A New York footwear brand was losing branded demand to competitors’ PPC campaigns and zero-click SERPs, while remaining largely invisible to AI answer engines except for its own name.
By rebuilding the site’s schema footprint, agent accessibility, and feed depth, we grew total AI citations from 2.4K to 5K quarter over quarter, opened up 316 citations on non-brand queries the site had never been cited for, and lifted organic shopping revenue by 119.34%.
Situation
A luxury women’s footwear store was recording a progressive dip in the volume of traffic emanating from its branded keywords.
- Competitor PPC campaigns were bidding against brand-modified versions of the store’s primary keywords, absorbing clicks that had historically been won organically.
- Zero-click searches and AI-generated answers were resolving product and informational queries in the results interface, reducing the clicks available per search.
- The brand’s presence inside AI responses was almost entirely name-driven. Descriptive, material, and product-terminology queries returned no citations at all, which removed the site from consideration at the exact moment of discovery.
The net effect was a shrinking pool of brand-led organic demand, with no compensating visibility in the AI and shopping surfaces that were consuming it.
Task
The objective was to rebuild discoverability on surfaces the brand did not previously occupy, and to make the site legible to the agents and models mediating that discovery. Specifically, the initiative aimed to:
(a) Ensure increased mentions in AI overviews and answer engine responses
(b) Improve shopping surface visibility across Google and Microsoft properties
(c) Improve referral traffic from AI engines such as ChatGPT, Gemini, and others
Success was to be measured by AI citation volume, cited page depth, and the revenue and engagement quality of the organic shopping channel.
Action
With these objectives in mind, we then took the following steps:
(i) We expanded the site’s schema footprint in both quantity and quality of attributes, introducing carousel and web page schema across all collection pages, while embedding keyword attributes to expand the semantic relevance of those pages.
(ii) We improved the merchant listings component of the product schema by providing detailed shipping information and merchant return policy data, so that the inventory could be parsed and compared without the model having to leave the markup.
(iii) At the robots.txt level, we introduced explicit accessibility directives for principal LLM training bots and crawlers to enhance the agentic friendliness of the site.
(iv) We reduced the content discrepancy between the raw and rendered versions of the site’s HTML, ensuring that crawlers operating without a rendering budget still received the full commercial payload of each page.
(v) For enhanced ingestion of product inventory and pricing, we uploaded all products to the Google Business Profile, while adding conversational attributes to the product feeds being served to Google and Microsoft Merchant Center.
(vi) To enhance provenance signals, we added publication and last update dates to the blog pages, raised the entity salience scores of target keywords across all product descriptions, and appended lastmod values in the XML sitemap.
Result
The interventions delivered measurable gains across both AI visibility and shopping performance:
(1) Total citations rose 108.33%, from 2.4K in Q1 2026 to 5K in Q2, as reported by Bing Webmaster Tools across Microsoft Copilots and Partners. Average cited pages per response also moved from 10 to 11, indicating that the citation growth came from more of the site being surfaced, not from a single page being reused.
(2) The site gained 316 citations across five non-brand queries for which it was previously invisible in AI responses.
These were not brand variants but discovery-stage terms:
| Grounding query | Prior citations | Citations gained |
|---|---|---|
| sacchetto | 0 | 186 |
| sandals that wrap around ankle | 0 | 62 |
| wrap around sandals | 0 | 31 |
| black crinkle patent shoes accessories | 0 | 20 |
| espresso patent flats shoes | 0 | 17 |
The distribution matters more than the total. One term is product terminology, two are style descriptors, and two are material and colourway queries, which together represent the unbranded intent the store had been ceding to competitors.
(3) Organic shopping revenue and traffic recorded compounding gains quarter over quarter:
- Revenue: +119.34%
- Total users: +33.12%
- New users: +35.26%
- Returning users: +39.02%
- Sessions: +44.37%
- Views: +48.24%
- Average session duration: +78.94%
- Average purchase revenue: +9.39%
Revenue growing at roughly three times the rate of user growth, alongside a 78.94% increase in session duration and a higher average purchase value, indicates that the traffic arriving through these surfaces was not simply larger but better qualified.
Collectively, these gains confirmed that the schema, accessibility, and feed interventions had restored the store’s discoverability at the point of intent and reduced its dependence on brand-led search in an environment where branded clicks are increasingly contested by paid media and answer engines.