# AI search results, proved with dated screenshots Source: https://www.smartecomseo.com/geo-agency/results/ Updated: 2026-09-29 A single screenshot of ChatGPT naming a brand proves very little. Answers change daily. This page sets out the evidence standard we hold ourselves to, and it is where client results go once each client approves them. ## Six things every AI result on this page will show ### The exact prompt Word for word, in the language and market it was asked, so anyone can run it again. ### The assistant and mode ChatGPT, Perplexity, Gemini, AI Overviews, AI Mode or Copilot, and whether search or shopping was involved. ### The date Every screenshot is dated. A result without a date cannot be checked against anything. ### Before and after Mention and citation counts across the whole tracked prompt set, at baseline and at each re-run. ### Competitor context Share of voice against the same competitors, so a gain reflects your brand, not the whole category moving. ### Business outcome AI referral sessions and orders from Shopify analytics, with the tracking gaps stated. ## What a documented result looks like ## What passes for AI proof, and what we publish | | What you often see | What we publish | |---|---|---| | Evidence | One undated screenshot | Dated screenshots for a fixed prompt set | | Prompt | Hidden or cropped | Full prompt text, market and language | | Scale | One lucky answer | Mention counts across 25 to 100 prompts | | Baseline | None | Counts before work started | | Competitors | Not shown | Share of voice against named competitors | | Revenue | Implied | AI referral sessions and orders, with gaps stated | ## How a result gets onto this page 1. **Baseline** (Month 0): The audit prompt set is run and every answer saved with a date, before we change anything. 2. **Fixed tracking set** (Month 1): A subset of prompts is locked for monthly tracking. We do not swap in prompts that happen to look good. 3. **Monthly re-runs** (Monthly): Same prompts, same markets, same method. Mentions, citations and share of voice counted each time. 4. **Revenue check** (Monthly): AI referral sessions and orders from Shopify and GA4 compared with the baseline period. 5. **Client approval** (Before publishing): Many clients prefer not to be named. Nothing is published without written approval, and anonymized results say so. ## Why this page is still filling up We publish AI search results only when they meet the standard above and the client has approved them. Results will appear here as clients approve them. Until then, this page shows the method, not invented examples. ### What we can already show Our SEO results are documented in full. A Chinese automotive accessories brand went from close to zero to almost 100,000 organic visits a month and $758,772.94 in organic revenue in seven months. A Chinese tech brand earned $150,000+ in organic revenue in four months, and Bing drove $25,000+ of it. That Bing number matters for AI search too, because Microsoft Copilot answers are built on Bing's index. Read the [automotive accessories case study](/case-studies/chinese-tech-brand-seo-case-study/) and the [1,000 clicks a day case study](/case-studies/chinese-tech-brand-1000-clicks-a-day/). ### What we measure in AI search - **Prompt tracking:** mentions, citations and share of voice across a fixed prompt set, with dated screenshots. - **Referral data:** sessions and orders from ChatGPT, Perplexity, Gemini, Copilot and others in Shopify analytics and GA4. ChatGPT links usually carry utm_source=chatgpt.com. Some AI visits arrive without a referrer, so these numbers undercount. - **Copilot citations:** Bing Webmaster Tools' AI Performance report, which shows how often Copilot and Bing's AI summaries cite your pages. - **Google:** Search Console, where AI Overviews and AI Mode are included in Web totals but not reported separately. ### What we will never publish Rankings in a tool that claims to show "your position in ChatGPT". Screenshots without dates. Results from prompts chosen after the fact. Numbers we cannot show the source for. To see where your brand stands today, start with an [AI visibility audit](/geo-agency/ai-visibility-audit/). For the full service, see the [GEO agency](/geo-agency/) page and [ChatGPT SEO](/geo-agency/chatgpt-seo/). ## Documented SEO case studies ## Questions about AI search results ## Start with a dated baseline Every result on this page will start the same way: an audit that records where a brand stood before the work began. [Get an AI visibility audit](/geo-agency/ai-visibility-audit/) ## Frequently asked questions ### Why are there no client AI screenshots on this page yet? Because we only publish results that meet our evidence standard and that the client has approved in writing. A dated baseline, a fixed prompt set and several months of re-runs take time, and many brands prefer not to be named. Rather than show invented examples, we publish the method now and add results as they are approved. ### How do you prove a brand is more visible in ChatGPT? By counting, not by showing one answer. We run a fixed set of buyer prompts before work starts and re-run the same set monthly in the same markets. The report shows how many prompts mention or cite the brand at each point, against the same competitors, with dated screenshots for every answer. ### Can you show revenue from AI search? Partly. Shopify analytics and GA4 show sessions and orders referred by ChatGPT, Perplexity, Gemini, Copilot and others, and ChatGPT links usually carry a utm_source tag. Some AI visits arrive with no referrer and some buyers research in an assistant and then search your brand on Google, so AI referral revenue is a floor, not the full picture. ### Are AI visibility tracking tools accurate? They are useful for trends and not precise for single answers. Assistants personalize results by location, history and session, and change often. Any tool, including our own tracking, is sampling. We trust results that hold across many prompts and several re-runs, and we treat a single-day jump or drop with suspicion.