Overview
Best Time to Buy, Consumer Reports
From 0→1, I helped identify, validate, and design the Best Time to Buy shopping agent. It saves consumers time and money by purchasing products at the lowest price while ensuring CR affiliate links are used, both boosting member retention and projecting $2.1M in ARR.
Role
Product Designer
Team
1 engineer, 2 designers
1 researcher, 1 PM
Timeline
7 months
Skills
Product strategy, agentic design,
Stakeholder management, UX research
Problem & constraints
Consumer Reports asked us to identify and design a high-value agentic AI use case that could launch in 3 months, run on their existing mobile platform, and drive revenue while fitting seamlessly within their brand of consumer trustworthiness.
Scoping
After evaluating two opportunies to focus on within the consumer journey, pre-purchase (personalization) and during-purchase (automation), we chose during-purchase because it offered three key business advantages:
Consumer Research
Immediate Buyers and Deal Waiters.
In the during-purchase phase these were two distinct behavioral groups we encountered.
But how many people are actually waiting to buy?
It's easy to assume that most consumers are overtaken by the cultural forces that incessantly aim to remove friction, training us all to crave instant gratification in ever-expanding aspects.
We found that the satisfaction of a good deal can overtake instant gratification.
93% of consumers in our survey (n=140) reported a specific item they were waiting to buy on sale. Many of the items listed included large and small appliances, considered part of the bread and butter of CR reviews.
Existing Feature Audit
Extending the existing feature to help with our tight timeline and adoption.
The previous "Best Time to Buy" was based on anecdotal data and lacked follow-through support, both desired changes according to CR’s user studies.
💭 How Might We
Help consumers decide if it's worth waiting and, if so, ensure they complete the purchase at the best time?
💡 Solution
A smart agent that monitors specific product prices and automatically purchases at the optimal time, saving consumers time and money while ensuring CR affiliate links are used.
The Landscape of Trust
Information is everywhere, but trust is scarce.
We found consumers are increasingly wary of big tech companies and financial platforms notifying them of discounts or automating purchases because of (real or perceived) deceptive patterns like inflated prices or kickbacks.
Designing for Trust
We used trust as both a differentiator and as a design constraint, meaning CR could lead by example in their use of agents.
Opt-in design for agentic features
We prioritized user control over full automation by providing the option to buy automatically or require confirmation when the price reaches their target. From user testing, we found some felt additional confirmation negated the agent's value, while others worried about forgetting they’d set up purchases or that their financial circumstances might change.
Criteria recommendations
Suggests optimal price & time criteria based on dynamic market data to guide consumers to confidently set up their criteria instead of forcing them to start from scratch. From testing, we found most users chose the agent’s recommendations with a high level of trust. They strongly preferred graphical representation over other formats.
In-the-loop expectation updates
Keeps users informed about shifting market conditions and guides them in updating their autobuy criteria to get the best deal. Users worried that setting criteria too low early on might cause them to miss a “good enough” deal while chasing the “best” one.
Modality Explorations
We tested and later intentionally avoided multi-modal features in our solution.
As interesting as it was to get closer to "human agent" experiences, utilizing voice and chat made our participants feel less confident and less in control of setting their time and price criteria. We also wanted to avoid alienating members, as CR's demographics skew older.
Reflections on AI UI
AI doesn’t need to look flashy (sparkles and all) when utilized in consumer products.
Focusing on the value it brings and garnering trust with the way people interact with it is much more important. We intentionally leaned into CR's branding and leveraging AI on the backend over dousing our product with AI ✨ UI language, which would have potentially alienated users.