not mean every detected button click should trigger the event. If an invalid form submission produces the same event as a received request, the report counts attempts as leads.

For a separate booking domain, review the supported integration rather than assuming the journey remains connected. Google's cross-domain measurement guidance explains how unified measurement can work across domains. A third-party system may limit what you can configure. In that situation, reconcile its confirmed bookings with the inquiry record and state the limits of the analytics connection.

Use DebugView or the appropriate platform testing tools to inspect event behavior, but also check the business destination. An event visible in a debugging screen does not establish that a human received the request. Keep test activity out of commercial results and document the exclusion. Repeat the path check after meaningful form, booking, or tracking changes.

Identity across devices deserves particular care. Google's reporting identity documentation describes User-ID, device identifiers, and modeling. A business should understand which identity methods its property uses. Do not interpret a modeled or aggregated journey as a directly observed personal history. For many service businesses, a transparent partial record is more useful than a complicated attribution diagram nobody can explain.

How should you handle direct traffic and customer-reported AI influence?

Keep technical source evidence and customer-reported influence in separate fields, and allow either field to be unknown. A customer saying an AI answer helped them is useful testimony about their decision. It does not justify rewriting a direct session as a confirmed AI referral, nor does a missing referral disprove the customer's account.

We shouldn't relabel that traffic as AI without evidence.

That is my written recommendation for this article. The practical distinction is between an observed route and a reported influence. A buyer might read an answer, remember the company name, and later type it into a search engine. Another might follow a visible link immediately. Both situations could involve AI discovery, but the records support different descriptions.

Ask an open question first: what led you to contact us? If the answer is broad, ask what they remember seeing and whether something helped them decide. Do not force every respondent to choose from a list that makes AI sound like the preferred answer. Keep their actual explanation, with an internal summary if useful. Someone saying they used online research has not identified a particular engine.

A hypothetical record might contain session source: direct; self-reported discovery: ChatGPT; remembered page: service overview; confidence: customer's account only. Another might contain session source: identifiable assistant referral; self-report: not asked. The first is not a confirmed click from ChatGPT, and the second does not tell you whether the answer was the first or decisive influence.

Use an unknown category that is visible in the report. If unknown grows, investigate collection or process changes before treating it as a channel trend. Perhaps the team stopped asking the discovery question, a booking integration changed, or the inquiry arrived by telephone. A missing answer should not be converted into an unfavorable source judgment either.

The same principle applies to assisted outcomes. You may report that an opportunity includes customer-reported AI influence, while preserving its recorded acquisition source. Those are overlapping descriptions of the same opportunity. Do not add them together as two opportunities. The report should help a reader understand the relationship without inflating totals.

How do you tell whether the inquiries are commercially useful?

Review the request, the buyer's understanding, and the next verified outcome against a consistent suitability definition. Useful inquiries concern work the business can provide and create a plausible next conversation. Quality should remain separate from source certainty: a strong prospect with unknown origin is still a strong prospect, and a confirmed AI referral can still be unsuitable.

Begin with service fit. Does the person need something you offer? Are the location, timing, scope, and expected commitment compatible with delivery? Distinguish a genuine incompatibility from a detail that simply needs clarification. If the person has misunderstood the offer, record the misunderstanding rather than labeling the entire channel low quality.

Next examine meeting outcomes. A booked meeting can be canceled, rescheduled, unattended, or completed. Those states should not collapse into one conversion total. A rescheduled appointment should remain linked to the same opportunity. If a prospect attends but discovers the service is different from what they expected, that is valuable evidence about the discovery and explanation path.

Review sales outcomes after enough time has passed for the relevant buying process. An open opportunity is not a loss simply because it has not closed by the reporting date. Keep created date, current stage, and outcome date distinct. Compare mature cohorts when evaluating customer conversion, and describe newer groups as still developing.

For profitability, use the financial measure the business can actually substantiate. Contract value, invoiced revenue, received revenue, and contribution after delivery costs answer different questions. This article does not prescribe a universal margin target. It recommends making the selected definition visible and involving the person who owns the financial record before calling a source profitable.

Finally, sample the language of the inquiry itself. Do buyers describe the right service? Do they expect an unsupported guarantee? Are they asking sensible questions the website could answer more clearly? Our copywriting work can address a repeated explanation gap. The source report becomes more useful when it shows what the incoming conversation reveals about the message.

A hand sketches arrows connecting a magnifying glass, people, and envelopes on paper.

How can you use an evidence worksheet to classify one inquiry?

Use the worksheet to record the strongest available source evidence and the current commercial stage without combining them into a performance score. Its output identifies the description the selected facts support and the next missing check. It is a local planning aid, not an attribution model, a lead score, or a prediction of sales.

Choose evidence and a stage, then review the inquiry.

The rule is deliberately conservative. Unknown source stays unknown. Customer testimony is labeled reported influence. A referral is labeled an observed referral only when it is attached to this inquiry. Without the record match, the tool tells you to reconcile the records first. No selection turns source evidence into proof that the source caused a purchase.

Consider a hypothetical suitable prospect who says an assistant recommended the business, but whose recorded session was direct. Select reported influence and suitable inquiry. If the testimony belongs to the inquiry being reviewed, mark the record match. The output preserves the testimony without claiming a referral click. This is a more accurate basis for a conversation about discovery than assigning the opportunity entirely to AI.

Now consider a hypothetical tracked referral whose form request is outside the service area. Select identifiable referral and unsuitable. The correct response is to examine the service-area explanation or the discovery context, not count the request as a qualified success. The tool intentionally preserves the unfavorable commercial outcome even when source identification is strong.

These inputs stay in the page and are not submitted by this worksheet. Avoid putting personal details into exploratory tools. Maintain actual inquiry evidence in the business systems authorized for that purpose. If your team needs additional report fields, Google's custom dimensions and metrics documentation explains their reporting role, but collecting a field should follow a defined question, not curiosity alone.

How do you compare performance without misleading yourself?

Compare like definitions, comparable periods, and sufficiently developed inquiry groups, while showing the underlying counts and unknowns. Separate a change in measurement from a change in buyer behavior. A higher rate from a very small group may be worth investigating, but it should not carry more certainty than the evidence can support.

Suppose, in a hypothetical example, one period contains 12 identified AI-referral inquiries and six suitable inquiries. A later period contains 20 and eight. The suitability rates are 50% and 40%, respectively, while the count of suitable inquiries rises from six to eight. Neither rate alone answers whether the business is better off. The sales outcomes, delivery fit, and cost of generating and handling those inquiries matter too.

Now suppose the later period also includes five inquiries awaiting review. Calling all five unsuitable makes the later rate prematurely pessimistic. Removing them without disclosure can make it optimistic. Show reviewed outcomes and pending status separately, then explain the denominator. The arithmetic is simple; deciding what belongs in the arithmetic is the essential work.

Preserve the source rules used in each period. If a newly recognized assistant domain moves from referral into AI Assistant, a channel increase may partly reflect classification. If a new form starts collecting a discovery question, reported AI influence may increase because the business finally asked. Annotate those changes in the same report as the trend.

Be equally careful about causation. A page update followed by more suitable inquiries is a reason to investigate, not proof that the update caused the increase. Other campaigns, seasonality, sales response, offer changes, and capacity can affect the outcome. Where a controlled comparison is practical, design it before making the change. Where it is not practical, describe the observation and competing explanations plainly.

A digital marketing company should provide enough detail for you to distinguish an operational improvement from a reporting improvement. Both can be valuable. They simply support different decisions about future investment.

What should you change when the evidence shows a weak point?

Choose the next change at the weakest demonstrated connection: discovery, explanation, inquiry capture, qualification, or sales follow-through. Use the records to identify a specific failure before changing several parts of the journey. The purpose of measurement is to make a defensible decision about the work, not to produce a more impressive chart.

If relevant pages are absent from observed answers, examine discoverability, accessible content, and whether the pages answer the buyer's actual question. If the business appears but is described inaccurately, inspect the underlying pages and sources for ambiguity. The remedy may be a clearer service explanation rather than publishing more pages on adjacent topics.

If identifiable visitors arrive but do not inquire, inspect the landing page and contact path. A lack of inquiries alone does not tell you whether visitors were unsuitable, the offer was unclear, or the form failed. Combine page evidence with actual path testing and buyer feedback. Avoid assigning a motive to anonymous visitors that the records cannot establish.

If inquiries arrive but the need is routinely incompatible, review what those people were led to expect. A page might accurately rank for a broad topic while failing to identify the service boundary. Explain who the service fits and what it requires. Making those boundaries clearer can reduce total inquiries while improving the usefulness of the conversations.

If suitable inquiries stall after contact, examine ownership, response content, meeting purpose, and commitments. An acquisition channel cannot compensate for a request that sits unanswered. Keep the source performance report connected to sales operations so that the team fixes the actual interruption.

Our Google and AI visibility service belongs in this connected view. Discovery work is easier to evaluate when the business has already decided what a useful next step looks like and can recognize it when it happens.

Before assigning budget, write the decision in a sentence that someone outside marketing can challenge. For example: we will clarify the service boundary on the page that produced these unsuitable requests, then review the next comparable inquiry group. That sentence names the evidence, action, and check. By comparison, increase AI visibility leaves the team without a clear account of which business problem the additional visibility should solve.

Also distinguish a learning budget from a proven acquisition investment. A business may reasonably spend effort exploring an emerging discovery route while its commercial evidence is incomplete. Describe that work as an experiment with a defined question and limit. Do not present exploration as demonstrated return, and do not reject it solely because the first report contains unknowns. The appropriate decision depends on the value of resolving the uncertainty and the resources the business can responsibly commit.

Keep an explicit stopping condition for each investigation. If a repeated test shows the booking confirmation fails, stop analyzing channel quality until the path is repaired. If inquiries are still awaiting review, assign that work before comparing qualification rates. If the proposed page change answers a concern nobody has expressed and no evidence reveals, return to the buyer question. These conditions protect the report from becoming a sequence of confident recommendations disconnected from what was observed.

A woman reads an open folder at a desk beside a telephone.

What should a practical monthly review contain?

A practical review should show the evidence collected, the commercial stages reached, the unknowns, and the next action with an owner. Include a few reconciled inquiry examples alongside aggregate counts. That combination helps the team understand both the scale of the activity and the specific reasons buyers progressed, misunderstood the offer, or stopped.

Begin with measurement health. Note broken integrations, changes to source rules, missing inquiry fields, and any backlog in qualification. Then show discovery and visit observations without presenting them as sales. Follow with inquiry outcomes by clearly named source-evidence category. Keep customer-reported influence visible as an overlapping field rather than an additional pile of leads.

Review exceptions as well as the dominant pattern. An excellent inquiry from an unexpected page might reveal a useful buying question. A stream of poor-fit requests from one page might reveal a misleading service explanation. A successful customer whose origin remains unknown should remain in the business total even if the attribution report cannot assign credit.

End with one or two concrete actions appropriate to the evidence. Name the record that prompted the action, who will own it, what will change, and when the team will check the result. If the next action is simply to repair a broken booking confirmation, say that. A precise small repair can be more useful than another broad visibility initiative.

Keep previous definitions and review notes so the next reader can understand a historical result. A report should remain interpretable after staff, platforms, and dashboards change. The most dependable measurement habit is a record of what was observed, what was inferred, and what the business decided to do with that distinction.

What else should owners know about AI search inquiries?

The most useful answers keep discovery evidence separate from commercial evidence and make uncertainty visible. Owners do not need perfect attribution to make a better decision. They need consistent definitions, a functioning contact path, and enough reconciled examples to understand whether the right people are arriving with an accurate view of the service.

Does an AI mention count as a lead?

No. A mention establishes that the business appeared in the observed answer or reported visibility context. A lead requires an actual inquiry under your stated definition. Keep mentions useful as discovery evidence, but connect them to received requests and commercial outcomes before describing them as evidence of customer acquisition.

Should I ask every prospect how they found us?

Ask when it fits naturally into the conversation and record the answer without leading the person. An open question can reveal both discovery and decision influences. Allow uncertain or incomplete answers. The aim is useful context, so the question should not make it harder for someone to receive help or take the next step.

Can I attribute all direct traffic growth to AI?

No. Direct traffic growth does not identify the missing source. Keep it separate from observed AI referrals and customer-reported AI influence. Investigate changes in tracking, booking routes, and buyer behavior, but do not replace an unknown with the explanation you prefer. A transparent unknown is more useful than an unsupported success claim.

What if very few inquiries come from identifiable AI referrals?

Inspect the individual records and preserve the small counts. Determine whether they are suitable and whether other buyers report AI influence. Avoid treating a tiny sample as a stable conversion benchmark. Continue checking the discovery and contact path while deciding whether the available evidence justifies more investment, a focused repair, or continued observation.

Should we stop measuring visibility if sales matter most?

No. Visibility helps explain whether relevant content can be discovered and how the business is represented. It becomes more useful when connected to visits, inquiries, and sales evidence. Keep it as an earlier stage with its own interpretation, rather than discarding it or treating it as a substitute for the outcome.

What should I bring to a review of our measurement?

Bring your service description, available search and analytics reports, a few inquiry records, and the known outcomes of those conversations. Include examples with unknown sources and poor fit. Explain recent changes to forms, booking, or qualification. Missing information is part of the diagnosis and does not need to be disguised with estimates.

How can The Mangione Group help make this measurement useful?

The Mangione Group can help connect what buyers discover with what happens when they contact your business. The work starts with the service, the people you can help, and the records you already have. From there, the recommendation should identify which measurement or communication gap is preventing a useful decision.

That may involve clarifying service pages, improving the path to an inquiry, defining qualification, or connecting source evidence with the sales record. It may also mean fixing a simple handoff before spending more on discovery. The aim is a practical view of the buying process that your team can understand and maintain.

A marketing company should be clear about what remains uncertain. We would not turn unidentified traffic into an AI success story or promise that more citations will produce profitable customers. If the service, delivery capacity, or follow-through cannot support additional demand, those conditions need attention too.

A 30-minute Marketing Review is a starting point for examining the evidence and deciding what to improve first. Bring the reports and inquiry examples you have, including the records that do not fit the story.

Michael Mangione

Michael Mangione, founder of The Mangione Group.

Founder, The Mangione Group

Michael Mangione brings 18 years of marketing expertise. He's assembled a leadership team whose track record includes more than $400 million in advertising spend managed, over 600,000 leads generated, and more than $1 billion in client outcomes.

He built The Mangione Group to help businesses compete with greater precision, win more customers, and grow revenue. His approach brings innovative technology and AI together with sharp strategy, persuasive copywriting, and compelling design to make businesses stand out, make their value unmistakable, and give customers a reason to choose them.

About Mike Mangione | Mike Mangione on LinkedIn

References and sources

  1. About traffic-source dimensions. Google. Undated living documentation. Checked November 13, 2026.
  2. Introducing Search Generative AI performance reports in Search Console. Google. June 3, 2026; rollout note August 31, 2026. Checked November 13, 2026.
  3. Performance report (Search results): Overview and basic setup. Google. Undated living documentation. Checked November 13, 2026.
  4. Default channel group. Google. Undated living documentation. Checked November 13, 2026.
  5. Lead acquisition report. Google. Undated living documentation. Checked November 13, 2026.
  6. AI features and your website. Google. December 10, 2025. Checked November 13, 2026.
  7. Recommended events. Google. Undated living documentation. Checked November 13, 2026.
  8. Set up cross-domain measurement. Google. Undated living documentation. Checked November 13, 2026.
  9. Monitor events in DebugView. Google. Undated living documentation. Checked November 13, 2026.
  10. Reporting identity. Google. Undated living documentation. Checked November 13, 2026.
  11. About custom dimensions and metrics. Google. Undated living documentation. Checked November 13, 2026.
  12. Identify Google Search traffic in your web analytics or server logs. Google. Added November 12, 2026. Checked November 13, 2026.
  13. Latest documentation updates. Google. November 12, 2026 entry confirms the new traffic-identification guide. Checked November 13, 2026.
  14. Generative AI performance report. Google. Living documentation. Checked November 13, 2026.
  15. Understand direct traffic. Google. Living documentation. Checked November 13, 2026.