@
Designing a Smarter,
Personalized Recommendation Experience
Imagine having a smarter, more personalized recommendation experience when checking out for a new meal plan. Instead of being recommended the cheapest plan(previously), your plan is personally tailored for your dog’s calorie group, intolerances, and plan preferences.
1.) Plan preferences question
Previously plan type recommendation was based soley on pricing. This new plan preferences allows for the best plan based on multiple factors
Increased plan type recommendation accuracy by more than 70%
2.) Calculating Screen
Utilizing the psychological principle of "The Labor Illusion Effect". A calculating page was created before recommended plan type to increase trust and personalization.
increased conversion rates by 5%.
3.) Combined Plans Page
After the recommendation algorithm was upgraded and calculating screen created, we combined the plans and entrees page to recommend the best plan and entree, ultimately to decrease decision making.
Drop off decreased by 30%. Time spent on page decreased by 65%.
FreshCooked not included
Despite Freshcooked having the highest checkout in all calorie groups, FreshCooked was not included in the recommendation algorithm at all.
PetPlate did not include FreshCooked to the recommendation despite historically, this plan had the highest checkout before the recommendation UI was added.
Low recommendation accuracy
The current recommendation algorithm proved inefficient as shown by the numbers before.
If PetPlate has different plans for multiple use cases, why aren't we recommending the right plan based on different use cases?
No clear basis for recommendation algorithm.
No research or data analysis was done to determine what plan to recommend the users. There is no question in the subscription flow that helps determine what plan is best for the user. The only factor PetPlate decided to consider was pricing.
To Summarize
There are multiple use cases for each plan. Pricing is just one of many aspects of why a user may pick a specific plan. Section 3 covers the UX Research that went behind how the question was created + the new point based system that covers multiple use cases of the best plan
UX Psychology at play
Adding a “Calculating Screen" isn’t just about aesthetics — it’s a psychologically informed UX decision that increases trust, engagement, and perceived personalization.
The Effort Heuristic:
Labor Illusion Effect:
Showing “effort” — even simulated effort — increases trust and satisfaction with the result.
Zeigarnik Effect:
People remember and stay engaged with unfinished experiences.
Competitor Analysis
Here are just a few of competitors and other companies in different industries that take advantage of these principles.
Calculating Page
New calculating screen that summarizes previous entries users inputted. Utilizing "The Effort Heuristic" and "The Labor Illusion Effect."
Drop Off Rates
The Plans and Recipe page saw the highest drop off rates consistently month over month.
Average Drop off rate Q1-Q2 2025
Hypothesis: Users are facing decision fatigue.
-After answering 10+ questions, users are forced to choose their own plan.
Time Spent on Page
The Plans and Recipe page had on average 250% higher time spent on page per second.
Hypothesis
Users are facing information overload
Users have to choose from up to 4 different plans and 8 different entrees.
Combined Plans & Recipe page.
Using the improved algorithm, we choose the best plan for our users.
Users still have the option to change plans
Section 3: The Process
For an in depth case study showing the process and detailed changes of the flow, please reach out :)