From 186 to 938 Impressions: How a Las Vegas Shopify Store Built Visibility Across Google and AI Channels

A practical IceStoreGroup case study on systematic SEO + GEO work in a policy-constrained product category, where local trust and data quality matter as much as the storefront itself.

Over four months, Google impressions grew roughly fivefold, clicks more than doubled, and ChatGPT became a measurable traffic source. The progress did not come from one setting or one file, but from improving the store as a connected system.

A finished storefront is not the same as a visible business

When we completed this Shopify store, the project looked finished from the outside. The catalog was in place, product pages worked, and customers could select products and place an order. But that is not enough for a business. A store can function correctly and still remain almost invisible to people searching on Google or asking an AI system for local recommendations.

This project had an additional layer of difficulty. The client is a local Las Vegas retailer selling personal wellness and health products. It is a sensitive category: customers expect trust, privacy, and clear delivery information, while advertising, shopping, and AI channels apply additional policy restrictions. Adding keywords or enabling a single Shopify channel was never going to be a complete strategy.

At first, the obvious question was how to grow organic traffic. A more useful question soon emerged: could Google, Shopify Catalog, and AI systems correctly understand what the store was, where it operated, which products were available, and why the business should be trusted? That question became the foundation of our SEO + GEO work. In this context, GEO means preparing a store so generative search systems and AI agents can find, interpret, and use its information more reliably. It does not guarantee inclusion in every AI answer; it creates a measurable layer of visibility that can be improved over time.

The baseline: the website existed, but its discovery layer was weak

On April 20, 2026, we recorded the starting point. Google Search Console showed 19 clicks, 186 impressions, and an average position of 46.2. The 10.2% CTR looked respectable, but it came from a very small and narrow query set. The store was appearing infrequently, and most visibility was concentrated around a limited number of searches. In two tested local AI queries, the business was not found in either ChatGPT or Gemini. Its readiness for AI agents was still at a basic level.

This did not mean the store was poor. It meant that search and AI systems did not yet have enough consistent evidence. The home page carried most of the visibility, while collections and products had little search presence. The local business entity was weakly supported, and information about delivery, policies, availability, and products was distributed across disconnected layers.

* The source export did not provide an overall August average position. August position was 18.02 in the US segment and 9.98 on mobile.

The first turning point: more visibility did not immediately look like success

The first meaningful checkpoint came in June. Impressions for the latest 28-day period increased from 352 to 520, a gain of 47.7%. Product-result impressions rose from 100 to 170, while their average position improved from 41.96 to 30.25. At the same time, clicks fell from 36 to 18.

Looking only at clicks would have made the work appear unsuccessful. Looking only at impressions would have produced an overly optimistic story. Both interpretations would have been incomplete. The evidence showed that Google was testing the store across a broader set of queries and exposing more categories and products. Those new impressions had not yet become reliable traffic because positions, snippets, and trust signals still needed work.

This is one of the most useful lessons from the project. SEO rarely moves in a straight line. When semantic reach expands, overall CTR can initially decline because a site begins appearing in searches where it has not yet earned strong positions. The right response is not to hide the decline or panic. It is to identify which pages gained impressions, which queries triggered them, and what is preventing those impressions from turning into clicks.

What we actually changed

There was no secret SEO tactic behind the result. We developed several connected layers in a deliberate sequence.

First, we strengthened the local business entity. The contact page was expanded with clear Las Vegas and service-area information. A Google Business Profile was created. Dedicated pages explained how ordering and local delivery worked. By August, two of those pages generated 225 impressions and 14 clicks in 28 days. The volume was still modest, but it represented validated demand and a new organic entry point that no longer depended only on the home page.

Second, we made the store's answers more structured. FAQs were cleaned up and expanded, valid FAQPage structured data was added, and service and local-business information was strengthened. Shopify Knowledge Base was populated with answers about delivery, privacy, policies, and products. This reduced uncertainty for customers and gave search and AI systems a clearer, more consistent business context.

Third, we developed the commerce discovery layer through Shopify Agentic Storefronts, Shopify Catalog, and product availability for AI channels. Discovery files and crawler rules remained accessible, but we deliberately did not reduce GEO to a single llms.txt file. One file cannot repair weak product descriptions, incomplete attributes, conflicting delivery claims, or missing external trust.

Finally, we monitored technical health. Real-world Core Web Vitals continued to pass. By August, mobile LCP had improved to 1.4 seconds and INP to 113 milliseconds. A decline in Accessibility was moved into a separate diagnostic task instead of being hidden behind a perfect SEO score. Technical optimization only creates value when it improves discovery without introducing a new customer-experience problem.

July: the system started to respond

By the July checkpoint, the accumulated changes were clearly visible in search performance. Over the latest 28 days, the site received 42 clicks versus 29, 970 impressions versus 658, and its average position improved from 28.64 to 20.24. Compared with the April baseline, search reach had increased more than fivefold.

More importantly, growth was no longer concentrated on the home page. A local delivery page, the how-it-works page, collections, and individual products began generating search activity. Product-result clicks rose from 4 to 12, impressions from 256 to 517, and average position improved from 35.18 to 27.22.

At the same time, Shopify Catalog contained 92 products, the Knowledge Base was available to AI channels, and Shopify attributed three sessions to ChatGPT. Three sessions were not yet a major sales channel, but they were measurable evidence that the store had started receiving visits from an environment where it had not appeared in our initial local tests.

August: the search gains held, and the AI signal strengthened

The next checkpoint mattered more than the July jump itself. A single increase can be temporary; progress becomes more credible when the result holds in the following period.

In the August Google Search Console export, the site recorded 45 clicks versus 44 in the preceding comparable period. Impressions were 938 versus 955, essentially stable, while CTR improved from 4.61% to 4.80%. In the target US market, average position improved from 20.69 to 18.02. On mobile, clicks increased from 32 to 42 and average position improved from 11.69 to 9.98, placing the store within the top ten on average for that device segment.

AI visibility also strengthened. Shopify attributed eight sessions to ChatGPT, up from two in the previous comparable period. The number of available products in Shopify Catalog increased from 92 to 114. This does not mean that every product will appear for every AI request. It confirms that the technical and content foundation had begun generating measurable visits and a broader product presence.

Not every metric improved. Google's product rich-result impressions fell from 509 to 315, clicks from 14 to 7, and average position moved from 25.35 to 30.42. We did not force the decline into one convenient explanation. Availability and channel eligibility, competition, Product/Offer quality, and incomplete product data were all plausible factors. Until tested, they remained hypotheses rather than a confirmed diagnosis.

Why a product catalog cannot be judged by one number

A separate product-data review identified 438 unique products in the export. Of those, 402 were Active, 374 were published, and 321 were active, published, and had positive inventory. Shopify Catalog, however, showed 114 available products.

The easiest conclusion would be to call the difference an error. That would have been premature. Shopify Admin status, channel publication, Markets, taxonomy, sellability, and the policies of each AI or shopping channel are separate layers. A sensitive product category adds mature-content restrictions.

The practical task is not to force every number to match, but to classify the gap and improve the elements the merchant can actually control.

The review also revealed actionable gaps. Most products lacked individual SEO descriptions and image alt text, several attributes were incomplete, and some descriptions were duplicated. At the same time, Catalog was already reading titles, images, and base descriptions correctly. The problem was not a complete integration failure; it was the quality and completeness of specific data. That is a much more useful conclusion than the generic instruction to 'improve SEO.'

Six practical lessons for Shopify merchants

1. Record a baseline before making changes. Preserve clicks, impressions, CTR, and average position by market, device, page, and search appearance. Without a starting point, progress cannot be demonstrated and declines cannot be isolated.
   
2. Do not judge performance by one metric. Higher impressions with fewer clicks may indicate broader reach, while a decline on one product with stable sitewide visibility may be a local issue rather than a sitewide collapse.
   
3. For a local business, prove that the entity is real. Contact details, service areas, a Google Business Profile, consistent operating hours, and accurate delivery terms build trust that city-name repetition cannot replace.
   
4. Connect content and data. Visible pages, FAQs, structured data, Shopify Knowledge Base, Catalog, and actual commercial policies should describe the same business without contradictions.
   
5. Check products across the full chain. Active status in Shopify Admin does not automatically mean publication, inventory, eligibility, or visibility in an AI channel. Diagnose the transitions between layers instead of searching for one universal switch.
   
6. Repeat measurement after every cycle. SEO + GEO is a management process: baseline, hypothesis, change, verification, conclusion, and next priority. Without a recheck, even good work remains unproven.
   

What this case means for other Shopify stores

I am deliberately not attributing sales growth to SEO + GEO when the available data cannot prove it. Paid traffic remained the dominant source in one reporting period, and fresh commercial attribution was not available for August. The confirmed result is narrower and stronger: organic visibility grew and held, target-market positions improved, more pages began contributing, Shopify Catalog expanded, and ChatGPT became a measurable traffic source.

Another store should not expect to reproduce the same figures automatically. Outcomes depend on domain history, competition, category, market, catalog size, and the quality of the starting data. The process, however, is repeatable: establish the current state, identify constraints and competitive gaps, create a prioritized plan, implement changes, and measure again.

This project matters to me not because every issue has already been solved. The product layer still needs further work, and the category restrictions have not disappeared. Its value is that a local Shopify store in a difficult niche moved from basic web presence to sustained organic visibility and its first measurable AI-driven visits. From April to August, impressions increased from 186 to 938 and clicks from 19 to 45. Google began seeing not only the home page, but useful content, collections, and products.

If a Shopify store is operating but its products remain hard to find, traffic is not growing, or different channels report conflicting data, the right starting point is not a promise of quick rankings. It is evidence-based diagnosis across technical health, indexing, product data, local trust, SEO, GEO, Shopify Catalog, and AI readiness. This is how we work at IceStoreGroup: connect technical changes to a business objective, measure the outcome, and clearly separate what has been confirmed from what still needs testing.

Case-study data is based on Google Search Console, Shopify Admin, Shopify Catalog / Agentic Storefronts, and PageSpeed Insights checkpoints from April through August 2026. The reports use consecutive 28-day comparison windows and do not treat those windows as direct sales attribution.