Improving deal discovery during high-traffic retail events.
When shoppers land on Amazon during major retail moments, the deals they see are shaped by a mix of automated systems and strategic merchandising decisions. My summer project sat with the team responsible for planning and prioritizing manual deal visibility for strategic products and campaigns across events like Prime Day, Big Spring Sale, Black Friday, and Cyber Monday.
While most deals are surfaced algorithmically, select high-priority campaigns may receive manual support in prominent onsite placements. The team needed better visibility into how these manually supported campaigns performed so they could make clearer decisions about which products, categories, and merchandising approaches deserved additional visibility during high-traffic events.
How should teams decide which campaigns deserve manual support?
Manual deal placements required significant time to review, approve, and execute, but teams did not have a consistent view of which campaigns, placements, or merchandising approaches actually performed best.
As category and vendor teams requested visibility for strategic products, the team needed a more data-supported way to decide when manual support was justified, where those deals were most likely to perform well, and how to reduce inefficient manual review.
A playbook for evaluating future manual deal placement.
I delivered and presented an Amazon-style narrative that translated campaign performance analysis into a proposed playbook for evaluating manual placements.
The playbook helped clarify how teams could prioritize high-impact campaigns, match merchandising approaches to event types, and identify where automation could support future planning.
The work was broken into three phases.
I moved from understanding the planning workflow, to analyzing performance across campaign and placement types, to packaging the findings into a stakeholder-ready playbook and final narrative.
Understand the workflow and define the metrics
I familiarized myself with the team’s tools, planning process, Ai Agents and manual merchandising workflows. I also interviewd product managers, marketing managers, category teams, and data partners to understand review pain points, stakeholder expectations, and which performance signals mattered most.
Analyze placement performance
I reviewed 400+ campaigns across Prime Day, Big Spring Sale, Black Friday, and Cyber Monday. The analysis looked across placement types including Single Cell Takeovers, Shovelers, Above the Fold, and Below the Fold.
I worked with data partners to understand attribution limits, compare placement types, and test early hypotheses using SQL, Amazon QuickSight, and internal reporting tools.
Build and present the playbook
I consolidated the research and analysis into a 6-page Amazon-style narrative, iterating with stakeholders to align on tradeoffs before presenting recommendations for evaluating manual placements across future retail events.
What the analysis helped clarify.
Some campaigns were better suited for manual visibility.
Campaign fit depended on category, message clarity, event context, and whether manual support could add value beyond standard placement logic.
Different events shaped how customers browsed.
Discovery-led events and high-intent shopping moments created different expectations for how shoppers engaged with deal placements.
Manual review needed clearer decision support.
Teams needed a more consistent way to compare campaign quality, evaluate placement fit, and reduce time spent on low-fit requests.
What the final playbook covered.
Campaign evaluation criteria
Defined clearer criteria for evaluating which campaigns should receive manual support and how placement requests should be assessed.
Review support opportunities
Identified where automation could support manual review while keeping human judgment in the final decision loop.
Reusable event planning framework
Packaged the logic into a repeatable framework for future high-velocity retail events.
Key takeaway: small, thoughtful changes in merchandising logic can shape how customers discover products, even on a platform operating at Amazon’s scale.