AI-search pipeline up 15%: what CRS tracked with Google Analytics and Looker

CRS reported a 15% increase in AI-search pipeline, a pre-sale measure rather than closed revenue. The team used Google Analytics and Looker to connect referrals with leads and deals.
A report saying “AI cited us” does not explain how much it helped sales. CRS, which provides access to credit information through an API, offers a case for looking further along the customer journey.
The company reportedly increased AI-search-attributed pipeline by 15%. Pipeline refers to potential business before a deal closes. Profound’s case study
Here, we consider how LLM optimization can connect with sales records. The article first explains the published case, then provides a process for checking a company's own GA4 and Looker data. Official documentation was checked on November 11, 2026.
CRS’s visibility, traffic, and pipeline figures measure different outcomes
The case reports 20-fold AI visibility, an 8% increase in weekly LLM referral traffic, and a 15% increase in AI-search-attributed pipeline. These relate to AI recommendations, website visits and potential business before closing. Profound’s case study
Illustration: Separate AI mentions, website visits, and open sales opportunities.
What CRS provides and what it wanted from LLM optimization
CRS provides services that connect credit, identity and fraud-related information to customer workflows. Its official website describes access through products including APIs, the interfaces through which software exchanges information. CRS’s official website
Profound's case includes comments from CRS marketing leader Ronak Patel. The stated objective was to connect AI visibility with business outcomes such as qualified leads, conversion rates and revenue.
Visibility concerns how a business appears in AI answers. The practical question is what happens after that appearance: whether people investigate the service and whether their enquiries progress toward sales.
The introduction to LLMO and its relationship with SEO covers the wider topic. This article focuses on the CRS case and on reading the measurement without confusing one outcome with another.
Visibility, visits and opportunities count different things
Citations, visits, and opportunities describe different stages. Define what counts as one event in a company's reporting first.
| Published measure | Reported change | What to distinguish |
|---|---|---|
| AI-search visibility | 20-fold | Not a multiplier for website visits or revenue |
| Weekly traffic from LLM citations | 8% increase | Different from appearances within AI answers |
| AI-search-attributed pipeline | 15% increase | Potential business, rather than closed revenue |
The source does not specify whether the pipeline figure is a count or monetary value or provide the full aggregation rules. It cannot be presented as a 15% increase in closed deals or revenue. Without baseline totals and comparison periods, it should not be adopted directly as a target for another business.
These are figures from Profound's customer case. AgentSignal has not obtained CRS's analytics and sales records and independently recalculated them. Profound’s case study
In an internal report, place the counting rule and reporting period beside each metric. That makes it easier to identify which stage needs investigation than a list of growth percentages alone.
Using Google Analytics and Looker to connect answers with sales activity
CRS describes using Google Analytics and Looker to connect AI answers with qualified leads and deals. The disclosed change was connecting Profound's analytics to its existing technology environment. Profound’s case study
Illustration: Connect answer observations, analytics, and sales notes while retaining attribution gaps.
What analytics, Looker and sales records help establish
For an internal process, identify which records can actually be connected rather than assuming every individual journey is observable. Analytics may show the landing page, while a customer-management record shows whether an inquiry became an opportunity.
Google Analytics collects and reports activity on websites and apps. Looker can aggregate data made available within an organization and display it in tables and charts. Sales records describe follow-up and progress through the sales process.
Connect the earliest visit and the eventual deal only where evidence supports the relationship. Keep an observed referral and a customer's statement that they discovered the business through AI in separate fields.
The source describes a view extending from AI responses to MQLs and closed deals. An MQL is a lead that marketing considers sufficiently qualified to pass toward sales. CRS's specific qualification criteria and record-matching implementation are not disclosed.
Content distribution and publisher relationships happened alongside the work
Content distribution and relationships with prominent publications were also part of CRS's work. The case concerns where information was distributed as well as how results were measured. Profound’s case study
The public material does not provide a complete record of pages edited, dates or outreach to individual publications. It also does not isolate how much each activity contributed to the reported figures.
For another company's report, record advertising, external coverage, and sales activity occurring during the same period. That helps avoid assigning every change in pipeline to an article edit. This is an editorial recommendation for internal reporting, not a reconstruction of undisclosed CRS activity.
After placing page edits alongside business results, add other relevant changes to the history. Keep the observed movement in a metric separate from an explanation of why it moved.
The public case does not provide a reproducible GA and Looker configuration
The disclosed fact is a connection to an existing environment including Google Analytics and Looker. The case is not a configuration manual for recreating CRS's implementation.
| Item | What the public material establishes |
|---|---|
| Analytics environment | Includes Google Analytics and Looker |
| Google Analytics version | Does not establish that CRS used GA4 |
| Connection to customer records | Does not disclose a product, matching key or process |
| Event names and settings | Does not provide a reproducible configuration |
| Reporting interface | Does not provide the settings needed to recreate it |
The process below uses records available within a reader's own organization. GA4 and Looker operations come from their official help, not from inferred details of CRS's setup.
CRS had created the FAQs; their contribution to the 15% increase is unverified
The case says CRS used Profound's FAQ Agent to create FAQ content for key product and blog pages. That passage appears in the source's “Looking ahead” section. Profound’s case study
Although the section is forward-looking, the FAQ creation is described as completed. The source does not establish the implementation date or its timing relative to the measurement period. It therefore does not show that FAQ production caused the reported 15% increase, or that the FAQs were introduced only after the results.
Illustration: Results and FAQ work are considered separately. The positions and lines do not represent CRS's dates, implementation sequence or causality. FAQ creation was reported as completed.
In an internal report, distinguish completed changes from planned experiments. A planned FAQ update should not become an explanation for a past result, and completed work should not automatically be labeled a proven cause.
To identify a next step, collect recurring pre-sales questions and check whether the website already answers them.
Prepare to connect observable AI visits with inquiries and opportunities
Begin by checking the records between an inquiry and an opportunity. This is our proposed process, not CRS's actual configuration. Separate reading analytics reports from changing measurement settings.
Illustration: Track inquiries and meetings alongside dates, source evidence, and page changes.
Define AI visits, completed inquiries and opportunities in a shared record
Before calculating totals, decide which record counts a visit, an inquiry and an opportunity. Copy this table and fill it with the definitions and storage locations confirmed internally. Blank cells are for completion; they are not CRS's rules or measured results.
| Subject | Definition to establish | Internal rule and location |
|---|---|---|
| AI-related visit | Referrer, channel and visit unit | |
| Inquiry | Completion condition and duplicate handling | |
| Opportunity | Sales qualification criteria | |
| Reporting date | Visit, inquiry or opportunity date | |
| Unmeasured or unknown source | How uncertainty is retained |
A click on a form's submit button does not necessarily mean submission succeeded. Validation errors may prevent completion. An inquiry count should use a record that matches the agreed completion condition.
Mark missing measurement as unmeasured. Instead of inventing a value to complete the table, identify which owner or record can resolve the gap.
Check inquiry storage and transitions to external forms
Check what the inquiry form saves to the customer-management system. Begin by asking the form owner and sales team whether they can locate the same inquiry in their respective records.
- Check what the inquiry form saves to the customer-management system.
- Connect available landing-page and referral information to the inquiry.
- Agree with sales on when an inquiry becomes an opportunity.
- Preserve unknown discovery sources as unknown.
If the form opens on another website, check whether measurement continues across that transition. The GA4 cross-domain guide for external bookings and applications (in Japanese) explains the relevant conditions.
A customer may report discovering the business through AI even when no referrer was captured. Retain that statement separately; it is useful evidence, but it is different from an observed analytics referral.
Define an opportunity and compare consistent time periods
Define the opportunity stage before comparing equal-length periods. An illustrative definition is an inquiry for which an initial sales meeting has been scheduled.
| Field | Purpose |
|---|---|
| Opportunity criteria | Keep counting consistent across staff |
| Inquiry and opportunity dates | The two events may occur in different months |
| Referral and customer statement | Preserve the evidence for AI attribution |
| Page change date | Make before-and-after review possible |
Start with verified counts. If reporting monetary values, separate expected value from closed revenue. The aim is to show how far people progressed after a citation.
An inquiry and the resulting opportunity may fall in different months. Dividing one month's opportunities by that month's visits does not automatically describe the same people's progress. State whether the report follows inquiries from their original date or counts opportunities created during the reporting period.
Keep that date rule consistent in before-and-after comparisons. Record the start of measurement and any definition changes separately from page publication dates.
Check AI-related visits and inquiries in GA4
Use GA4 to inspect recorded visit sources and the action configured as inquiry completion. Report viewing and measurement changes require different permissions; involve the appropriate owner when settings need to change.
Conceptual illustration: checking a referral source and confirming an inquiry are separate tasks. This does not reproduce GA4 screens or CRS’s configuration.
Find AI Assistant and the referrer in Traffic acquisition
Open Google Analytics, choose the site's property and navigate to Reports → Acquisition → Traffic acquisition. A property is the unit that holds the site's measurement data. Google’s report instructions
Choose a period and inspect Session default channel group for AI Assistant. Also examine Session source / medium to see the recorded origin of visits. Keep both the period rules and the dimension consistent when comparing results.
GA4 documents an “AI Assistant” channel for recognized AI referrers. Google AI Overviews and AI Mode are excluded from that channel. Google’s channel definitions
Treat it as the traffic captured by that definition rather than all AI-related visits. If the row is absent, inspect the date range and recorded referrers. If the entire report is absent, ask an administrator to check navigation configuration and access; Google notes that reports can be removed from the default navigation.
Verify the completion event before marking it as a key event
Counting completed inquiries requires an event that represents successful submission. An event records an action sent to GA4; marking it as a key event identifies it as important to the business. Google’s key-event instructions
The measurement owner can enable debugging for a test device using Tag Assistant and inspect events in DebugView during a test submission. Check that the event fires on completion and is not duplicated. Google’s DebugView help
Once an existing completion event has been verified, go to Admin → Data display → Events, find the event and select the star beside its name. Following the permissions note in the official existing-event instructions, an Editor or administrator should perform this step.
Marking an event does not implement missing form tracking. If no event arrives, the responsible team must inspect the form's completion condition and tagging. Record when the setting changed and distinguish earlier reporting periods.
Keep first acquisition, the current visit and sales qualification separate
Traffic-source dimensions describe different scopes. First user source concerns initial acquisition; Session source concerns the visit. Event-scoped reporting also involves how credit for an important action is assigned. Google’s scope explanation
For example, someone might first arrive from an AI link, then return through another channel and submit an inquiry. A difference between the first-user source and the later session source is not automatically a measurement error. Choose the field that answers the reporting question.
An inquiry-completion key event is also different from a sales-qualified opportunity in customer records. Do not rename the GA4 total as an opportunity count without checking the definition and supporting records.
The guide to AI traffic and outcomes in GA4 covers the foundational workflow. For the scope of AI Assistant, use the current official channel definition cited above.
Compare visits, inquiries and opportunities in Looker
Looker can display metrics based on data already connected and prepared within an organization. This section assumes an available Explore and appropriate save permissions. Access to GA4 alone does not establish those prerequisites.
Conceptual illustration: align periods and definitions when reconciling visits, inquiries and opportunities. This does not reproduce Looker screens or CRS’s data.
Confirm the data, Explore access and save permissions
Sign in through the organization's own Looker address and confirm the data available for the report. An Explore is an interface for selecting prepared fields and running a query. Its available fields depend on the organization's data model.
Explain the reporting need to the administrator: compare AI-related visits, completed inquiries and opportunities for a consistent period. Confirm the Explore, metric definitions, date fields, refresh timing and destination folder.
Looker distinguishes access to data, permission to explore it and permission to save dashboards. Editing access to the destination folder is also required. Being able to view a report does not establish permission to save another one. Looker roles
If the data is not connected, prepare the required fields and their source locations first. Do not infer an undisclosed join from the CRS case or substitute Looker Studio instructions for Looker. The internal data owner needs to establish the connection method.
Select dates and measures in Explore, then save a dashboard
In the available Explore, select the dates and measures that match the agreed definitions, apply the reporting-period filters and run the query. Check both the displayed values and what they count before saving.
In the current Explore interface, use Save or Share → Save to new dashboard. In a classic Explore, use Explore Actions → Save → As new dashboard. Looker’s dashboard instructions
Enter a title and select the destination folder on Settings. On Filters, decide whether to use the query's filters as dashboard filters, then save. The destination folder requires the Manage Access, Edit access level.
Reopen the saved dashboard and confirm that the period and measures are as intended. When adding another metric, check whether its date represents a visit, an inquiry or opportunity creation.
Use the fields actually defined in the organization's model. Do not assume a ready-made “AI opportunity count” exists. If the required definition is missing, resolve it with the data owner before relying on the total.
Reconcile with source records and preserve unknown paths
Finally, reconcile Looker values with the source GA4 and customer-management records. Match the site, date basis, counting unit and exclusions as well as the period.
| Reconciliation item | Findings and unresolved questions |
|---|---|
| Period and date basis | |
| GA4 dimension and included referrers | |
| Inquiry completion and duplicate rules | |
| Opportunity definition; count or value | |
| Source and Looker refresh times | |
| Unmatched values and next place to investigate |
If comparable totals differ, inspect filters, duplication and refresh timing. Visits and opportunities use different units and need not match. Preserve unknown discovery sources instead of treating them as zero or assigning them to AI without evidence.
The monthly AIO reporting guide provides a way to communicate metrics and change history. Keep confirmed observations alongside the next questions to investigate.
The initial task is complete when the records, period, definitions and reconciliation findings have been saved. Establish how far the business can trace the journey from AI discovery before treating CRS's reported growth as a target to reproduce.
FAQ
- Q. Is the 15% pipeline increase revenue?
- No. It concerns potential business before closing; the detailed count-versus-value basis is not disclosed.
- Q. Does GA4 identify every AI-related visit?
- No. Its AI Assistant channel covers recognized referrers and excludes Google AI Overviews and AI Mode.
- Q. Were CRS’s FAQs only planned for the future?
- No. Profound’s case says FAQ content had been created for key product and blog pages. The implementation date and its timing relative to the measurement period are not established, so its contribution to the 15% pipeline increase is unverified.
Sources
- [1] How CRS credit API increases AI visibility 20x with Profound (Profound) — accessed 2026-11-11
- [2] Default channel group (Google) — accessed 2026-11-11
- [3] CRS Credit API (CRS) — accessed 2026-11-11
- [4] Traffic acquisition report (Google) — accessed 2026-11-11
- [5] Traffic-source dimension scopes (Google) — accessed 2026-11-11
- [6] Mark events as key events (Google) — accessed 2026-11-11
- [7] Monitor events in DebugView (Google) — accessed 2026-11-11
- [8] Creating user-defined dashboards (Google Cloud) — accessed 2026-11-11
- [9] Admin settings - Roles (Google Cloud) — accessed 2026-11-11
About the author
Shogo MizushimaCEO of kairos Inc. / AgentSignal Developer
Develops AgentSignal, a tool for measuring AI crawler visits and AI-referred traffic, and diagnosing AIO readiness. Writes about measurement and practical improvements for AI search using observed data.
Related articles

AIO and AI search
How can gated content appear in AI search? Decide what to share and how to configure Google
Choose a public summary, indexable gated text or private access. Learn how Google's structured data, preview controls and authentication differ, and how to verify each view.
Published

AIO and AI search
How do AI crawlers discover new pages? Reading Cloudflare and OpenAI’s pilot results
Read the Cloudflare–OpenAI pilot results announced October 15, 2026. Understand page discovery, wasted fetches, change signals, access permissions, and the distinction between indexing and business outcomes.
Published

AIO and AI search
How can Bing Topics Portfolio help choose content topics for LLMO?
Use Bing Topics Portfolio beta to investigate content ideas for LLMO. Read citation metrics, evaluate Opportunity Queries, compare existing pages, and record decisions about new content or updates. Includes proposed editorial workflows and clear limits on what has been verified.
Published





