Does AI-driven traffic actually work? How to increase traffic and conversions using AI chats

Is artificial intelligence a threat to SEO or its ally? AI is evolving rapidly, and users are keeping up by adopting new trends andsolutions offered by the market — changing how they behave online and what theysearch for. Search engines are integrating AI-generated answers directly into searchresults, while users are increasingly relying on tools like ChatGPT and Geminiwhen searching for information. In this context, AI should be viewed not as athreat to organic traffic, but rather a new opportunity for growth. In this case study, you’ll learn how timely analysis and a clear strategy helpedAweb team to optimize the content so it appeared in AI-generated answers anddrove the traffic.

Client

A network of private clinics offering a full range of medicalservices

Our goal

Leverage the opportunities offered by AI-powered chatbots and virtual assistantsto grow the clinic network’s website by aligning content with modern searchtrends, using AI chats as a traffic acquisition channel, and increasing userengagement. Objective — keep up with the market, use new technologieseffectively, and create additional entry points for potential patients.

Initial data and project details

Niche: Private clinic Region: Kyiv Primary search engine: Google CMS: WordPress Promoted by Aweb: 4 years and 8 months Testing launch: Analysis and proposals were developed in March2025 with implementation starting as early as April 2025. Project details at the start of testing: According to a customGoogle Analytics report tracking AI-sourced traffic for the period of March1–31, 2025, the following findings were observed:

• over 100 sessions;
• over 90 active users;
• ChatGPT was the primary source of traffic.

By that point, no efforts had been made to optimize the content forAI.

Strategic approach by Aweb

1. Analysis and Reasoning At the research phase, Aweb evaluated how AI responds to popular medical queriesand discovered that the client’s brand was already appearing in ChatGPT answers,even without AI-focused optimization. Important note: The data below represents the current state of the analysisas of March 2025. As AI continues to evolve, the format of AI-generatedresponses may change. All queries were run in Incognito mode to ensure unbiased results. Sample queries: «Can you suggest a gynecologist in Kyiv?» — ChatGPT generated alist of doctors with active links. Aggregators such as doc.ua and likarni.comranked first, followed by physicians from private clinic websites. One of thecited sources was a FAQ page titled “Best Gynecologists in Kyiv,” highlightingthe effectiveness of clear titles and structured rankings. When the same query was run in a new chat, ChatGPT selected a broader list ofdoctors from doc.ua while still including private clinics. Similar results were observed for other queries, such as phlebologists. In onecase, AI suggested clinics, while in another it listed individual doctors usingdata from commercial pages. Our analysis revealed an interesting insight: one of the cited websites hadrecently changed its page structure by updating the H1 and Title to the “BestDoctors [Specialty]” format. Our team noticed this change while reviewingarchived versions of a competitor’s website. Despite limited page optimization, the Aweb client’s website was cited morefrequently in GPT responses, suggesting that concise, highly relevant contentmay outperform SEO-heavy pages.

Analysis summary:

• Structured FAQs and list-based formats (tops/rankings) consistently performwell in AI-generated responses;
• Even without large-scale page optimization (e.g. pages without backlinks,well-structured texts) it is absolutely possible to be mentioned in ChatGPT —highlighting the importance of clear titles and a logical contentstructure;
• Concise and relevant content proved to be more efficient than pagesoverloaded with SEO-phrases.

Sample structure: 2. Recommendations for AI traffic optimization of the client’swebsite Based on these insights, our team developed a set of solutions aimed at scalingresults and systematic use of AI traffic as an additional acquisition channel.

1. scaling the “Top Doctors” format:
   • lists of doctors by specialty (top gynecologists, top phlebologists inKyiv);
   • niche queries (doctor + specific condition, e.g., “gynecologist forendometriosis treatment”).
2. top treatment approaches (e.g., “Best methods for treating cervicaldysplasia”).
3. optimizing meta tags and H1–H5 headings with a clear, clean structure(avoiding unnecessary symbols and filler wording).
4. enable indexing of doctor filter pages to create relevant query combinations(e.g., “gynecologist Pozniaky,” “gynecologist for pregnancy planning”).
5. designing a template for new pages and content, allowing the client to addthese pages independently.

The proposed AI optimization plan served as a baseline entry point, enabling theAweb team to further develop the broader strategic approach.

Initial results

Effective collaboration on the client’s side allowed rapid implementation andscaling of our recommendations. Sample pages created following Aweb’s recommendations: Results:

• Pages generally appeared in ChatGPT responses within two weeks following thepublication.
• Example of initial results: a new page being included in a ChatGPT-generatedresponse.

By late May, following two months of dedicated content work, the first measurable results appeared in Google Analytics.

• sessions increased by 30%;
• active users increased by 21%;
• events increased by 70%;
• key events increased by 12.5%.

Screenshot from a Google Analytics report (analysis data comparing April 1–30, 2025 and May 1–31, 2025)

Progress update

As we continued analyzing ChatGPT responses and refining our optimizationstrategy, the client consistently implemented updates. Within three months oftesting, the website demonstrated significant growth in key metrics compared tothe three months prior to the experiment:

• The number of sessions increased by 109%;
• Active users increased by 109%;
• Events increased by 145%;
• Key events increased by 192%.

Meanwhile, the average duration of session declined by 62% (from 6:24 to 2:24),which was expected, as user behavior from AI-driven sources is still evolving.We consider this a normal adaptation phase for a new channel of traffic. ChatGPT drives the highest volume of traffic. Perplexity is showing the fastest growth, although its absolute volume remainslow. Screenshot from a Google Analytics report (analysis data comparing January 1 –March 31, 2025 and April 1– June 30, 2025):

Results after five months of work

After five months of testing, the client’s website appears in nearly all ChatGPTresponses for relevant queries. Sample queries in Incognito mode (client mentions highlighted in red): 1.Query: “Can you recommend a good gynecologist in Kyiv?”: 2. Query: “Can you recommend an experienced urologist in Kyiv?” Beyond this, the growing volume of brand mentions resulted in another positiveoutcome — a stronger clinic reputation within ChatGPT. As a result, for thequery “best clinics in Kyiv,” ChatGPT ranks the client’s clinic first andhighlights it as the most frequently chosen option. According to Google Analytics, five months of working with AI chat platformsdelivered the following results:

• sessions increased by 258%;
• active users increased by 255%;
• number of events increased by 323%;
• the number of key events increased by 407%.

Meanwhile, this traffic shows a relatively high conversion rate compared to itssession volume. Screenshot from a Google Analytics report (analysis data comparing November 1,2024–March 31, 2025 and April 1–August 31, 2025): The results are reflected not only in Google Analytics metrics but also in areal feedback provided by the client. According to the client, clinic doctorsreported a growing number of patients who discovered the clinic via ChatGPT. Onesenior doctor alone saw six such patients in a single month.

Summary

Optimizing content for AI and integrating chatbots as a traffic source is areal, measurable, and promising channel for website growth. A well-structuredcombination of research, analysis, and a clear action plan drives growth insessions, engagement, and conversions while strengthening brand reputation. Why are we confident that our actions delivered results? Some may assume that the results are driven solely by the general advancement ofAI and that increased traffic from chatbots is inevitable. This is onlypartially true, as more users are turning to chat-based tools as a primarysource of information. However, we believe the results were a direct outcome of our targeted actions.This is supported by insights from other medical projects where ourrecommendations have not yet been implemented or are still at an early stage ofadoption. Example 1: A private clinic in Kyiv. Monthly organic traffic: 69,000. Data for a 1.5-month period: 137 sessions and 16 key events. Example 2: A network of private clinics in Kyiv. Monthly organic traffic:26,000. Data for a 1.5-month period: 103 sessions and 7 key events. Data for the client featured in this case study over a 1.5-month period: monthlyorganic traffic — 74,000; sessions — 645; key events — 70. So, the answer is clear: yes, traffic from AI-driven sources does reachwebsites. However, the difference between projects that actively leverage AItraffic and those that don’t is significant — around 5–6x on average.

Project team

Project Optimizer: Anastasiia Hrytsenko Project Manager: Aliona Lagunina Reviewer/Supervisor: Nataliia Pronina