8,148 Impressions, Three Clicks: The AISymantix Conversion Study

AISymantix recorded 8,148 Google page impressions and three clicks in 28 days. All three disclosed clicks came from the branded query symantix. This study follows the next question: can the evidence, the checker, and a clear audit offer turn that visibility into qualified inquiries?

The visibility was real. So was the distance between being shown somewhere in Google and having someone arrive with a question worth answering. I wanted to preserve both facts before changing the site.

The Baseline

This is the August 13 through September 9, 2026 window from the live Digital Karma Data Warehouse. September 9 was the latest imported Google Search Console day when the baseline was captured. These are page-level impressions, not 8,148 people.

The site total comes from page records. The table below comes from disclosed query records, a different aggregation. The numbers are not added together, and withheld queries are not invented.

Disclosed queries, August 13 through September 9, 2026

Query	Impressions	Clicks	Weighted position
search visibility	1621	0	58.1912
ai search visibility	591	0	48.7496
search visibility score	384	0	52.4505
symantix	88	3	1.8409

The Smaller Opportunities Were Closer

The head terms still averaged positions deep enough to make clicks scarce. A shorter look at specific query and page pairs showed something more actionable. The Searchmetrics article was already appearing near the top for one narrow term, while the definition article was approaching a more competitive range.

Exploratory query-page observations, September 3 through September 9, 2026

Query	Impressions	Clicks	Weighted position
searchmetrics seo visibility	41	0	3.7317
search visibility definition	38	0	17.7632

Forty-one impressions do not establish a stable click-through rate. The zero beside that row is a reason to inspect intent and presentation, not a verdict that the page failed. It also shows why a sitewide average can hide the page that deserves the next hour of work.

The existing search visibility definition remains the main explanation of that metric. The Searchmetrics article answers the narrower historical and calculation question. This study measures what happens around those pages; it does not create another competing definition.

A Visit Needed a Clear Next Step

The audit page had a very ordinary problem. It described a paid review priced from $299 to $999, then offered a free audit through a button that returned to the same page. There was no request form on that page. A reader could understand the subject and still have to work out how to ask for help.

On September 12, I separated the free page check from the paid audit and added the request form. A completed check now points to a review with a stated price range and an agreed scope. The inquiry keeps its referring source, and qualification requires a human decision after the request is read.

The Research Travels With a Purpose

AISymantix owns the original finding and method. Signal Architect Group explains the implementation as a case study. DataSetSEO publishes the reproduction material. RealSEOLife examines the diagnostic decision, and Krisada.com explains the portfolio decision. These are related properties under the same ownership, so this is a first-party portfolio study.

Read the Signal Architect Group case study or inspect the DataSetSEO reproduction record. Both point back to the original evidence here.

What We Will Measure

The first complete post-launch window is September 13 through October 10, 2026. The primary outcome is qualified audit inquiries, shown alongside the actual visit and request counts. Completed checks, referring sources, Google clicks, and the fixed query-page pairs are separate readings.

The structured-data and entity-architecture library pages provide context because their editorial content stays unchanged. They still share the sitewide offer and measurement changes, so they are not untreated controls. Several changes launch together. A later improvement would not identify which one caused it.

There is no comparable historical checker or inquiry conversion rate under this new measurement. Those baseline fields are null in the dataset. Future results will be reported as observed counts; missing history will stay missing.

What Would Change My Mind

If readers use the checker but do not request audits, the next investigation is whether the offer answers their question. If requests arrive but are unsuitable, the promise or audience needs work. If the publications produce no visits, adding more requests to the same pages will not fix the distribution problem. The measurement should tell us which decision to revisit.

The earlier five-site crawler study found no reliable sitewide visibility forecast from rising Good Bot counts. That remains a useful boundary here. Crawler access, an observed AI citation, an AI referral visit, and an audit request each need their own evidence. Google also includes AI-feature traffic in overall Web reporting, so these GSC tables do not isolate AI Overviews or AI Mode.

Download the complete baseline JSON. It includes all 28 daily page rows, 173 disclosed queries, 39 observed page URLs, the selection rules, and the planned comparison. Reuse is available under CC BY 4.0, with attribution to AI Symantix. No post-launch conversion improvement is claimed at publication.