Exon Games knew the revenue was in the catalog. They just couldn't reach it.
The store sells digital game keys, software licenses, and subscription services across every major platform: Steam, Xbox, PlayStation, Nintendo, Microsoft Office, Adobe, Netflix. Thousands of SKUs, and the relationships between them are obvious once you see them. Buy FIFA on Xbox and you probably want an Xbox Game Pass subscription. Buy Windows 11 and you probably need an Office license. Buy a base game and you're a candidate for the expansion pack, unless you already bought the deluxe edition that includes it.
A human merchandise team could map those relationships for a hundred products. At thousands of SKUs, with new titles, DLCs, and subscriptions launching every week, manual cross-selling breaks down before it starts. Every unlinked product was a missed opportunity, and Exon Games had thousands of them.
This case study covers what happened when AI product recommendations were built into the store: 13.46% more revenue from existing traffic, a 26% AOV lift, and 20x ROI.
About Exon Games
Exon Games is a high-volume Shopify store based in Israel, serving the local gaming community with digital game keys, software licenses, and subscription services. Their catalog spans every major gaming platform alongside productivity software, creative tools, and premium subscriptions including Netflix, Spotify, and YouTube Premium.
The store operates entirely in Hebrew with a right-to-left layout and Israeli Shekel pricing. Thousands of digital products, where the same title exists across multiple platforms and each version comes with its own DLCs, season passes, and add-ons.
The challenge: cross-selling a digital catalog that generic AI can't read
Platform context matters more than any other signal
A customer buying "GTA V for Xbox Series X" and a customer buying "GTA V for PlayStation 5" are not the same customer from a cross-sell standpoint. The first needs Xbox Game Pass, Xbox gift cards, Xbox-compatible DLC. The second needs PlayStation Plus. Surfacing PlayStation content to an Xbox buyer is worse than showing nothing — it trains customers to ignore the recommendation widget entirely.
Most recommendation engines treat platform as a product attribute, like color or size. They don't account for the fact that platform determines a whole ecosystem of compatible products. An Xbox game pairs with Xbox subscriptions not because of shared keywords, but because of logical relationships that standard AI isn't built to track.
Version and edition relationships change with every release
Digital game catalogs aren't static product grids. Games have standard editions, deluxe editions, DLCs, season passes, and expansion packs, with overlapping content across all of them. The AI needs to know that someone buying a base game is a candidate for the expansion, but someone buying the deluxe edition that already bundles it isn't. These relationships shift with every new release. A cross-sell that was accurate last month may be wrong this month.
A catalog that changes too fast for manual curation
New games, DLCs, and subscription tiers launch constantly. Exon Games' catalog changes weekly. Manual product pairing couldn't keep up, and every new product added without cross-sell connections was a missed opportunity from day one. The team needed a system that could learn the gaming ecosystem and adapt automatically, not a static configuration that required ongoing maintenance.
The solution: platform-aware AI and bundle pricing that gives customers a reason to add more
PersonalizerAI deployed a recommendation engine trained specifically on Exon Games' catalog structure, order history, and the platform relationships within the gaming and software ecosystem. Two capabilities drove the result.
1. Platform-aware recommendations
PersonalizerAI treats platform as a primary dimension in every recommendation decision, not an attribute to filter on. It understands that "XBOX Game Pass Ultimate 3 Months" and "Forza Horizon 5 — Xbox Series X|S" belong in the same recommendation set because they share platform context. A Windows license pairs with Microsoft Office. An antivirus subscription pairs with VPN services. A gaming subscription pairs with titles available on that specific platform.
The same logic handles version relationships. The AI distinguishes base games from deluxe editions and knows which DLCs and season passes are already bundled in. A customer who bought the deluxe edition doesn't see a recommendation for content they already have. A customer who bought the standard edition does. That distinction matters because getting it wrong teaches customers to ignore the widget — and recovering from that is much harder than getting it right from the start.
2. Frequently Bought Together with automatic bundle discounts
When a customer adds a game to their cart, the FBT widget surfaces complementary products — DLC, subscriptions, related games — with an automatic bundle discount applied through Shopify Functions. "Buy FIFA 25 + Xbox Game Pass Ultimate 1 Month — save 10%." The discount makes the bundle feel like a deal rather than an upsell. Customers perceive value and Exon Games captures revenue that would otherwise go unclaimed.
Shopify Functions handles the discount logic natively at checkout. No workarounds, no codes for customers to enter — the pricing adjustment applies automatically when the bundle is added to cart.
The combination matters: platform-coherent suggestions remove the relevance problem, and automatic discounts remove the friction. Those are the two reasons customers skip cross-sell widgets. Address both and conversion follows.
The full customer journey
PersonalizerAI deployed across five touchpoints.
At the collection and homepage level, product grids adapt by platform and browsing behavior. A customer browsing Xbox games sees Xbox-relevant content throughout the store, not a generic bestseller grid.
On product pages, FBT bundles with automatic discounts surface platform-matched DLCs, subscriptions, and related titles alongside whatever the customer is viewing.
The implementation includes full right-to-left support for Hebrew and localized Israeli Shekel pricing. The AI and widget layer adapts to the store's language and layout without any loss in recommendation accuracy.
The results
+13.46% revenue lift. New revenue from existing traffic, attributed directly to AI-powered recommendations and bundle pricing. No additional ad spend, no new customer acquisition — the same visitors spending more because the catalog's cross-sell relationships were finally visible.
20x ROI. Every dollar spent on PersonalizerAI returned twenty. Platform-aware recommendations and automatic bundle discounts paid for themselves within weeks and continue to compound as the AI refines its understanding of customer behavior.
+26% AOV increase. Customers who added bundled games and subscriptions at a discount spent significantly more per order than those buying individual items. The bundle pricing made higher spending feel like saving — which is a different psychological ask than "would you like to add something else."
What 13.46% more revenue means at different store sizes
An additional 13.46% in revenue from existing traffic adds up quickly. At different monthly baselines:
Monthly revenue baseline | Additional monthly revenue |
|---|---|
$50,000 | +$6,730 |
$100,000 | +$13,460 |
$500,000 | +$67,300 |
$1,000,000 | +$134,600 ($1.6M+ annually) |
Based on Exon Games' verified 13.46% revenue lift applied to different revenue scales. Actual results vary by catalog, traffic, and vertical.
What other Shopify merchants can take from this
The Exon Games result is most directly applicable to stores where products have ecosystem-level relationships that generic AI can't surface. But a few of the underlying principles apply more broadly.
Cross-sell relevance has a threshold. Below it, recommendations don't just fail to convert — they actively damage the experience. In a digital gaming catalog, a cross-platform suggestion signals to the customer that the widget doesn't know what they bought. Once customers register that, they stop engaging. The goal isn't just to surface more recommendations. It's to surface ones that are specific enough to earn attention.
Bundle pricing is an underused AOV lever for digital merchants specifically. Physical product stores use free shipping thresholds to push average order value. Digital products don't ship, so that mechanism doesn't apply. Automatic FBT discounts fill the same role: they give customers a concrete reason to add a second item that reads as value rather than an add-on pitch. The FBT widget does what the shipping threshold does in a physical store — but the incentive is tied to a specific complementary product rather than an arbitrary cart total.
The relationship between catalog complexity and recommendation quality is also worth addressing directly, because the conventional assumption gets it backwards. Simple catalogs with obvious product relationships can be cross-sold manually, or even with generic templates. Complex catalogs — where thousands of SKUs have platform-specific, version-specific, and edition-specific relationships — are where the gap between a purpose-trained model and a generic one is largest. The harder the catalog is to curate manually, the more an AI built for that specific structure outperforms one borrowed from somewhere else.
For digital product merchants: the recurring revenue angle is worth building toward. A customer who buys a game subscription this month is a candidate for renewal, an upgrade, and related titles next month. Post-purchase recommendations on the thank-you page are a low-cost way to start that conversation without waiting for the customer to come back on their own.
Is there revenue hiding in your catalog?
Exon Games' result came from a catalog where the product relationships were real but invisible platform logic, version hierarchies, and subscription bundles that no manual process could maintain at scale. AI found the revenue the team already knew was there.
In 20 minutes, the PersonalizerAI team will audit your product discovery setup, walk through projected revenue lift based on your actual store data, and show you how recommendations would work across your catalog.
Built for scaling Shopify brands with 100+ SKUs and active ad spend.