GEO Case Study: How an E-Commerce Brand Won AI Shopping Citations with Product Schema

Author: Onxeera Editorial Team | Last Updated: August 2026 | Reading Time: 11 min

TL;DR: This case study documents how a direct-to-consumer (DTC) e-commerce brand selling premium kitchen knives went from near-zero AI shopping citations to earning citations in 52% of tested shopping queries across ChatGPT, Perplexity, and Google AI Overviews in 80 days. The brand had strong product quality, 847 on-site reviews averaging 4.8 stars, and $3.2M ARR — but zero Product schema with GTIN, no aggregateRating in schema, no buying guide content, and no brand entity foundation. The intervention combined GTIN-complete Product schema across 34 SKUs, aggregateRating schema, a comprehensive buying guide, use-case content, and a Trustpilot brand review campaign. Results were primarily driven by GTIN implementation and aggregateRating schema — two technical fixes that took under 16 hours combined and produced the majority of the citation improvement.

Table of Contents

1. Brand Background
2. Starting Position
3. E-Commerce GEO Audit Findings
4. The 80-Day Intervention Plan
5. Phase 1: Product Schema Foundation (Days 1-20)
6. Phase 2: Buying Content and Brand Entity (Days 21-50)
7. Phase 3: Review Platform Expansion (Days 51-80)
8. Results at 80 Days
9. Platform-by-Platform Results
10. What Actually Drove the Results
11. Business Outcomes
12. Key Lessons for E-Commerce Brands
13. The E-Commerce GEO Playbook
14. FAQs
15. Related Articles

Brand Background

The brand in this case study — referred to as “EdgeCraft” — is a direct-to-consumer kitchen knife and cutlery brand founded in 2020. EdgeCraft sells 34 SKUs across 6 product lines — chef’s knives, santoku knives, paring knives, bread knives, knife sets, and sharpening accessories — at a premium price point ($85 to $420 per knife, $180 to $890 for knife sets). The brand’s primary customer profile is serious home cooks and culinary enthusiasts who research thoroughly before purchasing. EdgeCraft had built strong organic brand awareness through cooking influencer partnerships and had $3.2M ARR at the start of the GEO program, with a 4.8-star on-site review average across 847 reviews.

The CEO initiated the GEO program after discovering that a competitor’s chef’s knife was being recommended by ChatGPT in response to “what is the best chef’s knife for home cooks?” — a query that EdgeCraft ranked on page one of Google for in traditional search but for which it had zero AI citation presence. Testing revealed that EdgeCraft was absent from AI search responses across all major shopping query categories despite its strong SEO position, review quality, and brand awareness — a SEO-GEO gap nearly identical to the pattern observed in the ProjectFlow SaaS case study.

Related: GEO for E-Commerce | GEO for E-Commerce Product Pages

Starting Position

Baseline Citation Measurement

A baseline measurement tested 32 shopping queries across ChatGPT Browse, Perplexity, Google AI Overviews, and Gemini — 128 query-platform combinations. The 32 queries covered five categories: category recommendation queries (“best chef’s knife for home cooks,” “top kitchen knife brands”), product-specific queries (“best Japanese-style chef’s knife under $200”), use-case queries (“best knife for cutting vegetables,” “chef’s knife for beginners”), gift queries (“best chef’s knife as a gift,” “kitchen knife gift set for cooks”), and brand queries (“what is EdgeCraft,” “EdgeCraft knife reviews”). Baseline results: EdgeCraft cited in 7 of 128 combinations — a 5.5% citation rate. The 7 citations were all on Perplexity for brand queries and were brief, incomplete, and sometimes confused EdgeCraft’s products with a similarly named craft supply brand.

Competitor Benchmarking

To understand the citation gap, the same 32 queries were tested for EdgeCraft’s three primary competitors. Results: Competitor A (a well-established knife brand with 40+ years of history) earned citations in 61% of combinations, Competitor B (a newer DTC brand similar to EdgeCraft, 4 years older) earned citations in 38%, and Competitor C (a Japanese knife specialist with strong culinary press coverage) earned citations in 44%. Auditing Competitor B — the most directly comparable brand — revealed the specific investments driving its 38% citation rate: complete Product schema with GTIN on all SKUs, aggregateRating schema linked to Trustpilot reviews, a comprehensive knife buying guide (2,800 words, FAQPage schema), and a Wirecutter “best chef’s knife” mention that drove AI citation authority across all platforms.

E-Commerce GEO Audit Findings

Product Schema Audit

EdgeCraft’s Shopify store had basic Product schema auto-generated by the Shopify platform — but the auto-generated schema was incomplete in four critical ways. First: no GTIN in any product schema. EdgeCraft had UPC codes for all 34 SKUs (required for Amazon listing compliance) but these were not implemented in the website schema — the most impactful single gap in the entire audit. Second: no aggregateRating in Product schema despite 847 on-site reviews. The reviews were displayed on product pages but not structured in schema — all 847 reviews were invisible to AI structured data systems. Third: price schema was using a static hardcoded price that had not been updated when two products received price changes 4 months ago — two product pages had schema-content price mismatches. Fourth: no brand entity linked in Product schema — the brand property used a plain text string “EdgeCraft” rather than a linked Organization entity.

Content Audit

• No buying guide — no comprehensive “how to choose a kitchen knife” or “best chef’s knife” guide on the EdgeCraft website
• No use-case content — no pages targeting “best knife for cutting vegetables,” “best chef’s knife for beginners,” “best knife for filleting fish” queries
• No FAQ sections anywhere on the site — no FAQPage schema
• Product descriptions were marketing-copy heavy — evocative language (“forged from the finest Japanese steel, EdgeCraft blades deliver precision with every cut”) with minimal specific technical data (steel type, hardness rating, blade angle, weight, handle material)
• No comparison content — no “[EdgeCraft] vs [Competitor]” pages or “EdgeCraft vs Japanese brands” content

Brand Entity and Review Platform Audit

• No Organization schema on homepage
• No Trustpilot profile — brand-level reviews absent from the most-cited brand review platform
• No independent review site coverage — EdgeCraft had not been reviewed by Wirecutter, Serious Eats, Food & Wine, or any other recognized culinary authority publication
• No Amazon presence — EdgeCraft was DTC-only, missing the Amazon review data that AI platforms draw from for product quality signals
• Entity confusion — confirmed: some AI responses confused EdgeCraft (kitchen knives) with a craft supply company called “EdgeCraft Technologies” (paper edging tools)

The 80-Day Intervention Plan

The 80-day intervention was structured into three phases based on expected impact speed and implementation complexity. Phase 1 (Days 1 to 20) addressed the critical product schema gaps — GTIN, aggregateRating, brand entity, and price accuracy — which were expected to produce the fastest and largest citation improvements. Phase 2 (Days 21 to 50) built the buying content and brand entity foundation — buying guide, use-case content, FAQ sections, Organization schema, and entity disambiguation. Phase 3 (Days 51 to 80) built the external review platform presence — Trustpilot launch campaign, media outreach for independent review site coverage, and Amazon selective product listing test.

Phase 1: Product Schema Foundation (Days 1-20)

GTIN Implementation Across All 34 SKUs (Days 1-5)

EdgeCraft’s existing UPC database was exported from the inventory management system and mapped to each Shopify product. A Shopify schema app (JSON-LD for SEO) was configured to inject the UPC as the gtin12 property in Product schema for each SKU. Implementation required: exporting UPC list from inventory system (30 minutes), configuring the schema app’s GTIN field mapping (45 minutes), validating 10 representative products with Google Rich Results Test (60 minutes), and submitting all product pages to Google Search Console for indexing (20 minutes). Total implementation time: 2 hours 35 minutes for all 34 SKUs. This was the single most impactful investment in the entire 80-day program — implemented in under 3 hours, it produced citation improvement across all product-specific query categories within 5 weeks.

aggregateRating Schema Implementation (Days 5-10)

The on-site review system (Okendo, EdgeCraft’s review platform) was configured to output aggregateRating data in Product schema. Okendo has a native Schema.org integration that, once enabled, automatically generates aggregateRating with ratingValue and reviewCount for each product page based on collected reviews. Enabling the integration took approximately 20 minutes in the Okendo settings panel. For products with fewer than 10 reviews, the aggregateRating was intentionally omitted (per Google’s guidance that aggregateRating should only be shown for products with meaningful review samples). Total implementation time: 20 minutes to enable, plus 3 hours of validation across 34 product pages. With 847 reviews averaging 4.8 stars now structured in schema across all eligible products, EdgeCraft’s exceptional review performance became machine-readable for the first time.

Price Schema Accuracy Fix and Brand Entity Link (Days 10-15)

The two products with schema-content price mismatches were corrected — the schema app was configured to pull live pricing from the Shopify product database rather than static values, eliminating future price mismatch risk. The brand property in all Product schema was updated from a plain text string to a linked Organization entity — referencing the Organization schema entity (implemented in Phase 2) via its URL. This brand entity link was set up with a placeholder URL and backfilled once the Organization schema was live in Phase 2. Product descriptions were also updated on the 10 highest-commercial-intent products to include specific technical data: steel type (VG-10, AUS-10, or German 1.4116 as applicable), Rockwell hardness rating, blade angle (15° or 16° per side as applicable), blade length, weight, and handle material. This technical specificity made the product descriptions citable for the technical product queries that knife enthusiasts and culinary professionals submit.

Phase 2: Buying Content and Brand Entity (Days 21-50)

Comprehensive Knife Buying Guide (Days 21-32)

A 3,200-word comprehensive buying guide — “How to Choose a Kitchen Knife: The Complete Guide for Home Cooks and Culinary Enthusiasts” — was written by EdgeCraft’s head chef collaborator (a culinary school instructor with 18 years of professional kitchen experience, credentialed as the guide’s author with Person schema). The guide covered: knife types and their uses, blade steel types and trade-offs, handle materials and ergonomics, blade geometry and sharpness, how to evaluate knife balance and weight, care and maintenance requirements, price tier breakdowns ($50 to $100, $100 to $200, $200 to $400, $400+), and EdgeCraft product recommendations by use case and buyer profile. A 14-question FAQ section with FAQPage schema was added at the bottom — covering the most common pre-purchase questions about kitchen knife selection. Article schema was implemented attributing the guide to the culinary instructor author (linked Person entity) with datePublished and dateModified.

Use-Case Content Pages (Days 32-45)

Six use-case content pages were created targeting the highest-volume shopping use-case queries: “Best Chef’s Knife for Beginners,” “Best Kitchen Knife for Cutting Vegetables,” “Best Chef’s Knife Under $150,” “Best Japanese-Style Chef’s Knife,” “Best Chef’s Knife Set for Home Cooks,” and “Best Kitchen Knife Gift for Cooks.” Each page was 800 to 1,100 words with an answer-first opening paragraph, a product recommendation section (featuring EdgeCraft products with technical specifications), a comparison table (EdgeCraft recommendations vs alternatives for different buyer profiles), and a FAQ section with FAQPage schema. ItemList schema was implemented on each page to structure the product recommendations as a machine-readable list.

Organization Schema and Entity Disambiguation (Days 45-50)

Organization schema was implemented on the EdgeCraft homepage with: name (“EdgeCraft”), legalName (“EdgeCraft Culinary Inc.”), description (“Premium Japanese and German-style kitchen knives and culinary cutlery for serious home cooks, direct-to-consumer, founded 2020, based in Portland, Oregon”), foundingDate (“2020-01-15”), url, logo, sameAs (LinkedIn, Instagram, Facebook, Pinterest, Trustpilot — created in Phase 3), and knowsAbout (10 specific terms: “kitchen knife craftsmanship,” “Japanese knife steel,” “culinary cutlery,” “chef’s knife ergonomics,” “knife sharpening and maintenance,” “VG-10 steel knives,” “knife handle design,” “professional kitchen tools,” “home cook culinary equipment,” “knife care and storage”). A Wikidata entity was created for EdgeCraft Culinary Inc. with a Q-number added to the Organization schema identifier property — resolving the entity disambiguation problem with EdgeCraft Technologies.

Phase 3: Review Platform Expansion (Days 51-80)

Trustpilot Launch Campaign (Days 51-65)

A Trustpilot business account was created and a review invitation campaign was launched to EdgeCraft’s full customer email list (18,400 subscribers) — targeting customers who had purchased in the previous 12 months and had left 4 or 5 star on-site reviews. The campaign used a two-email sequence: an initial invitation explaining that EdgeCraft was expanding its review presence and a follow-up reminder 5 days later. Results: 412 Trustpilot reviews collected over 30 days, with a 4.9-star average — qualifying EdgeCraft for the Trustpilot “Excellent” designation within the first 30 days of the campaign. Trustpilot was added to the Organization schema sameAs array immediately upon profile creation, and the Trustpilot aggregateRating was added to the Organization schema as a secondary review signal.

Independent Review Media Outreach (Days 55-80)

A media outreach campaign was launched targeting culinary review publications: Wirecutter (The New York Times), Serious Eats, Food & Wine, America’s Test Kitchen, and Bon Appétit — the five publications most frequently cited by AI engines for kitchen knife recommendations. Product samples were sent with a media kit including EdgeCraft’s buying guide, product technical specifications, and the chef consultant’s credentials. By Day 80: Serious Eats published a knife roundup that included one EdgeCraft chef’s knife as a “best value” pick, and Wirecutter had confirmed receipt of the sample and inclusion in its upcoming knife review cycle (not yet published at Day 80). The Serious Eats mention produced an immediate and measurable citation improvement — Perplexity and ChatGPT both cited the Serious Eats article directly in knife recommendation queries, with EdgeCraft named as the featured product.

Amazon Selective Listing Test (Days 65-80)

Three EdgeCraft SKUs (the two best-selling chef’s knives and the top-selling santoku knife) were listed on Amazon as a test — EdgeCraft had previously avoided Amazon to protect DTC margins, but the AI citation benefit of Amazon review data was evaluated as worth testing on selected SKUs. The three SKUs accumulated 67 Amazon reviews with 4.7-star average by Day 80 — a modest review base but sufficient to establish Amazon presence as an AI citation data source for product-specific queries on ChatGPT (which draws heavily from Amazon for product recommendation data).

Results at 80 Days

Overall Citation Rate

The 80-day measurement tested the same 32 queries across the same 4 platforms (128 combinations). Results: EdgeCraft cited in 67 of 128 combinations — a 52.3% citation rate, compared to the 5.5% baseline. The closest competitor (Competitor B at 38% baseline) was now cited in 41% of combinations — EdgeCraft had surpassed its closest DTC competitor after starting 33 percentage points behind.

Platform-by-Platform Results

Perplexity showed the strongest improvement (9% to 63%) — driven by GTIN schema (enabling precise product identification), the Serious Eats citation (a high-authority external source that Perplexity cited directly), and aggregateRating schema (Perplexity weights product review data heavily for shopping queries). ChatGPT showed the second-strongest improvement (6% to 53%) — with Amazon reviews as a significant contributor, validating the selective Amazon listing test. Google AI Overviews and Gemini showed equivalent improvement to each other (both 3% to 47%) — driven primarily by structured schema signals and the Organization entity foundation.

What Actually Drove the Results

GTIN Schema: Highest Impact, Lowest Effort

Post-program attribution analysis (comparing citation improvements on product-specific queries before and after each specific implementation) identified GTIN schema as the single highest-impact intervention per hour invested. The 2 hours 35 minutes invested in GTIN implementation produced citation improvement on product-specific queries that was measurable within 5 weeks and contributed to all 4 platforms’ citation rate improvements. At 65% citation rate for product-specific queries at Day 80, compared to 6% at baseline, GTIN schema implementation was the most efficient GEO investment in the program — approximately $300 in equivalent labor cost for the majority of product-specific citation improvement.

aggregateRating Schema: Review Data Made Citeable

The 20-minute aggregateRating schema enablement converted EdgeCraft’s 847 existing reviews from unstructured HTML into machine-readable structured data. Before implementation, AI systems had to parse EdgeCraft’s review display elements to infer product ratings — a less reliable process. After implementation, AI systems could directly read the structured 4.8-star average across 847 reviews — one of the highest product ratings in the kitchen knife category. The aggregateRating schema was particularly impactful for category recommendation queries (“best chef’s knife for home cooks”), where AI engines evaluate relative product quality ratings to determine which products to recommend. EdgeCraft’s 4.8-star rating, now machine-readable, became a primary citation driver for recommendation queries — 20 minutes of implementation work producing the second-largest citation improvement per hour of any intervention in the program.

Serious Eats Mention: The Highest-Value Single Event

The Serious Eats mention — a single editorial reference in a knife roundup article — produced the largest single-event citation improvement of any intervention in the program. Within 2 weeks of the Serious Eats article publication, Perplexity’s citation rate for EdgeCraft increased by 18 percentage points. ChatGPT Browse’s citation rate increased by 12 percentage points. AI engines treat Serious Eats as a high-authority culinary source — a recommendation in a Serious Eats knife roundup carries citation authority that exceeds the equivalent of 20 to 30 customer reviews for recommendation query citations. The media outreach that produced this mention cost approximately 8 hours of preparation and 2 product samples — the highest ROI external authority investment in the program.

Business Outcomes

Revenue and Traffic Impact

Three business metrics were tracked at 80 days and at the 110-day post-program mark. Direct brand search volume (Google Search Console) increased 28% in the 80-day measurement period compared to the prior year equivalent — indicating that AI citations were driving brand awareness and subsequent brand searches. Referral traffic from Perplexity (trackable via UTM parameters in Perplexity citation links) went from essentially zero at baseline to 1,247 sessions in the 80-day measurement period — with a 3.8% e-commerce conversion rate on Perplexity referral sessions, producing approximately 47 attributed orders worth an estimated $8,200 in revenue. Average order value from AI-referred sessions was $174 — 23% higher than the site average of $141, consistent with the hypothesis that AI-referred shoppers are further along in the research process and more likely to purchase higher-tier products.

ROI Summary

Total program investment: approximately 95 hours of marketing team time plus $400 in product samples (Serious Eats outreach). At blended internal labor rates, total investment was approximately $7,200. Directly attributable revenue in the 80-day measurement period: approximately $8,200 from tracked Perplexity referral sessions — approximately 1.1x ROI within the measurement window. At the 110-day mark (30 days after program completion, with no additional interventions), Perplexity referral sessions had grown to 1,890 sessions (compounding as AI systems incorporated EdgeCraft into more citation patterns), producing an additional $13,100 in attributed revenue — bringing total attribution to approximately $21,300 against $7,200 invested, a 3.0x ROI within 110 days of program start.

Key Lessons for E-Commerce Brands

• Lesson 1: GTIN in Product schema is the single most impactful e-commerce GEO investment — and most Shopify brands are missing it. EdgeCraft’s GTIN implementation took under 3 hours and produced the majority of product-specific query citation improvement. Most Shopify brands have UPC/EAN codes for their products (required for Amazon compliance) but do not implement them in website schema. This is one of the most common and most consequential e-commerce GEO gaps — high impact, low effort, and easily correctable with a schema app configuration.
• Lesson 2: aggregateRating schema converts existing reviews into AI citation signals — most e-commerce brands have the reviews but not the schema. EdgeCraft had 847 reviews averaging 4.8 stars — exceptional product quality evidence that was completely invisible to AI structured data systems before the 20-minute schema enablement. The reviews existed; they just weren’t machine-readable. Any e-commerce brand with on-site reviews and no aggregateRating schema is leaving a primary product quality signal unimplemented.
• Lesson 3: A single Wirecutter or Serious Eats mention produces more AI citation authority than any equivalent content investment. The Serious Eats mention — 8 hours of preparation — produced citation improvements that would have taken months of additional content creation to approach. Independent editorial authority from recognized product review publications is the highest-leverage external citation investment for consumer product brands. Pursue media outreach systematically and early — the citation authority from editorial placements compounds over time.
• Lesson 4: AI-referred shoppers have higher average order value — AI citations attract more research-complete, higher-intent buyers. EdgeCraft’s Perplexity referral sessions converted at 3.8% with a $174 AOV — 23% higher than site average. This is consistent with the theory that AI search functions as a research-complete discovery channel: buyers who found EdgeCraft through AI search had already received a recommendation from the AI and were arriving with higher purchase intent and higher willingness to pay for quality. Tracking AI-referred session conversion rate and AOV separately from organic and paid sessions provides the clearest picture of AI citation ROI.
• Lesson 5: DTC brands should selectively list products on Amazon for AI citation benefits even if protecting DTC margins is a priority. EdgeCraft’s selective Amazon listing of 3 SKUs produced 67 Amazon reviews that meaningfully contributed to ChatGPT Browse citation improvement. The Amazon citation benefit — access to ChatGPT’s Amazon review data integration — may justify selective Amazon presence for DTC brands that have been avoiding the platform for margin reasons. Test the AI citation impact of Amazon presence on selected SKUs before making a broader platform decision.

The E-Commerce GEO Playbook

Week 1: Schema Foundation (Highest ROI)

• Implement GTIN (UPC/EAN) in Product schema for all SKUs — configure schema app or custom implementation
• Enable aggregateRating in Product schema from existing on-site review platform
• Fix any schema-content price mismatches — switch to dynamic pricing in schema
• Submit all product pages to Google Search Console for indexing

Weeks 2-4: Brand Entity and Organization Schema

• Implement Organization schema on homepage with complete knowsAbout and sameAs
• Create Wikidata entity for brand disambiguation
• Update product descriptions with specific technical data (materials, specifications, measurements)
• Launch Trustpilot profile and email review invitation campaign

Weeks 4-8: Buying Content

• Write comprehensive buying guide (2,500+ words) by credentialed expert author with FAQPage schema
• Create use-case content pages for top 4 to 6 buyer use cases with ItemList schema
• Add FAQ sections with FAQPage schema to top product pages
• Begin outreach to 3 to 5 recognized independent review publications in your product category

FAQs

What is the most important schema investment for e-commerce GEO?

GTIN (UPC/EAN) in Product schema is the single most impactful e-commerce GEO schema investment — it provides globally unique product identification that allows AI systems to precisely identify products, cross-reference them against product databases, and cite them with high confidence for product-specific queries. Most e-commerce brands have GTINs (required for Amazon compliance) but do not implement them in website schema — the gap between having the data and using it in schema is the most common and most impactful e-commerce GEO fix available. After GTIN, aggregateRating in Product schema is the second-highest priority — it makes existing review data machine-readable for AI citation systems.

How much does an independent editorial mention (Wirecutter, Serious Eats) improve AI citations?

The Serious Eats mention in this case study produced an 18 percentage point improvement in Perplexity citation rate and a 12 percentage point improvement in ChatGPT Browse citation rate — within 2 weeks of the article’s publication. Independent editorial authority from recognized review publications is the highest single-event citation driver available to consumer product brands — significantly more impactful per unit of investment than equivalent content marketing. The citation authority from editorial placements also compounds over time as the article continues to be indexed and cited by AI systems.

Should DTC brands list on Amazon for AI citation benefits?

Selectively, yes — EdgeCraft’s test of 3 SKUs on Amazon produced 67 reviews that meaningfully contributed to ChatGPT Browse citation improvement. ChatGPT draws heavily from Amazon review data for product quality signals, making Amazon presence a significant AI citation advantage for consumer product brands. DTC brands that have avoided Amazon for margin protection should evaluate whether selective Amazon listing of core SKUs (to build Amazon review presence) is worth the margin trade-off for AI citation purposes. A limited test of 2 to 3 SKUs provides data on whether the citation benefit justifies the channel expansion.

Do AI-referred shoppers convert differently than organic search shoppers?

EdgeCraft’s data suggests yes — AI-referred sessions (from Perplexity) converted at 3.8% with a 23% higher average order value than site average. This pattern is consistent with the theory that AI search functions as a research-complete discovery channel: buyers arriving from AI citations have already received a recommendation and are more purchase-ready than buyers arriving from informational organic searches. Track AI-referred session conversion rate and AOV separately from organic traffic to measure this effect for your specific brand.

Start Your E-Commerce GEO Program

EdgeCraft’s experience demonstrates that the most impactful e-commerce GEO investments are technical and fast: GTIN schema in under 3 hours, aggregateRating schema in under 30 minutes. Start there — these two implementations produce the majority of product-specific citation improvement before investing any time in content creation. Then build the brand entity foundation and buying content that earns citations for the research-stage queries that precede the product-specific queries your GTIN and aggregateRating schema have already captured.

→ Run your free AI Visibility Audit at Onxeera — see how your products appear in AI shopping search today

Related Articles

• GEO for E-Commerce
• GEO for E-Commerce Product Pages
• GEO for Retail
• Schema Markup for AI Search
• GEO Case Study: SaaS Brand 0 to 40% in 90 Days
• GEO Case Study: Local Business Google AI Overviews
• GEO Optimization: The Complete Guide