Members using more benefits showed stronger renewal.
I owned Prime's global lifecycle engagement platform, helping country teams connect benefit discovery and use to renewal across a 200M+ paid-member ecosystem.
Senior Manager, Prime Worldwide Engagement (L7 PMT)Shipped work
MEMBER CONTEXT
Lifecycle stageBenefit useCountry
Relevant next action
DiscoverUseReturn
KELVIN ZEE
AMAZON PRIME · 2022 - 2023
01 · THE MANDATE
02 · THE FINDING
03 · THE PLATFORM
04 · WHAT CHANGED
05 · THE TEAM
US ANNUAL PRIME PRICE
$139
after the February 2022 increase from $119
Would members still seeenough to find it worth it?
ASSOCIATION
~70%
stronger 60-day yield for members using three or more benefits
SCOPE
26
countries served
PROGRAM RESULT
4 weeks4 hours
4 WEEKS, TO SCALE
approximate localized customer experience build time
Self-service targeting and experimentation for PMs in 26 countries
PROGRAM RESULT · BROADER LIFECYCLE WORK
+23%
higher renewals
+17%
higher shopping activity
30M+monthly benefit uses
1B+impressions
SCOPE
130+
people across product, engineering, design, data science, and analytics
5
Senior PMTs, on a team grown from one
KELVIN ZEE · AMAZON PRIME
Members usingmore benefits showedstronger renewal.
Role reel, at your own pace
The mandate
In February 2022 the US annual Prime price rose from $119 to $139. The question was whether members would still see enough of the membership to find it worth it.
The findingASSOCIATION
Members using three or more benefits showed roughly 70 percent stronger 60-day yield. That is an association, not proof of cause, and it pointed the work at recommendations across video, music, and shopping.
The platformSCOPE · PROGRAM RESULT
The platform served 26 countries. It gave PMs self-service targeting and experimentation, and took the approximate build time for a localized customer experience from 4 weeks to 4 hours.
What changedPROGRAM RESULT · BROADER LIFECYCLE WORK
Across the broader lifecycle work: 23% higher renewals and 17% higher shopping activity, with more than 30 million monthly benefit uses and more than one billion impressions.
The teamSCOPE
A 130+ person cross-functional organization spanning product, engineering, design, data science, and analytics. I grew the PM team from one to five Senior PMTs.
The role in 20 seconds, using only facts published on this site.
26countries served
4 weeks → 4 hoursapproximate localized customer experience build time
+23%renewal outcome in the broader lifecycle work
Country teams owned the local commercial outcomes. Renewal, shopping and engagement results belong to the broader lifecycle work rather than to any one recommendation, and the 200M+ paid-member figure describes the ecosystem the platform served rather than membership I created.
I joined Prime Worldwide Engagement immediately after the February 2022 US annual price increase, from $119 to $139. The mandate was revenue defense, and the product question was whether a member could see enough of the membership to still think it was worth it.
Members arrived through different benefits. A video-first member and a shopping-first member did not begin with the same understanding of Prime. The next recommendation needed to reflect that context.
I led lifecycle evaluation across acquisition mix, plan type, geography and benefit-activation depth. An internal analysis I led found that members using three or more benefits showed roughly 70 percent stronger 60-day yield, an association that shaped the activation strategy. That made activation depth the thing to design for, rather than impressions.
Give country teams a common platform
I led product development of a shared platform that brought member preferences and retail product information together. It supported recommendations across video, music, and shopping on Amazon and off-Amazon surfaces.
Targeting could change by ingress point, lifecycle stage, benefit use, and market context. A configurable incentive layer let country teams adapt customer experiences through common capabilities.
Local teams owned their commercial outcomes. The central team supplied dedicated development capacity and translated country needs into reusable platform capabilities. That distinction kept local judgment close to customers while reducing bespoke engineering.
Build the team and the operating model
I grew the PM team from one to five Senior PMTs. I led product direction and investment priorities for a 130+ person cross-functional organization spanning product, engineering, design, data science, and analytics.
The engineering, design, and data-science reporting lines were separate. Product direction depended on shared priorities and clear requirements across those functions.
We built self-service targeting and experimentation for PMs in 26 countries. Localized customer experience build time fell from roughly four weeks to four hours. Country teams could adapt a customer experience without waiting for a separate engineering build each time.
Connect engagement to the lifecycle
I established 30/60/90-day activation, retention, shopping, and customer lifetime value measures. Benefit activity mattered because of its relationship to the membership, so impressions alone could not decide whether a customer experience was useful.
The work connected benefit teams' objectives with the broader membership objective. That required explaining how engagement with Prime could support the video, music, and shopping experiences those teams already owned.
What changed
Across the broader lifecycle work, we measured 23% higher renewals and 17% higher shopping activity. The work reached more than 30 million monthly benefit uses and more than one billion impressions.
The transferable lesson is to build the shared capabilities that make local judgment faster. The platform needed to support different member contexts without forcing each country to rebuild the same machinery.
FROM THE WORK / 2022-2023
One instance of a control plane, not one campaign.
Prime Stampcard asked a member to use four parts of the membership, a purchase, a show, a song and a book, for a $10 credit. It is one output of what my team actually built: a self-service front end that let product managers in 26 countries compose their own do-X-get-Y experiences without waiting on bespoke engineering. Localized build time went from roughly four weeks to roughly four hours.
Each experience was configurable by audience, lifecycle stage, benefit, ingress and channel, which is more than 520 base country-benefit combinations before any variant. The incentive was optional, because the goal was completing benefit activation rather than buying a single transaction, and the placements reached well past one page on Amazon.
The cadence underneath was 30, 60 and 90 day activation and retention measurement, so a country team could tell whether their variant moved the behavior or only the click.
Beside it, the savings view that presented the value back to the member. One surface asks for a benefit to be used, the other shows what using them was worth.
