Case studyClothing and outdoor retail
Name under NDA
CAD1m in additional search revenue for a Canadian clothing and outdoor retailer
- CAD1m
- +60%
- 2.5x
Who the client is
Client profile. The retailer sells clothing, footwear and outdoor products through four Canadian stores and an English/French website. Customers include everyday shoppers, hikers and recreational skiers. Before the engagement, annual company revenue was approximately CAD24m, including approximately CAD6m online.
- Four storesCAD18.0m
- WebsiteCAD5.9m
- Google organicCAD1.4m
- Paid search, direct, email and otherCAD4.4m
Catalog scope. At kickoff, each product model had a page in both languages, with sizes and colors held as SKU variants. Our work covered the opening catalog and new seasonal products as they arrived.
- 3,000
- 18,000
Engagement terms. We worked together from April 1, 2025 to March 31, 2026. The fee was EUR14,000 monthly across 12 payments: EUR168,000 total, or approximately CAD269,000 at the engagement-period average exchange rate.
What they came to us with
Business goals. The retailer wanted more annual online revenue from search, priority pages ready before winter and useful guidance maintained across the full assortment.
Search positioning. The team wanted recommendations for relevant shopping needs and accurate AI answers about products, delivery, returns and support.
Team guidance. Ecommerce, merchandising and editorial staff needed shared criteria for choosing opportunities, approving content and interpreting results. They wanted strategic guidance alongside production.
What they got
Additional revenue. Google-organic net revenue in the engagement year exceeded our estimate for the previous approach. The chart compares both years; the comparison assumptions are explained under Business results.
AI comparison. Citation counts measured answers containing a link to the retailer’s website across 240 unbranded shopping prompts, split equally between English and French. Three repetitions across ChatGPT Search, Perplexity Search and Gemini produced 2,160 answers per snapshot.
Team capability. Category owners brought supplier and stock changes into planning. Reviewers used source-backed corrections, and the ecommerce lead used reporting to agree on publishing priorities with us.
The situation before the engagement
Business context
Everyday decisions. Buyers needed help choosing soft jacket linings, coats comfortable on heated trains, boots for slush and layers that managed sweat. Advice had to remain useful as seasonal demand and available stock changed.
Website and content
Existing inventory. The website had a product page for every model, 300 category and brand collections, 60 guides and 80 home, service and utility pages: 3,440 logical pages and 6,880 language URLs. Supplier descriptions varied in quality, fit and material fields were inconsistent, and guides received irregular updates.
Traffic sources
Acquisition mix. In the previous year, after refunds, Google organic search supplied about a quarter of online net revenue, as the revenue split above shows. Paid search and Shopping were another major source, alongside direct visits, email, other search engines and referrals.
Search presence
Discovery gaps. Search coverage favored named products over detailed buying comparisons. AI answers mentioned competitors more often and frequently omitted or misstated the retailer’s service policies.
Previous approach
Merchandising focus. The team concentrated on supplier descriptions and merchandising updates. Product information varied between brands and languages, making consistent buying guidance difficult to maintain across the assortment.
The strategy we planned and why
Commercial and search priorities
Retailer positioning. We focused on useful multi-brand advice and dependable service. Shoppers needed help choosing a product and understanding what happened if delivery was delayed, an item arrived damaged or a return was necessary.
Joint priorities. We worked with category owners to compare buyer demand, competitor coverage, stock depth and implementation effort. We chose the collections where missing buying guidance and sufficient stock justified editorial work. Shared technical fixes covered the wider catalog; comparisons and use-case guides followed the commercial priorities.
Technical SEO priorities
Discovery barriers. We prioritized the shared problems affecting product discovery and buying decisions:
- Make essential product information available in the initial HTML.
- Improve mobile rendering so shoppers could reach product details more quickly.
- Replace missing or duplicated titles and descriptions with clear product information.
- Correct product, offer and variant structured data, including shipping and returns fields.
- Fix discovery rules for filtered collections and English/French page versions.
Content strategy and production order
- Commercial content. Improve product and collection pages so shoppers can compare fit, materials and intended use.
- Buying guidance. Publish comparisons and guides for practical choices, including commuting layers and footwear for slushy conditions.
- Service explanations. Clarify delivery, returns and support procedures using approved policies.
- Content maintenance. Update both languages when supplier information or available stock changes.
Why the strategy fit
Commercial fit. The multi-brand assortment supported comparisons, while product specialists and customer service supplied the facts. Improving commercial pages first connected that expertise to products shoppers could buy; guides then addressed questions that supplier descriptions left unresolved.
What we did
Agency and client responsibilities
Our services. The engagement combined:
- SEO and AI Search Strategy.
- Technical SEO.
- AI Search Optimization.
- Content Research and Strategy.
- AI Content Production Systems.
Delivery roles. We handled strategy, technical specifications, system engineering, drafting, bilingual editing and reporting. The ecommerce lead approved publication, merchandising supplied records, product specialists checked claims, customer service approved policies and the developer released changes.
Weekly guidance. Weekly working sessions addressed implementation and editorial decisions. Monthly reviews with category owners connected search performance with stock and supplier changes. We taught the team how to read traffic reports, follow visits through to purchases and check how AI answers described the retailer.
- 1Research, strategy and technical fixes
- Buyer and competitor research
- Technical SEO audit
- Shared-template fixes
- Strategy and execution plan
- 2AI content production system
- Structured AI workflow
- Bilingual pilot
- Automated and human checks
- 3Ongoing production, measurement and improvements
- Catalog rollout
- Guides, collections and service pages
- Updates when stock changes
Weekly working sessions on implementation and editorial decisions, with monthly reviews
Phase 1: Research, strategy and technical fixes
April planning. We researched buyers, competitor coverage, content gaps and brand representation alongside a technical SEO audit. The team received a strategy and a 90-day execution plan, with implementation priorities explained to the ecommerce lead and developer.
Technical delivery. The developer released the shared-template fixes while we checked their implementation. We verified that essential product details appeared in the initial HTML, checked metadata and structured-data fields, and tested filtered collections and language versions. Technical monitoring continued during production.
Phase 2: Building the AI content production system
May build. We built a structured AI workflow using approved catalog and supplier records, field schemas, terminology rules, factual checks and CMS staging. Missing or conflicting specifications went to product specialists; commerce feeds controlled price and stock.
- 1Approved recordsCatalog and supplier data
- 2AI draftingSchemas and terminology rules
- 3Automated checksEvery output
- 4Human reviewFlagged records and samples
- 5CMS stagingEnglish and French
Missing or conflicting specifications went back to product specialists; commerce feeds controlled price and stock.
Bilingual pilot. The pilot covered 40 product models, eight collections and four existing guides, fully reviewed in both languages. Reviewers learned to distinguish factual corrections from wording preferences and improve source fields and generation rules.
Production checks. Automated checks covered every output. People fully reviewed flagged product records and checked samples across groups of unflagged records. Collections, guides and service pages received full review.
Phase 3: Ongoing production, measurement and improvements
Catalog rollout. Following the May pilot, we processed the opening catalog by September. New seasonal products entered the same workflow throughout the year. Approximately 900 models were added and 600 were retired.
- Added
- Retired
Numbers above the bars are the active models at each month end.
Publishing schedule. Guide and collection work continued through March, alongside catalog maintenance. The eight service pages explained delivery, delays, lost parcels, damaged items, exchanges, returns, refunds and warranty support. All content was produced in English and French.
120 new buying, comparison and use-case guides
60 guides refreshed
120 collections refreshed
8 service pages clarified
One square is about 10 pages, each in English and French.
Ongoing improvements. We reran changed catalog records, updated recommendations when stock changed and corrected outdated guidance. Category owners learned to bring supplier changes into planning, while reviewers supplied source-backed corrections. We continued operating the system and guiding priorities throughout the engagement.
Results
Google search and AI referrals
All queries, Search Console
Year before
Engagement year
Includes product-brand searches, Search Console
Year before
Engagement year
Identifiable external AI referrals, analytics
Year before
Engagement year
Buying intent. Commercial landings covered the homepage, products, category and brand collections, site search and transaction pages. Informational landings covered guides, service, company and legal pages. Informational traffic’s share rose from 21.0% to 22.4%. Commercial conversion improved from 2.74% to 3.17%, and informational conversion from 0.82% to 1.05%.
Other acquisition. Total paid-media spending increased from approximately CAD418,000 to CAD452,000. Within paid acquisition, search and Shopping revenue rose from CAD1.68m to CAD1.94m over the same annual periods. Search growth occurred alongside continued investment in other acquisition channels.
AI visibility and brand representation
March 2025March 2026
Answers linking to the retailerShare of all sampled answers
Share among six tracked retailersClient appearances among all retailer appearances
Positive answersAmong answers mentioning the client
Correct factual claimsAmong checked claims
Service-policy accuracy
Service accuracy. A separate panel used 30 English and 30 French prompts about delivery, returns and support. Three repetitions across the same engines produced 540 answers per snapshot. Correct answers rose from 151 to 451, or 28.0% to 83.5%; answers with materially wrong or unsupported policy claims fell from 83 to 27.
151 correct, 83 wrong, March 2025
451 correct, 27 wrong, March 2026
- Correct
- Missing required details
- Wrong or unsupported
One square is about 10 of the 540 answers in each snapshot.
Business results
Annual outcomes. Rounded results compare April 2024-March 2025 with April 2025-March 2026, using website analytics joined to commerce records. Google-organic figures cover consented Canadian sessions and same-session purchases by new and returning customers. Most of the traffic gain arrived in the second half of the engagement, following the catalog rollout.
Google-organic sessions
Sessions with a purchase
Purchase conversion
Net revenue per order
- Year before
- Gain in the engagement year
Comparison assumptions. We allowed 15% base traffic growth covering ordinary demand and assortment changes, adjusted to 11%-21% by month for disruption and recovery. We retained 25% of conversion improvements for routine trading outside the engagement. Both comparisons used the same monthly basket values, discounts, cancellations and refunds within commercial and informational traffic groups.
- Estimated with the previous approach
- Additional revenue
- Below the estimate
- Same month, year before
Most of the gain arrived from October, after the catalog rollout.
Estimated range. Testing 10%-35% base traffic growth and 0%-100% of conversion improvements from ordinary activity produced approximately CAD476,000-CAD1.17m in additional revenue. These judgment-based assumptions considered retail growth, postal interruptions, competitor closures and weather. Our selected estimate sits near the top of that range.
What the estimate covers
Joint delivery. The engagement included our strategy, production and technical guidance together with the client’s implementation and approvals. The comparison estimated revenue above the selected allowance for improvements from ordinary trading activity.
Guide contribution. Buying guides supported both immediate purchases and later consideration. The reported revenue counted purchases completed during the Google-organic visit. AI referrals were measured separately; their growth also reflected increasing use of AI shopping tools.
Conclusion
Business outcome. Google-organic revenue exceeded the selected comparison estimate by approximately CAD1.03m. Shoppers had clearer bilingual product information, buying guidance and service explanations across the clothing, footwear and outdoor assortment.
Team practices. Weekly guidance and review coaching gave category owners, editors and the ecommerce lead a shared way to choose priorities and check content. Those practices connected supplier and stock changes to the information shoppers found through search.
How we measured
Search measurement
Annual search. Rounded results compare April 2024-March 2025 with April 2025-March 2026. Search Console covers property-wide web-search clicks and impressions; analytics covers consented Canadian sessions. Counts and percentage changes are rounded independently from their underlying values.
AI answer measurement
AI sampling. Both March snapshots used the same 240 unbranded questions about shopping in Canada, producing 2,160 answers across the three engines. Share of voice used a fixed retailer set: the client, Altitude Sports, MEC, Sport Chek, Latulippe and Bushtukah. Each retailer counted once per answer; rates and changes are rounded independently.
Metric definitions. An appearance counted an answer naming the retailer; a citation counted an answer linking to its website. Automated scoring classified recommendations and favorable evaluations as positive sentiment. It checked factual claims against the approved business facts applicable to each snapshot to calculate precision.
Scoring method. Automated scoring used the same rubric and each snapshot's dated business facts and policies. Materially incorrect or unsupported policy assertions counted as wrong; omitted required details counted as missing. Correct answers supplied the applicable eligibility, deadlines, fees or next steps specified for that question.
Orders and revenue
Purchase definitions. Purchase conversion divided sessions containing an order by all reported consented Canadian Google-organic sessions. Multiple orders could occur in one session. Net revenue covered merchandise after discounts, cancellations and refunds, excluding tax and shipping. Online checkout orders, including store pickup, belonged to online revenue; store-till purchases belonged to store revenue.
Channel definitions. Both years used the same consent and transaction-tracking rules. Paid search, tagged Google Merchant Center free listings and AI referrals were counted separately from Google organic search. Orders with unknown sources remained in a separate category.
Reporting period. Refunds were assigned to original purchase cohorts through June 30, 2025 and June 30, 2026, for the July 15, 2026 report. Both comparison years covered the full assortment, including peak winter demand and quieter months. The estimated additional revenue covered the combined result of agency and client work.
Become the brand
AI recommends
Book a free 30-minute call with Eugene. Before the call, we check what Google and AI say about you. On the call, you get the first steps to make AI recommend you.