How I designed 'Find Similar Stores', a tool to help retailers predict a new store's performance
How I designed 'Find Similar Stores', a tool to help retailers predict a new store's performance
role
Founding Designer
Founding Designer
Collaborated with
ML Engineers, CPO, CEO
ML Engineers, CPO, CEO
Finding Analogs lets businesses compare a potential location to similar, proven sites-helping predict performance and make smarter, lower-risk decisions.
Finding Analogs lets businesses compare a potential location to similar, proven sites-helping predict performance and make smarter, lower-risk decisions.
Problem Statement
Retailers struggle to find accurate and relevant analogs to compare potential locations, often relying on inconsistent data sources, making it difficult to assess a site’s potential.
Retailers struggle to find accurate and relevant analogs to compare potential locations, often relying on inconsistent data sources, making it difficult to assess a site’s potential.
Context and human psychology at play:
Anchoring Bias,
Social Proof,
Familiarity.
Ashby, Greenhouse, Workday are all recruiting platforms in the industry. Imagine there's a brand new, 4th platform in the picture.
Ashby, Greenhouse, Workday are all recruiting platforms in the industry. Imagine there's a brand new, 4th platform in the picture.
Fig: Nobody likes Workday. That's the lesson here.
Fig: Nobody likes Workday.
That's the lesson here.
If I tell you a "New Platform" is similar to a platform that's widely disliked, you're not going to spend time learning about the "New Platform".
If I tell you a "New Platform" is similar to a platform that's widely disliked, you're not going to spend time learning about the "New Platform".
Human psychology means users don’t evaluate new products in a vacuum—they compare, anchor, and judge based on prior experiences and emotional responses.
Human psychology means users don’t evaluate new products in a vacuum—they compare, anchor, and judge based on prior experiences and emotional responses.
How does this translate for retailers? Let's explore :)
How does this translate for retailers? Let's explore :)
Analogs is industry term.
How does it help the retailers?
Analogs is industry term.
How does it help the retailers?
Analogs is industry term.
How does it help the retailers?
Retailers face high uncertainty when evaluating new store locations, since it’s hard to predict how a site will perform before opening.
Retailers face high uncertainty when evaluating new store locations, since it’s hard to predict how a site will perform before opening.
Fig: similar stores are called analogs
Fig: similar stores are called analogs
Earlier, it used to be based on gut feeling -- which, let's be honest doesn't cut it anymore.
Earlier, it used to be based on gut feeling -- which, let's be honest doesn't cut it anymore.
When they see that a new store is similar to their 3 (or x number) high-performing stores, their confidence increases in it. They know they can invest time in learning more about the new store.
When they see that a new store is similar to their 3 (or x number) high-performing stores, their confidence increases in it. They know they can invest time in learning more about the new store.
It helps them with:
predicting new store sales
forecast its performance
set realistic expectations
It helps them with:
predicting new store sales
forecast its performance
set realistic expectations
Challenge #1: Technical
Defining what makes two stores “similar” is a complex, data-driven problem.
Our ML engineers built a model that compares locations based on factors 3 main factors (NDA, sorry).
Defining what makes two stores “similar” is a complex, data-driven problem.
Our ML engineers built a model that compares locations based on factors 3 main factors (NDA, sorry).
To design the right user experience, I had to dive deep into how the model worked-learning about its inputs, how it weighed different attributes, and what its limitations were.
To design the right user experience, I had to dive deep into how the model worked-learning about its inputs, how it weighed different attributes, and what its limitations were.
Challenge #2: Translating complexity to the users
While the model could surface hundreds of potential analogs, I knew users needed clarity, not complexity.
I worked to prioritize results based on what mattered most to retailers i.e. which factors do they care most about?
While the model could surface hundreds of potential analogs, I knew users needed clarity, not complexity.
I worked to prioritize results based on what mattered most to retailers i.e. which factors do they care most about?
I collaborated with the ML team to tune the ranking, and designing a simple interface that highlighted the most relevant matches, so users could make quick, confident decisions.
I collaborated with the ML team to tune the ranking, and designing a simple interface that highlighted the most relevant matches, so users could make quick, confident decisions.
Typical User Journey
Users have a list of addresses from the brokers that they need to investigate into. They use our tool first, for Site Evaluation, and if it passes their base set of requirements, they look into the 'analogs factor'. So there's essentially 2 kinds of input:
Users have a list of addresses from the brokers that they need to investigate into. They use our tool first, for Site Evaluation, and if it passes their base set of requirements, they look into the 'analogs factor'. So there's essentially 2 kinds of input:
Fig: Broadly, 2 kinds of inputs
Fig: Broadly, 2 kinds of inputs
For every new store, if the user gets 'these are x no. of similar stores' + factors on which they're deemed similar, that would help the user in gauging the store's future performance.
For every new store, if the user gets 'these are x no. of similar stores' + factors on which they're deemed similar, that would help the user in gauging the store's future performance.
Iterative Solutions
My iterative solution for a user to find all necessary information when they look new site's similarity to existing sites ->
My iterative solution for a user to find all necessary information when they look new site's similarity to existing sites ->
This is also taking into account what factors matter most to the user and which ones do they want to see every time they look at analogs
This is also taking into account what factors matter most to the user and which ones do they want to see every time they look at analogs
Feedback & Tests
We're building solutions, testing and iterating continuously but so far, our customers have been giving us 100% positive feedback
We're building solutions, testing and iterating continuously but so far, our customers have been giving us 100% positive feedback
Next:
Next:
See how I revamped the website, boosting sign-ups by 166% and revenue to $26,000