AI search
What did 344k AI-result impressions teach us on Moolwan?
That AI visibility follows from pages that answer real questions and from clean product data, and that it has to be measured separately. Moolwan earned 344k impressions in Google's AI results in three months, two guides carried thousands each, and chatgpt.com sent 368 new users. Impressions are visibility, not sales.
344k impressions in Google's AI results for MoolwanSearch Console · Generative AI features · 25 Jun–23 Sep 2026 · real data
What you'll see in 10 minutes
- Whether the brand appears in Google AI results, ChatGPT and Perplexity for its category questions
- A site and SEO audit with findings ranked by severity
- Draft SEO articles and schema for the team to review before publishing
Why are we writing about our own brand?
Agencies ask us the same question about AI search that brands ask them: is any of this real, or is it a new name for SEO? We run Moolwan, a Bengaluru D2C home décor brand, and ai-dm is built and run on it every day. So instead of telling you what AI search might do, here's what we can see on our own brand, where each number comes from, and what we don't know yet.
What does it cost to ignore this?
If an agency reports only clicks and sessions, AI visibility doesn't appear anywhere. The brand sees flat organic traffic and concludes that content isn't working, while its pages are being shown inside Google's AI answers. The agency loses credit for work that is doing something, and the brand may cut the content programme that produced it.
What do the numbers show?
Every figure below is from Moolwan's own accounts.
| What we see | Number | Source |
|---|---|---|
| Impressions in Google's AI results | 344k | Search Console · Generative AI features · 25 Jun–23 Sep 2026 |
| Impressions in AI results for our two strongest guides | 8,479 and 7,853 | Search Console · Generative AI features · 25 Jun–23 Sep 2026 |
| Products eligible for rich results | 1,400; 24 of 1,424 flagged | Search Console · Product snippets · 26 Sep 2026 |
| Valid merchant listings | 1,380 valid, 1 invalid | Search Console · Merchant listings · 26 Sep 2026 |
| New users from chatgpt.com | 368 | GA4 · First user source · 29 Jun–26 Sep 2026 |
| States with active users | 8 | GA4 · Region · 29 Jun–26 Sep 2026 |
Swipe the table sideways to see every column.
The Search Console and GA4 windows differ by a few days, so read them side by side, not as one funnel.
Guides carry a visible share
Two guides alone account for 8,479 and 7,853 AI-result impressions. Both answer a specific buyer question rather than describing a product range. The articles and their schema were made with the same SEO articles and schema workflow that agencies get in ai-dm.
Product data is mostly clean
1,400 products are eligible for rich results, with 24 of 1,424 flagged, and 1,380 merchant listings are valid with 1 invalid. We kept product data clean so these checks pass. That matters for rich results and merchant listings, which are separate from AI answers.
Some AI traffic shows up in GA4
GA4 records 368 new users whose first source was chatgpt.com. That's a different surface from Google's AI results, and it's the part of AI search we can actually see arriving on the site.
What do we think helped?
This is our reading, not a controlled test.
- Pages that answer one question well. The guides that lead are the ones written around a question a buyer would type, answered early and in plain text.
- Schema that matches the page. Google's page on AI features and your website says there's no special markup for AI features, and that structured data should match the visible text. We think consistency helped more than any extra markup.
- Product data with few errors. Clean product data keeps rich results and merchant listings healthy. Whether it also affects AI answers, we can't show.
- Watching it. We track AI impressions as their own line, so we notice which pages earn them.
What don't we know yet?
- Impressions are not sales. Google's Generative AI performance report shows impressions by page, country, device and date. It doesn't show clicks or queries, so we can't tie the 344k to orders.
- Which change caused what. The guides, the schema and the product data were all in place over the same window, so we can't separate their effects.
- Whether it lasts. AI answers change as models and indexes change. Three months is a window, not a trend.
- What the 368 users went on to do. ai-dm doesn't do attribution modelling, and we won't claim a revenue figure we can't trace.
What should agencies take from this?
- Report AI visibility separately. If it isn't in the report, the brand can't value it.
- Write for questions, not catalogues. Our strongest AI-result pages are guides, not collection pages.
- Treat product data as hygiene. A short list of flagged items is easier to fix than to explain.
- Label confidence. Say "what we see" and "what we think helped" separately. Brands trust an agency that says what it doesn't know.
- Keep it light to run. One person runs Moolwan's nine marketing channels in about an hour a day. The monitoring and drafting are done by ai-dm; the person approves what goes out.
What should you do with a brand?
Open the brand's Search Console and check the Generative AI performance report for the last three months. Note the top pages, then check whether they're guides or product pages. Run the Product snippets and Merchant listings reports and list the flagged items. Check GA4 first-user source for chatgpt.com and other AI assistants. Put all of it on one page with the source next to each number, as in the table above, and agree with the brand which questions to write for next. The AI search guide for D2C brands covers the content side step by step.
How does ai-dm show this?
AI search visibility monitoring tracks whether a brand appears in Google AI results, ChatGPT and Perplexity for its category questions. The Moolwan case study explains how we run the brand, and Moolwan's live numbers show the figures as they stand. ai-dm drafts the articles and schema; the team reviews and approves them before anything is published.
Related pages
More posts
Meta ads stopped converting: what changed?
When a brand's Meta ads stop converting, check in this order: tracking, attribution settings, learning phase, fatigue, auction, landing page, season.
Read the post →Google Ads vs GA4 ROAS: why don't they match?
Attribution, counting, dates, import lag and modelling make Google Ads and GA4 ROAS differ. Which to use for which decision, and what to tell the brand.
Read the post →
Questions agencies ask
Did the 344k AI-result impressions turn into sales?
We can't say. Search Console's Generative AI report counts impressions, not clicks or orders, so we report it as visibility and don't claim revenue from it.
What do we think helped Moolwan appear in AI results?
Guides that answer specific buyer questions, made with the same article and schema workflow agencies get in ai-dm, and product data kept clean enough to pass rich-result and merchant-listing checks. That's our reading, not a proven cause.
How many people does it take to run this?
One person runs Moolwan's nine marketing channels in about an hour a day, with ai-dm doing the monitoring and drafting and that person approving what goes out.
No plan to pick. Billed in USD and INR. A brand's data is deletable on request.