Streamlining the subscription purchase flow at Later
Improving the purchase experience for Later's prospect customers while increasing conversion, one experiment at a time.
For years, upgrading to a higher subscription tier on Later was a complex, multi-step process that disrupted users' core task of scheduling social posts.
Challenge
55% of freemium users are abandoning the purchase flow
Our goal is to streamline the subscription experience, beginning from starting a 14-day trial to becoming a repeat Later user.
From start to finish, I led: research support, experimentation strategy and ideation, and end-to-end design. Collaborating with leaders at Later to influence the strategic vision and product roadmap, the goal was to simplify the checkout process to start a trial.
Role
Product designer
Year
2022
Disciplines
Interaction & experience design
A/B test strategy
Data gathering
Roadmap planning
What
impact
did we make?
adoption
+38%
new trial starts
conversion
+23%
completed orders
usability
<42 sec
avg. spend time on checkout
Quick preview only. See the rest of what we launched below ☺
Freemium Woes
Freemium Woes
What's happening to
What's happening to
freemium users
freemium users
?
?
Only freemium users can see upgrades for paid features.
From scheduling a video on the post builder alone, there are at least 7 trial start entry points a freemium user would encounter.
By design, freemium users on Later get to explore, build trust, and see the product's value on their own terms. Upgrade paths are intentionally discoverable when going through Later's core task of scheduling social media posts on the post builder.
However, 55% of the sample group chose to abandon the upgrade instead…
…and we don't know why yet.
To start a premium trial, the freemium user will lose their scheduling work-in-progress.
Patterns in User Behaviour
Patterns in User Behaviour
What did
What did
user behaviour
user behaviour
reveal?
reveal?
1.
The substantial friction occurs even before reaching the checkout
We only scratched the surface when learning that over 50% of the total traffic drops off at the post builder (instead of during checkout, which is a more common scenario in a typical purchase flow).
2.
Different segments prioritize different features
Later's users vary from businesses to individuals and agencies. It turns out that upgrades are typically motivated by one specific feature, rather than a bundle of different features within a premium plan.
3.
Banner-blindness at the second most-used upgrade entry point
Insight
Later's nav bar upgrade path showed limited effectiveness, likely due to banner blindness and that freemium users aren't thinking about upgrading for a bunch of features at once.
Limited performance of Later’s nav bar upgrade path suggested two issues: users tuned it out due to banner blindness, and freemium users typically upgrade for one key feature rather than a broad package.
Implication
Together with the early drop-off metric, this reinforces that effective upgrade prompts need to be contextual and low-friction, surfacing at the moment of user intent without requiring extra effort or breaking flow.
4.
Prospects and freemium users revealed the upgrade was risky
Understanding concerns from prospects and freemium users going through the checkout upgrade within the last 2 weeks.
When we asked freemium users and prospect customers who dropped off from starting a trial within the last 2 weeks, much of the apprehension came and delay resulted from not wanting to lose their work.
One important validation that came out of this is that there is interest for starting a trial, but how it's currently done in-app needs some re-thinking.
Additionally, we found some of the interviews suggesting that there are other trial start paths outside of the post builder that should be reviewed.
Here's my main hypothesis:
A one-step purchase flow
will increase checkout completion rate
The thinking
If we replace the current multi-page checkout with a reduced flow, completion rates will increase because users can upgrade without leaving their workflow.
Success metrics
Completed orders increases by 12%
Trial start rate increases by at least 20%
Time to completed order decreases by 1 minute
Tal Raviv's framework came in really handy when we were asking ourselves if we should A/B test a new purchase flow.
Shaping Modal Checkout
Shaping Modal Checkout
How did we get to a
modal checkout
?
What did
stakeholders say
?
Understanding the landscape with stakeholders was a non-negotiable
The checkout process isn't as easy as it looks. Considerations regarding plan types, pricing points, upgrade paths, and edge cases had to be well-understood, requiring collaboration with developers and other designers early. Throughout regular check-ins, I began to understand what we want to prioritize for our test from the business and technical standpoints, and learned new insights that could be turned into future improvements.
One point that kept being brought up was that we know that the existing checkout and upgrade modals issues were highly prioritized as they were user-facing.
Quantifying the unquantifiable
With the help of the psych framework, our team of three (UX researcher, design systems lead, and I) scored the customer-facing portion of the subscription funnel.
This step was helpful for getting into the mindset of ideation. Breaking down and scoring different parts of the journey was essential to ideate the endless possibilities of solutions and prioritizing them.
What we noticed
Lack of visual and information hierarchy
Too many factors driving prospects away from making a purchase
Many redundant steps
Inconsistency in upgrade paths
A modal-based checkout, checks out
We explored and came down to three potential solutions for a new checkout display. Ultimately, with scalability and feasibility considerations in mind, we opted to further develop a modal-based checkout.
Taking the early iterations from our initial ideation workshop, I went through multiple iterations and feedback sessions with developers and designers to ensure our learnings and constraints are met. While this was well-received at the start, it eventually led to some issues.
Challenge
Engineers and I kept running into misunderstandings while we were iterating the solution.
How did I overcome it?
Instead of gathering everyone on my feature team every single time questions and updates arose, I worked one-on-one with our lead developer first.
This created a strong partnership that kept progress aligned, allowed us to quickly resolve blockers, and ensured the rest of the team received only the most relevant, timely updates (reducing the information overload for engineers; after all, everyone retains new information differently).
The ideation workshop dove deeper into each possible option. We discussed the pros and cons before voting on how the checkout would look and feel like, resulting in the selection of Option 2: Modal Checkout.
The Purchase Flow A/B Test
The Purchase Flow A/B Test
Trial starts and completed orders increased by 38% and 23% respectively
Over a three-week experiment involving more than 1,000 users interacting with both variations of the launch, the modal checkout generated strong positive results. These outcomes supported its adoption as the standard checkout pattern across the Later platform, demonstrating its scalability as the company expanded.
The updated checkout now handles complex use cases, including AI Credit purchases, address validation, and tax collection (read more here), while also enabling future optimization opportunities (read below).
Our team's growth mindset treats major launches (the new checkout) as a foundation, creating momentum for ongoing testing and iteration rather than a finished endpoint.
Further Optimization Efforts Post-Launch
After launching the new checkout design and validating it through our initial A/B test, we shifted focus to smaller, high-impact experiments in the spirit of growth-thinking. These optimizations built on the new foundation and further reduced friction in the subscription flow.
14-Day Trial Messaging
Problem
Users were hesitant to start a trial because they assumed their card would be charged immediately.
Research also revealed that 27% of the hesitation came from concerns about forgetting to cancel and being charged later.
Change
Inspired by Jaycee Day's experiment at Blinkist, we tested a version of the create account step with a goal of educating prospective customers what a 14-day trial entails.
Impact
Increased new account sign-ups by 35%
Increased checkout conversion by 18%
Dynamic Address Form
Problem
Usability testing showed users did not see the correllation between credit card number and mailing address when entering their information, making a lot of the saved addresses invalid for taxation.
Change
We simplified the step by automatically pre-selecting and displaying the correct address form based on the country detected from the credit card input.
Impact
Reduced invalid address entries by 40%
Shortened checkout completion time by 30 seconds
Gave confidence to roll out the feature to other countries as part of future taxation initiatives
Responsive to mobile browser
Problem
Analytics revealed a significant share of trial start traffic was coming from mobile browser, but our checkout was only optimized for desktop.
Change
We designed a responsive mobile web version.
Impact
Improved checkout completion rates on mobile web browsers by 29%
Drove >38% lift in subscription revenue from mobile users
Demonstrated the value of optimizing for mobile, sparking discussions on implementing an in-app checkout for Later's mobile app
For a better experience, see this on a larger screen ☺
For a better experience, see this on a larger screen ☺
Further Optimization Efforts Post-Launch
UP NEXT
Automating sales tax collection
View case study
UP NEXT
Automating sales tax collection
View case study
View case study
UP NEXT
Automating sales tax collection
View case study
View case study