Home/Case Studies/evolveamz.com AI Search Growth
[01] Case Study: Our Own Site

How evolveamz.com Went From 65 to 412 Organic Sessions a Month and Became an AI-Cited Source

Twelve months of GA4 data from our own site, published with the parts most case studies leave out: the bot traffic, the broken lead tracking, and the things that went backward. Numbers pulled September 15, 2026.

65 → 412

Google and Bing organic sessions per month, Oct 2025 to Mar 2026 average vs Jun to Aug 2026 average

GA4

+385%

Organic sessions, last 30 days vs the same 30 days in 2025 (383 vs 79)

GA4, Aug 16 to Sep 14

~670

Visits from AI assistants in 12 months

ChatGPT, Claude, Gemini, Perplexity

48%

AI-assistant visitors engaged, vs 21% sitewide

GA4, last 30 days

[02] Starting Point

Winter 2025: A Services Site Nobody Was Finding

Ian Smith has sold on Amazon since 2015 and founded Evolve Media Agency in 2017. Through 2025, evolveamz.com was a typical agency site: a homepage about our photo, video, and listing services, a pricing page, a portfolio, and a blog that had been updated occasionally since 2022. It worked as a brochure for people who already knew us. It did not bring anyone new.

From October 2025 through March 2026, Google and Bing organic search averaged about 65 sessions a month. Earlier in 2025 it had been closer to 116 a month, so the site was not just flat; it was decaying. AI assistants were not a measurable channel at all, partly because GA4 had no AI channel grouping until June 2026 and partly because nothing on the site was worth citing.

We had one advantage: we write and ship content for clients every week, and we were already using AI-first production internally. The question was whether the same approach, applied to our own site at volume, would move the numbers.

[03] What We Published

Long-Form Guides, Daily, Plus the Technical Layer

1

A guide library, published in daily batches (April 2026 onward)

Long-form guides of roughly 3,300 to 4,100 words each, scheduled one per day in batches of 20 to 40, across Amazon operations, listing optimization, photography, PPC, email, Shopify, TikTok Shop, and AI search. Every post carries a quick answer near the top, a stat bar, a table of contents, 10 to 12 FAQs with FAQPage schema, a key takeaways block, sources, and related guides. By September 2026 the site had 346 live URLs. The reasoning is in the compounding content moat.

2

The AI Search Hub (15 definitional and comparison guides)

A dedicated cluster at /ai-search/: what AI search optimization is, how each assistant picks products, agency comparisons that name competitors, and decision frameworks. This is the cluster that earns the most AI-assistant entries.

3

A content-first homepage rebuild (June 2026)

We replaced the services-first homepage with a hub that features guides and posts, keeps services to one section, and uses click-to-load video and lazy images to stay fast. The goal was to give new visitors somewhere to go instead of a pitch.

4

The technical layer

A robots.txt that welcomes GPTBot, ClaudeBot, PerplexityBot, Google-Extended, and Applebot; llms.txt and llms-full.txt; a schema graph on every post (Article, FAQPage, BreadcrumbList, Organization, Person); and consistent entity information. Method in the schema implementation guide.

5

Original first-party data

Posts built on our own numbers, our own tests, and our own mistakes. Assistants cite sources that say something specific; this case study is one of them.

[04] Results

What Moved, With the Fine Print

All figures are GA4 for evolveamz.com, pulled September 15, 2026. Raw session totals are excluded on purpose; see the next section for why.

Organic search: 65 a month to 412 a month

Google and Bing organic went from about 65 sessions a month (October 2025 to March 2026) to 412 a month (June to August 2026), peaking at 497 in July. March 2026 was 75; July was 497, more than 5x in four months. The last 30 days were 383, up 385% on the same window in 2025.

AI assistants: about 670 visits in 12 months, and still growing

ChatGPT about 316, Claude about 173, Gemini about 103, Perplexity about 51, plus smaller sources. GA4's own AI Assistant channel labeled only 409 of them; the rest were filed under Referral or (not set) and had to be recovered by source. In the last 30 days the channel had 126 sessions, up 15% month over month, the only sizable channel still growing.

AI visitors read more than anyone else

AI-assistant sessions ran 48% engaged with 48 seconds of average engagement in the last 30 days, against 21% and 7 seconds sitewide. Organic search was 44% and 48 seconds. The assistants send people who want the answer.

The guides became the front door

Almost all organic and AI traffic lands on guides and posts, not the homepage or service pages. The local-business AI search guide is the single top AI-assistant entry page; the A+ Content guide drives the most key events of any content page.

Leads: not measurable for this period

Our GA4 key events were built for the previous site's buttons and stopped firing as the site changed. Calendly clicks were never a key event. We cannot say AI or organic visitors converted to booked calls at any particular rate, and we are not going to guess. New events went in September 2026; the next version of this study will have the number.

[05] What Did Not Work

The Part That Makes the Rest Believable

About half of raw traffic was bots

The site logged 32,686 sessions in 12 months. Singapore alone was 39% of users at about one second per visit (50% in the last 30 days), plus US data-center cities and email link scanners from our newsletter platform. Sitewide averages are meaningless because of it, which is why this study quotes organic and AI sessions only. We are now challenging automated traffic at Cloudflare and building a "humans" comparison in GA4.

The homepage rebuild halved homepage Google entries

Google entries to the homepage went from 53 to 26 after the rebuild, and the title tag still read like the old positioning. Content-first was the right call for the site; we still lost something on the homepage itself and are re-checking Search Console queries to find out what.

Ranking without clicking

An earlier Search Console review showed about 91,000 impressions on ranking guides at roughly 0.2% click-through. Pages sat on page one and nobody clicked, because the titles were written for the topic, not the searcher. Title and meta rewrites on the top-impression guides are the highest-return item on our current list.

Some winners decayed

The AI crawler list guide, our second-biggest Google entry page over three months, fell from 37 entries to 18 in the latest month. A 2024 post on search shadowbans went from 100 Google entries to 9. Content that ranks still needs refreshing.

We leaked our own traffic

Hundreds of internal links still pointed at an old domain and an off-site PDF page; 163 clicks left for the old portfolio, 64 for the lead magnet, in a year. Two lead pages returned 404. Nobody noticed until the crawl. The fix is a redirect plugin and a link rewrite across every post.

Measurement lagged the strategy

No Search Console link in GA4, an internal-traffic filter stuck in testing, newsletter clicks without UTMs, and AI referrals mislabeled. The content worked before we could prove it, which is the wrong order.

Disclosure: this is our own site, we sell AI search services, and we chose which numbers to show. What we could not isolate: how much of the organic growth came from the content volume versus the homepage rebuild versus Google's own algorithm changes over the same period. We did not run a controlled test; nobody does on their own site.
[06] What We Would Tell a Client

Five Things This Year Taught Us

01

Volume works, and it compounds slowly. Organic took about six months of daily publishing before it moved; AI referrals took longer and were still climbing at month twelve.

02

Guides beat service pages for discovery. Nearly every new visitor arrives on a guide. Service pages exist to convert people the guides bring.

03

AI visitors are few and valuable. 670 visits is small next to organic, but they read, and the channel is the only one growing every month.

04

Fix measurement before the content, not after. We are rebuilding the tracking now and lost a year of lead attribution to it.

05

Crawl your own site. Old links, 404s, and bot traffic hide in plain sight. Half of what we found needed a spreadsheet, not a strategy.

06

Publish the caveats. This page names them because an assistant, or a buyer, that reads only the wins has no reason to trust the source.

Free · 30 minutes · No pitch deck

Want the Same Playbook on Your Site?

Book a free call, or read what the AI search optimization blocks include. We will show you your current organic and AI baseline and what the first 90 days would look like.