What My Shopify Data Taught Me About AI Traffic
Gary WhittakerPart 3 · Measurement without the scoreboard
The data mattered because it changed what I paid attention to. It did not give me a secret formula.
AI Search Visibility for Creators · 6-Part Series
This series began with a real 2025 discovery experiment and has been updated for 2026. The lesson is not a secret ranking formula. It is how to build useful, crawlable, connected content over time and measure what happens.
The useful question changed
When I first wrote this article in 2025, I led with exact Shopify numbers. That was useful as a snapshot, but it put too much weight on a small historical data set. A better question is:
Over time I learned to separate several things that are easy to blur together: referral source, landing page, engagement, product interest, checkout activity, repeat visits and actual orders. One number rarely explains the whole journey.
What I measure now
Source
Where did the visit appear to come from: search, AI assistant, email, social, referral or direct?
Landing page
Which page did the visitor enter through, and what question was that page answering?
Next click
Did the page lead naturally to another useful resource, product, signup or support route?
Outcome
Did the visit create something useful: time spent, return behavior, an email signup, an inquiry or a sale?
Why exact attribution is harder than it looks
Traffic sources can be mislabeled, collapsed into direct or referral buckets, or split across different source/medium names. AI products can also change how they append referral data over time. That is why I look for repeated patterns instead of treating one dashboard row as proof of causation.
OpenAI says ChatGPT search referral URLs can include utm_source=chatgpt.com, which can help publishers identify inbound traffic. Google Search Console is also expanding reporting for generative and multimodal search experiences. Those tools improve visibility into the journey, but they still do not tell you exactly why one page was chosen over another.
Primary references: OpenAI: ChatGPT Search · Google: Search Console multimodal reporting
The operating rule I use
Make one meaningful change, give it time, then evaluate the page in context. A title change, stronger opening answer, better internal link or cleaner product path may help. Changing everything at once makes it harder to know what improved.
Next: make the page easier to understand
Part 4 explains how I format content for clarity without pretending a specific paragraph length or list style is an AI ranking signal.
Continue to Part 4 — Content Formatting for AI Search →
ASK JACK Member Case Study
See the full JackRighteous.com implementation
The member case study shows the slower work behind the results: what I changed, how I connected pages, what I measured, what I ignored, and how I am adapting the site for AI-assisted customers.
Available with JR Creator Library, Creator Pro and Complete Access.
