Optimizing our upsell strategy
at TurboTax
Optimizing our upsell strategy
at TurboTax
Optimizing our upsell strategy at TurboTax
© 2025 Aran Kim
© 2025 Aran Kim
OVERVIEW
OVERVIEW
As a product designer at TurboTax, I was tasked with improving the upsell strategy in the TaxHub experience by leveraging AI to deliver the most relevant offers at the right time, driving engagement and sign-ups.
Teaming up with a UX Researcher and Content Writer, we launched the enhanced experience just in time for the first peak release.
As a product designer at TurboTax, I was tasked with improving the upsell strategy in the TaxHub experience by leveraging AI to deliver the most relevant offers at the right time, driving engagement and sign-ups.
Teaming up with a UX Researcher and Content Writer, we launched the enhanced experience just in time for the first peak release.
TIMELINE
3 months (2024)
TOOLS
Figma
Figjam
ROLE
Product Designer
TEAM
1 UX Researcher
1 Content Writer
ROLE
ROLE
Product Designer
Product
Designer
TEAM
TEAM
1 UX Researcher
1 UX Researcher
1 Content Writer
1 Content Writer
TIMELINE
TIMELINE
3 months (2024)
3 months (2024)
TOOLS
TOOLS
Figma
Figma
Figjam
Figjam
CONTEXT
CONTEXT
The unexpected journey of contextual upsells
The unexpected journey of contextual upsells
We set out to rethink how Intuit presents upsell opportunities, but the journey was full of twists and surprises. After months of research, iterations, and conversations with stakeholders, the project was ultimately cut. But along the way, we uncovered insights that completely changed how we approach contextual upsells at TurboTax.
We set out to rethink how Intuit presents upsell opportunities, but the journey was full of twists and surprises. After months of research, iterations, and conversations with stakeholders, the project was ultimately cut. But along the way, we uncovered insights that completely changed how we approach contextual upsells at TurboTax.
*User journey of the NBA Card recipe
*User journey of the NBA Card recipe
PROBLEM & OPPORTUNITY
PROBLEM & OPPORTUNITY
How can we make upsells feel natural, not pushy?
How can we make upsells feel natural, not pushy?
Users felt our upsells were random, too frequent, and disruptive to their tax-filing experience. Instead of driving engagement, they caused frustration. To fix this, we needed a way to make upsells feel seamless—offering value without getting in the way.
Users felt our upsells were random, too frequent, and disruptive to their tax-filing experience. Instead of driving engagement, they caused frustration. To fix this, we needed a way to make upsells feel seamless—offering value without getting in the way.
“These offers feel random— none of them actually apply to my tax situation.” — Janette (age 36)
“These offers feel random— none of them actually apply to my tax situation.” — Janette (age 36)
MY ROLE & APPROACH
MY ROLE & APPROACH
Testing 2 approaches
Testing 2 approaches
I explored multiple ways to address user concerns and ultimately landed on testing two different UI patterns:
I explored multiple ways to address user concerns and ultimately landed on testing two different UI patterns:
Recipe A: NBA Card
Recipe A: NBA Card
As known as Next Best Action card, it's an inline component utilized in our TaxHub experience.
Recipe B: Persistent Banner
A fixed upsell banner appearing throughout the tax filing process.
Recipe B: Persistent Banner
A fixed upsell banner appearing throughout the tax filing process.
To make sure these upsells were timely and relevant, we leveraged our AI-powered recommendation engine to determine the most appropriate moment and placement for each upsell.
To make sure these upsells were timely and relevant, we leveraged our AI-powered recommendation engine to determine the most appropriate moment and placement for each upsell.
SOLUTION
SOLUTION
Recipe A: NBA Card
Recipe A: NBA Card
I designed five variations of the upsell, each with different visuals and messaging, allowing our AI model to deliver the most relevant option based on the customer’s tax situation.
I designed five variations of the upsell, each with different visuals and messaging, allowing our AI model to deliver the most relevant option based on the customer’s tax situation.
Recipe B: Persistent Banner
Recipe B: Persistent Banner
Our second approach used a persistent banner that stayed visible throughout the TaxHub experience. Just like our NBA card, clicking the CTA took users to the comp chart, where they could see all the upsell benefits.
Our second approach used a persistent banner that stayed visible throughout the TaxHub experience. Just like our NBA card, clicking the CTA took users to the comp chart, where they could see all the upsell benefits.
MAX
MAX
DIWM
DIWM
Super Bundle
Super Bundle
THE DECISION
THE DECISION
Why we had to pivot…
Why we had to pivot…
Once the contextual upsells launched within Taxhub, we noticed a sharp drop in S2C (Start-to-Complete rates) for tax filing. This led the Taxhub team to reduce the taxhub experience from 100% to just 10% of users, reverting the majority back to the traditional linear flow.
This change had a direct impact on our experiment: with only 10% of users seeing our contextual upsell experience, we wouldn’t be able to gather statistically significant data. As a result, we made the difficult decision to cut the test and revert to our previous upsell approach.
Once the contextual upsells launched within Taxhub, we noticed a sharp drop in S2C (Start-to-Complete rates) for tax filing. This led the Taxhub team to reduce the taxhub experience from 100% to just 10% of users, reverting the majority back to the traditional linear flow.
This change had a direct impact on our experiment: with only 10% of users seeing our contextual upsell experience, we wouldn’t be able to gather statistically significant data. As a result, we made the difficult decision to cut the test and revert to our previous upsell approach.
KEY LEARNINGS
KEY LEARNINGS
A lasting influence on TurboTax's upsell approach
A lasting influence on TurboTax's upsell approach
Rather than viewing this as a failed project, we treated it as a valuable learning opportunity. As a team, we regrouped and brainstormed next steps, shifting our focus to:
Rather than viewing this as a failed project, we treated it as a valuable learning opportunity. As a team, we regrouped and brainstormed next steps, shifting our focus to:
Optimizing our
AI model
Optimizing our
AI model
Optimizing our
AI model
To better predict the ideal moments of our upsells
To better predict the ideal moments of our upsells
Exploring new interaction models
Exploring new interaction models
Exploring new interaction models
That are less invasive and more contexually relevant
That are less invasive and more contexually relevant
Refining our messaging strategy
Refining our messaging strategy
Refining our messaging strategy
To ensure upsells feel helpful rather than promotional
To ensure upsells feel helpful rather than promotional
REFLECTION
REFLECTION
Turning setbacks into strategic wins
Turning setbacks into strategic wins
Not every project gets launched, but great design is about solving the right problems, not just shipping features.
This experience showed me how important data-driven design is in cross-functional teams. By using user insights, we pushed for meaningful changes, even when there was resistance. More importantly, we set the stage for smarter, less intrusive upsells that will improve the user experience in the long run.
Not every project gets launched, but great design is about solving the right problems, not just shipping features.
This experience showed me how important data-driven design is in cross-functional teams. By using user insights, we pushed for meaningful changes, even when there was resistance. More importantly, we set the stage for smarter, less intrusive upsells that will improve the user experience in the long run.
