Case studyComputer hardware retail
Name under NDA
£3.2m in additional organic revenue for a UK computer hardware retailer
- £3.2m
- +35%
- 3x
Who the client is
Two storefronts. The retailer sells PC components and accessories through one storefront and assembled, configurable gaming PCs through another. It serves UK enthusiasts building or upgrading PCs and gamers choosing complete systems, in English. Before the engagement, estimated annual online hardware revenue was approximately £24m.
Engagement terms. The completed engagement ran from 26 August 2025 to 25 August 2026. The fee was €12,000 monthly across twelve payments, totaling €144,000. Results were reported on 12 October 2026 using the fulfillment and refund assumptions below.
What they came to us with
Revenue priorities. The client wanted more organic-search revenue from buyers researching compatibility and gaming requirements. It also wanted AI recommendations to explain its offers accurately: stocked parts and built-to-order systems have different delivery, warranty and support terms.
What they got
Additional revenue. We compared estimated fulfilled organic hardware revenue, after a refund provision, with a monthly estimate for the previous approach that allows ordinary growth and partial purchase-funnel improvement. The chart below shows the difference. Both totals use the accounting assumptions below.
- £9.7mestimated with the previous approach
- £3.2mestimated additional revenue
Our fees were €144,000 for the year.
Search growth. Annual unpaid search sessions across both storefronts rose 35%, while paid orders attributed to those sessions increased by nearly half. These results cover the organic channel’s customers, including returning buyers.
AI discovery. In a fixed sample of 250 UK-English buying questions, collected three times on each of ChatGPT, Perplexity, Gemini and Claude, answers linking either storefront tripled.
August 2025August 2026
Answers linking to the retailerShare of all sampled answers
Answers mentioning the retailerEither storefront counted once per answer
Share among four tracked retailer groupsClient appearances among all tracked group appearances
Favorable sentimentAmong answers mentioning the client
Claim precisionCorrect product and service claims among assessed claims
The situation before the engagement
Business context
Different buying decisions. Component buyers compare price, compatibility and availability. Complete-PC buyers also need confidence in game performance, configuration choices, assembly and after-sales support. Black Friday, holiday shopping and hardware releases changed demand throughout the year.
Website and content
Catalog scope. Around 6,000 items comprised 5,300 physical products and 700 services or licenses. Listed variants counted separately; each configurable base PC counted once. Approximately 6,300 product URLs represented storefront offers, within around 7,600 website pages.
Disconnected maintenance. The retailer already had buying guides and technical specifications, but descriptions, specification fields and advice followed separate workflows. Component, stock and policy changes could leave conflicting information across pages.
Traffic sources
Established organic sales. The chart shows the approximately 3.2m annual sessions by source. Estimated annual organic hardware revenue was £8.4m after the refund provision.
- 1.12mOrganic search
- 0.96mPaid acquisition
- 0.64mDirect visits
- 0.48mEmail and referrals
Search presence
Uneven buying coverage. The grid below shows how 500 tracked Google buying queries ranked. AI answers often described the retailer generally, missing compatibility, assembly and support distinctions that could help buyers choose.
145 in the top ten at the start
235 a year later
- Google top three
- Positions four to ten
- Below the first page
One square per tracked buying query.
Previous approach
Separate workstreams. Marketing managed category maintenance, technical fixes and occasional articles without shared search priorities or an approved product-fact register. The audit found weak metadata, difficult-to-discover filtered categories, incomplete product markup, rendering inconsistencies, internal-link gaps and slow commercial templates.
The strategy we planned and why
Commercial and search priorities
Buyer decisions first. SEO & AI Search Strategy prioritized profitable needs the retailer could fulfill. The parts storefront needed to help buyers choose compatible upgrades; the systems storefront needed to explain which gaming PC suited their requirements and how it would be supported.
Specific purchase questions. AI Search Optimization (GEO & AEO) aligned answers across commercial pages, advice and support content:
- Component fit. Explain GPU clearance and power connectors, plus CPU, motherboard and memory compatibility.
- Gaming requirements. Connect complete-PC choices to games, resolution and budget.
- Service expectations. Explain the delivery and warranty terms that apply to each offer.
Technical SEO priorities
Discoverable commercial pages. Technical SEO prioritized rendering and crawl paths so search engines could access commercial content. Filter indexation and internal links helped direct discovery toward relevant categories and products while preserving distinct storefront offers and controlling redundant paths.
Clear product information. Metadata made each page’s product or category easier to understand in search listings. Structured data helped search engines interpret published product details. Performance work improved the commercial templates buyers used to compare offers.
Content strategy and production order
Commercial publishing order. Content Research & Strategy put existing pages with buying traffic first, followed by missing decisions and comparisons. Guides connected compatibility, upgrading and gaming requirements to suitable available products. Broad technology news was secondary to questions linked to purchasing.
Maintained source facts. AI Content Production Systems used manufacturer specifications, product identifiers, compatibility rules and approved service terms. Specialists checked variant-specific fit and gaming recommendations against applicable technical evidence; component substitutions triggered review. Client approval connected publication to current stock and policies.
Why the strategy fit
Existing product expertise. The retailer already had buying guides and technical specialists. We prioritized purchase questions that benefited from that knowledge and led to available products, ahead of broad technology news. Weekly working sessions taught the teams how buyer needs, accurate facts and availability shaped those priorities.
What we did
Agency and client responsibilities
- Agency ownership. We led strategy, research, technical diagnosis, AI monitoring, workflow development and editorial operations.
- Client implementation. Developers changed the website. Merchandising and technical specialists checked specifications, compatibility and stock; support approved service claims.
- Publication decisions. The marketing lead owned priorities and final publication. Weekly calls combined results review, team education and decisions on the next work, with Slack supporting coordination.
- 1Research, strategy and technical fixes
- Buyer segments and positioning
- Template and performance repairs
- Release checks
- Strategy and execution plan
- 2AI content production system
- Product-fact register
- Drafting workflow
- Checks for each product type
- Approval rules
- 3Ongoing production, measurement and improvements
- Catalog-wide product checks
- New and refreshed guides
- Category page updates
Weekly sessions on buying traffic, AI descriptions and support questions
Phase 1: Research, strategy and technical fixes
First-month priorities. Technical diagnosis and engineering fixes ran alongside market research, buyer segmentation and positioning. The month ended with an agreed strategy and 90-day execution plan connecting engineering and content work to buying needs.
- Template repairs. Developers corrected commercial-page rendering, metadata and structured-data handling, then improved performance, navigation, internal links and crawl/indexation controls.
- Release checks. We compared crawls, inspected representative rendered pages and server logs, validated eligible physical-product markup, and monitored lab and field performance after releases.
- Service distinctions. Approved facts separated stocked-part dispatch from custom-PC build times and parts coverage from labor-only warranty coverage.
Phase 2: Building the AI content production system
Second-month production. We built the product-fact register, drafting workflow and approval rules, then validated and published initial catalog batches. Type-specific checks made the workflow usable across the retailer’s different offers.
- Automated checks only
- 600 exception items reviewed by our team
- 300 stratified samples reviewed by our team
One square is 100 items. Automated checks covered every item; client specialists checked the high-risk exceptions and samples.
- Physical products. Drafts were checked against specifications and compatibility rules.
- Services and licenses. Checks used applicable coverage and eligibility terms.
- Release approval. Our reviewers and client specialists split the checks as shown above. Replacement batches followed the same review rules.
Phase 3: Ongoing production, measurement and improvements
Catalog-wide delivery. By month six, checks covered the whole catalog and approximately 6,300 product URLs, assessing identifiers, specifications and offer terms. Publication required supported claims; unresolved compatibility or service claims were blocked pending specialist correction. Shared products retained storefront-specific price, availability and terms.
- Guide publishing. New guides were published across months three to twelve, about 22 a month, and existing guides were refreshed.
- Page maintenance. Overlapping pages were consolidated into retained pages and category pages updated. Around 800 replacement catalog items also passed through the workflow.
- Operating reviews. Weekly sessions examined useful buying traffic, inaccurate AI descriptions and support questions. Teams updated recommendations when stock or specifications changed; product-feed checks and technical monitoring continued throughout the year.
220 new guides
200 guides refreshed
415 category pages updated
55 overlapping pages merged
One square is about 10 pages.
Results
Search and AI visits
Both storefronts, analytics
Year before
Engagement year
Includes manufacturer and product searches, Search Console
Year before
Engagement year
Counted as referrals, outside organic revenue
Year before
Engagement year
Business results
- Year before
- Gained in the engagement year
Each band is drawn to scale. Paid orders rose from 2.50% to 2.74% of organic sessions.
Estimated additional revenue. The comparison assumes 10% ordinary organic traffic growth and retains 25% of each month’s relative improvement in session-to-cart, cart-to-checkout and checkout-to-payment rates, separately for components and systems. Monthly orders then receive the same segment prices, cancellation rates, refund provision and fulfillment timing as the engagement scenario.
- Engagement result
- Estimated with the previous approach
- Additional revenue
Each month runs from the 26th to the 25th: Sep covers 26 August-25 September.
Scenario sensitivity. Varying ordinary growth from 0%-20% and retained funnel improvements from 0%-50% gives approximately £2.1m-£4.3m additional revenue, holding other inputs fixed. The central difference is an assumption-based estimate for the joint work. Market shifts and channel switching remain unresolved influences.
Segment and company context. The chart splits organic revenue by storefront, under the same accounting assumptions. Estimated all-channel online hardware revenue increased from approximately £24m to £30.3m. Commercial selection also considered product contribution and client delivery costs.
Year before
£5.04m components, £3.35m complete PCs
Engagement year
£7.63m components, £5.30m complete PCs
- Components and accessories
- Complete gaming PCs
Conclusion
Continued ownership. The retailer finished with agreed buyer priorities, repaired technical foundations and a publishing process tied to product facts. Marketing retained publication decisions; merchandising and technical specialists maintained specifications and compatibility checks, while support owned service terms. These responsibilities carried the process beyond the engagement.
Commercial results. Organic orders and assessed AI claim accuracy improved during the engagement. The monthly comparison above estimates the additional organic revenue under its stated assumptions.
How we measured
Search measurement
Annual search comparison. Annual results compare 26 August 2024-25 August 2025 with 26 August 2025-25 August 2026. Both years use consistent cross-domain session and order-source rules. Display values are rounded where appropriate.
Query measurement. Non-retailer-branded clicks include manufacturer and product searches. Search Console clicks and analytics sessions have different collection scopes.
Ranking comparison. Rankings use the same 500 UK-English buying queries at the engagement's start and end.
AI answer measurement
AI sample. The fixed panel contained 100 compatibility, 100 complete-PC selection and 50 retailer-trust questions without naming the client. Captures on 26-28 August 2025 and 23-25 August 2026 each contained 3,000 answers. Each engine carried equal weight; the rates describe this panel.
Retailer comparison. Each answer counted either storefront as one client-group mention. Client mention rate rose from 420/3,000 (14%) to 870/3,000 (29%). Share of voice compared the client with three fixed competing retailer groups, giving four groups in total.
Evaluation rules. Automated sentiment classified client-mentioning answers against a fixed rubric. Claim precision assessed individual product/service statements against date-applicable manufacturer specifications and approved policies, counting unsupported claims as incorrect. It measured correctness among assessed claims; service-topic error reduction remained unmeasured.
AI referrals. Identifiable AI referrals increased from approximately 1,800 to 12,500 annual sessions and from 32 to 250 paid orders. These visits were classified as referrals and excluded from organic-search revenue.
Orders and revenue
Purchase funnel. The calculation assumes one hardware order per purchasing session, completing each stage in order. Sessions and orders enter one segment; complete-PC baskets include component add-ons. Mixed hardware/service baskets contribute hardware value; service-only orders are excluded. Returning buyers remain included, and orders retain their purchasing-session source through fulfillment.
Revenue scope. Revenue estimates fulfilled hardware value after discounts and cancellations, excluding VAT and delivery. A fixed 5% refund-value provision applies to both segments in both years. Orders lacking acquisition data remain unassigned within hardware-only company totals.
Delivery timing. Periods run from the 26th through the 25th. Fixed assumptions recognize 96% of uncanceled component value and 85% of system value in that period; the balance follows next period. Both scenarios share the same opening backlog; closing backlog moves forward. The provision supplies the entire refund deduction.
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.