Pizza Hut has already reshaped how it understands customers, using Fullstory's behavioral data to move beyond commercial metrics and see the “why” behind customer behavior across 110+ markets. But behavioral data doesn't just explain what happened—it fuels what comes next. For Fabian Alvares, Global Head of CRO and Experimentation, and his team, Fullstory has become the engine behind a fast-evolving experimentation program: one that's moving from manual A/B testing to AI-built product affinities, and toward a self-improving, agentic future.
For the full picture of how Pizza Hut transformed its global customer experience with Fullstory, read the customer story: Pizza Hut transforms global customer experience with Fullstory →
Great experiments start with data, not opinions
Where do good A/B tests come from? Too many organizations still rely on the opinion of the highest-paid person in the room. Pizza Hut takes a different view: the best experiments are rooted in high-fidelity customer data. With 15 years of experimentation experience and thousands of tests behind him, Fabian treats behavioral data as the raw material for every hypothesis. Comprehensive behavioral capture lets the team do more than track what users, they can intuit what users are thinking and feeling, understand why an experiment succeeds or fails, and surface the friction that dashboards hide.
From one side item to a multivariate matrix
Pizza Hut's experimentation program spans standard A/B testing, multivariate testing, and web personalization. A recent basket-page test shows how the team scales an idea. They started simple: a pop-up modal on the basket page asking customers whether they'd like to add a single side of fries before checkout. It worked, resulting in a 1.5% increase in total transactions.
But Pizza Hut offers hundreds of side items, and testing each one in isolation wasn't feasible. So the team expanded the experiment into a multivariate test spanning a range of sides like mac and cheese, chicken, wedges, Pepsi Max, and more, capturing how customers actually responded to each option with Fullstory.
Turning behavioral data into AI-built product affinities
This is where the program takes its biggest leap. Rather than manually combing through the multivariate results, the team pushed the raw Fullstory interaction data into an AI model. Within 15 minutes, a workflow that would have taken one to two weeks by hand, the AI produced a targeted product-affinity matrix, mapping which side to surface for which customer, and when:
The output isn't just a smarter upsell. Pizza Hut is treating these product relationships as company-wide intelligence—sharing them beyond CRO with food development, product innovation, and marketing teams to shape omnichannel strategy. Behavioral data captured in a single digital experiment now informs decisions well outside the online storefront.
“Where this would have taken us maybe one to two weeks, with Fullstory and AI I was able to do it in about 15 minutes. There's a hidden value in running an experiment and using AI to claw through all that insight.”
— Fabian Alvares, Global Head of CRO & Experimentation, Pizza Hut
A blueprint for agentic, self-improving experimentation
The next frontier for Pizza Hut is a “human plus AI” agentic ecosystem. Fabian envisions a scaled optimization framework built on three principles:
The goal is a system that moves from enrichment to automated execution to continuous self-improvement—always with human-guided guardrails to evaluate and validate what the AI proposes.
Why behavioral data changes the game for experimentation
By treating behavioral data as the fuel for experimentation, Pizza Hut is changing how a global brand tests, learns, and personalizes. The team is now preparing to launch its newly mapped AI product affinities, establishing a scalable, omnichannel blueprint for digital retail.
Behavioral data is the best source of hypotheses. The strongest tests come from what customers actually do, not opinions or guesswork.
AI turns behavioral data into scale. Analysis that once took one to two weeks now takes fifteen minutes, making it feasible to personalize across hundreds of variables.
Insight shouldn't stay in a silo. Behavioral findings from a single experiment can inform product, food development, and marketing across every channel.
