Our Website Traffic Increased. Our Leads Didn’t. Here’s What AI Changed.

For years, the basic logic of a successful SEO campaign was straightforward: improve rankings, earn more qualified search traffic, and create more opportunities for phone calls and estimate requests. That logic shaped our own expectations when Garza Web Design began a focused SEO campaign on October 15, 2025.

The visibility gains arrived. The business outcome did not arrive in the same proportion.

Demand for our work softened after AI use accelerated in the fourth quarter of 2025. At the same time, our website became substantially more visible in traditional search. Those two facts can exist together, and that is the uncomfortable part of this story.

We are not presenting this as proof that AI alone caused fewer inquiries. Many factors influence demand, including economic conditions, competition, budgets, seasonality, referrals, and changes in buyer behavior. But the timing was meaningful enough that ignoring it would have been irresponsible. So we rebuilt our website, changed our content strategy, and turned our own marketing into a documented experiment.

SEMrush estimated that monthly organic traffic increased from 338 in October 2025 to 761 in September 2026, a gain of 125%. During the same period, the number of keywords ranking in Google’s top 100 increased from 176 to 337, a gain of 91%.

Third-party estimates are not the same as measured visits, so we checked Google Search Console as well. Across equal 169-day periods before and after the campaign began, organic clicks increased from 113 to 189. Average click-through rate rose from approximately 0.045% to 0.115%, and average position improved from 36.2 to 22.7.

One commercially important phrase, “web design richmond va,” illustrates the pattern. Clicks increased from 17 to 24 even though impressions declined from 5,154 to 3,487. The approximate click-through rate therefore more than doubled, from 0.33% to 0.69%. We were appearing in stronger positions and earning a larger share of the searches in which we appeared.

These are the kinds of improvements agencies normally celebrate. We do celebrate the technical progress. But clicks and rankings are intermediate measures. They matter only when they help create conversations, opportunities, and revenue.

Despite the stronger search performance, the campaign did not produce the increase in phone calls and estimate-request submissions we expected. This is not a story about an SEO campaign failing to create visibility. It is a story about visibility becoming less reliable as a predictor of action.

AI-powered answers can satisfy part of a buyer’s research process without sending that person to a website. A prospective client can ask for recommendations, compare approaches, learn typical costs, build a shortlist, and even draft an internal project brief before visiting an agency’s site. When a visit finally happens, it may occur later in the decision process—or not at all.

This creates a measurement problem. A business may be mentioned or cited inside an AI answer and still receive no attributable session. A buyer may read an AI summary, remember a company name, and return through another channel days later. Traditional analytics were built around visits and clicks. They do not fully describe a journey that may begin inside an answer engine.

Our Analytics data also contained a large amount of referral spam during the post-campaign period, including 500 sessions attributed to trafficheap.cc. That contamination makes headline traffic and conversion-rate comparisons unreliable. For this analysis, we give more weight to Search Console, SEMrush trends, verified form submissions, and actual phone activity than to unfiltered session totals.

In August 2025, SEMrush reported no AI mentions for Garza Web Design. By September 2026, that figure had risen to 15. The SEO campaign began on October 15, 2025, so the increase gives us a useful trend to follow—but it does not, by itself, prove that the campaign caused those appearances.

Google Search Console’s generative AI reporting provides another view. From June 9 through September 8, 2026, Garza Web Design received 949 impressions in Google generative AI features: 150 during the partial month of June, 417 in July, 278 in August, and 104 during the first eight days of September. The homepage accounted for roughly 70% of those impressions.

This is progress. It tells us that our content is increasingly available to AI-assisted search experiences. It does not tell us how many people visited, called, requested an estimate, or became a client. That distinction is central to the experiment.

Hiding from the change would not restore the old buying journey. We chose to respond by learning in public.

The redesigned Garza Web Design website is structured around the questions people ask before they contact an agency. Service pages explain scope and fit more directly. Case studies connect strategy to documented evidence. Technical markup helps search engines and AI systems identify who we are, what we do, where we work, and which claims are supported. Calls to action are clearer, but the site does not force every visitor into an immediate sales conversation.

We are also changing what we measure. Rankings and organic traffic remain useful, but they sit beside AI mentions, cited pages, generative-search impressions, branded searches, qualified phone calls, estimate submissions, and the quality of the resulting opportunities. No single dashboard explains the whole journey.

This is an experiment, not a victory announcement. The new website is our response to what we have learned so far. In the coming weeks and months, we will watch whether stronger explanations, clearer evidence, and better alignment with AI-assisted research produce more meaningful conversations.

Many owners and managers are being told either that AI will solve everything or that it will destroy everything. Neither claim is useful when you are trying to make payroll, plan next quarter, or understand why a once-reliable marketing channel feels different.

Begin by separating visibility from outcomes. Are rankings down, or are they up while inquiries are down? Are fewer people visiting, or are similar numbers visiting without acting? Are branded searches changing? Are prospects mentioning ChatGPT, Gemini, Copilot, or Google’s AI results during sales conversations? Are strong leads arriving later and better informed?

Clean the measurement before drawing conclusions. Remove referral spam. Confirm that forms and phone calls are tracked correctly. Compare equal time periods. Label third-party estimates as estimates. Separate current AI mentions from cumulative appearances. Most importantly, record lead quality and revenue rather than treating every session as equally valuable.

Then make your expertise easier to verify. Publish clear service information, real examples, responsible claims, useful answers, authorship, dates, sources, and structured data. The goal is not to write for robots. The goal is to make your business understandable and credible wherever a human or an AI system evaluates it.

We will continue comparing Search Console performance, AI visibility, qualified calls, estimate requests, and lead quality after the redesign. We expect the picture to change as the new pages are indexed and as AI platforms evolve.

If the redesign improves inquiries, we will show the evidence. If traffic continues rising without corresponding demand, we will say that too. The purpose of this experiment is not to defend an old playbook. It is to learn what visibility means now—and to help our clients adapt with clearer evidence and less fear.

AI has changed the digital marketing landscape quickly. Businesses do not need false certainty from their web partners. They need honest measurement, thoughtful experiments, and a willingness to change course when the results demand it. That is the approach we are taking with our own business.