Mars Case Study: Driving Tens of Millions in Incremental Revenue in Just 4 months, leading to a Global AI Visibility Partnership

Mars has driven tens of millions in incremental revenue through its wide-ranging partnership with Azoma, seeing a +17.7% average AI Share of Voice uplift, +68% YoY SKU revenue uplift, and +43% revenue uplift vs. control.

A Historic Brand Embracing the Future

For more than a century, Mars Wrigley has been synonymous with innovation - from electrifying chocolate production in the 1920s to transforming global snacking culture. Today, Mars is once again redefining what it means to stay ahead of technological change — this time, through AI and Generative Retail Search.

In partnership with Azoma, Mars Wrigley is pioneering how global brands can leverage Generative Engine Optimization (GEO) to drive visibility, relevance, and sales in the era of Agentic Commerce, defined by AI-powered search engines and retail shopping assistants.

The Challenge: Search Is Changing Faster Than Ever

Retail search has entered a new era. Platforms like Amazon’s Rufus, Walmart’s Sparky, and ChatGPT Shopping are fundamentally changing how consumers discover products online. On Black Friday’25 alone, AI Chatbots & Agents drove $14.2 billion in sales, including $3 billion in the U.S. In Amazon’s Q3 earnings call, influencing 20% of purchases. On the Q3’25 Amazon earnings call, Andy Jassy stated that Customers using Rufus are 60% more likely to complete a purchase; and in Q4’26 stated that Rufus delivered $12bn in incremental annualised sales, whilst the Walmart CFO said on their earnings call that roughly half of Walmart’s app users have used Sparky, and that Shoppers using Sparky place orders that on average 35% larger. For a global CPG like Mars Wrigley, this presents both a challenge and an opportunity - how do you win in this new era of consumer discovery?

The Solution: Partnering with Azoma to Power Agentic Commerce Optimization

Mars Wrigley partnered with Azoma to run a structured pilot across two divisions (Snacking and Pet) and eight brands spanning three markets (US, UK, and Canada). The pilot followed a phased approach: first measuring baseline AI search visibility across AI Shopping Agents like Rufus, Sparky and ChatGPT, then identifying the key conversation themes and content signals influencing AI recommendations, and finally optimising content to improve discoverability.

Azoma's work covered five core content levers: PDP written content (titles, bullet points, descriptions), PDP visual content (enhanced product images, alt text, image keywords), Q&A seeding on marketplaces, review strategy, and citation building across the open web. Each lever was calibrated to how retail AI shopping assistants actually retrieve and synthesise product information.

The Results: Measurable Lift Across Visibility, Content Quality, and Revenue

Across all eight pilot brands, the results were significant:

• +17.7% average Share of Voice uplift across all brands, with standout gains for Extra Refreshers (+28.3%), M&M's (+28.6%), and 5 Gum (+18.3%)

• +41.8% average Content Quality Score improvement, driven by a +28.3% COSMO score uplift and a +65.7% Rufus Score uplift across the portfolio

• +68.0% average YoY ASIN revenue uplift for optimised products, with M&M's (+45.9%), Extra Refreshers (+82.8%), and Pedigree (+69.7%) leading the pack

• +43.0% revenue uplift vs. control, isolating the incremental impact of Azoma's optimisations from broader brand and category trends. The control group (total brand revenue YoY) grew +25.0% on average, meaning Azoma-optimised ASINs outperformed the brand baseline by 43 percentage points

• Best Seller Rank improvements of -15.8% on average (number 1 is the best), with Pedigree improving by -24.8% and M&M's by -26.5%

In such competitive categories, this is worth tens of millions in additional revenue,

Speaking midway through the pilot, Tory Bradley (Global e-commerce retailer Search Director) was so pleased with the results she commented publicly about them to Allison Smith of ModernRetail.

Listen to the full podcast

Looking Ahead: Global Rollout and the Agentic Commerce Infrastructure

The pilot has paved the way for a much larger, global-scoped partnership between Mars Wrigley and Azoma. The next phase will extend coverage across additional segments, markets, and retailers, building on the playbook established during the pilot.

Beyond content optimisation, Azoma is helping Mars build the full infrastructure layer for agentic commerce: connecting product catalogues and brand guidelines into a centralised knowledge graph, syndicating structured product intelligence across AI ecosystems, and establishing direct machine-readable presences inside platforms like ChatGPT (via dedicated ChatGPT Apps and MCP servers). Via the Agentic Merchant Protocol (AMP), Azoma is enabling brands like Mars to move from reactive content updates to a programmatic, always-on approach to how their products are understood, ranked, and recommended by AI agents worldwide.

About Azoma

Azoma is the enterprise infrastructure layer for Agentic Commerce. The platform enables global brands to control, measure, and optimise how their products are understood, ranked, cited, and recommended by AI shopping agents and assistants, including Amazon Rufus, Walmart Sparky, ChatGPT, and Google Gemini.

Book a demo with our team to find out more.