How I used ChatGPT to make letter writing feel fun; decreasing time to completion by 91% and improving experience ratings by 23%.
How I used ChatGPT to make letter writing feel fun; decreasing time to completion by 91% and improving experience ratings by 23%.
How I used ChatGPT to make letter writing feel fun; decreasing time to completion by 91% and improving experience ratings by 23%.
How I used ChatGPT to make letter writing feel fun; decreasing time to completion by 91% and improving experience ratings by 23%.
How I used ChatGPT to make letter writing feel fun; decreasing time to completion by 91% and improving experience ratings by 23%.
How I used ChatGPT to make letter writing feel fun; decreasing time to completion by 91% and improving experience ratings by 23%.
My role
I was the sole designer working on the proof-of-concept. I did initial user research and data analysis, built and tested a working prototype and pitched the new approach to the wider organization.
I was the sole designer working on the proof-of-concept. I did initial user research and data analysis, built and tested a working prototype and pitched the new approach to the wider organization.
I was the sole designer working on the proof-of-concept. I did initial user research and data analysis, built and tested a working prototype and pitched the new approach to the wider organization.
I was the sole designer working on the proof-of-concept. I did initial user research and data analysis, built and tested a working prototype and pitched the new approach to the wider organization.
I was the sole designer working on the proof-of-concept. I did initial user research and data analysis, built and tested a working prototype and pitched the new approach to the wider organization.
I was the sole designer working on the proof-of-concept. I did initial user research and data analysis, built and tested a working prototype and pitched the new approach to the wider organization.
Summary
In spring 2023, I tested a new letter writing approach for Compassion Canada, enlisting a conversational interface and chatGPT. I used surveys, user interviews and market research to find the right problem along with prototyping and user testing to find the right solution. In trials, key experience ratings were boosted by 23%. Also, the time it took users to complete their first letter plummeted by more than 91%! 🤯
In spring 2023, I tested a new letter writing approach for Compassion Canada, enlisting a conversational interface and chatGPT. I used surveys, user interviews and market research to find the right problem along with prototyping and user testing to find the right solution. In trials, key experience ratings were boosted by 23%. Also, the time it took users to complete their first letter plummeted by more than 91%! 🤯
In spring 2023, I tested a new letter writing approach for Compassion Canada, enlisting a conversational interface and chatGPT. I used surveys, user interviews and market research to find the right problem along with prototyping and user testing to find the right solution. In trials, key experience ratings were boosted by 23%. Also, the time it took users to complete their first letter plummeted by more than 91%! 🤯
In spring 2023, I tested a new letter writing approach for Compassion Canada, enlisting a conversational interface and chatGPT. I used surveys, user interviews and market research to find the right problem along with prototyping and user testing to find the right solution. In trials, key experience ratings were boosted by 23%. Also, the time it took users to complete their first letter plummeted by more than 91%! 🤯
In spring 2023, I tested a new letter writing approach for Compassion Canada, enlisting a conversational interface and chatGPT. I used surveys, user interviews and market research to find the right problem along with prototyping and user testing to find the right solution. In trials, key experience ratings were boosted by 23%. Also, the time it took users to complete their first letter plummeted by more than 91%! 🤯
In spring 2023, I tested a new letter writing approach for Compassion Canada, enlisting a conversational interface and chatGPT. I used surveys, user interviews and market research to find the right problem along with prototyping and user testing to find the right solution. In trials, key experience ratings were boosted by 23%. Also, the time it took users to complete their first letter plummeted by more than 91%! 🤯
Context
Compassion is a child development non-profit that works in 27 countries around world. They empower children to break free from the cycle of poverty through community lead education, health care and mentorship. Though the work is co-ordinated and carried out locally, child “sponsors” can play a big role too.
Compassion is a child development non-profit that works in 27 countries around world. They empower children to break free from the cycle of poverty through community lead education, health care and mentorship. Though the work is co-ordinated and carried out locally, child “sponsors” can play a big role too.
Beyond financial support, one of the most powerful ways they can encourage children is through writing letters to them. So many children and Compassion alumni talk about how life changing it is to know that someone so far away cares for and believes in them. On the other side of the experience, sponsors who write letters have a much stronger connection with the children they support and will sustain their sponsorship for a longer period than those that don’t write.
Compassion is a child development non-profit that works in 27 countries around world. They empower children to break free from the cycle of poverty through community lead education, health care and mentorship. Though the work is co-ordinated and carried out locally, child “sponsors” can play a big role too.
Compassion is a child development non-profit that works in 27 countries around world. They empower children to break free from the cycle of poverty through community lead education, health care and mentorship. Though the work is co-ordinated and carried out locally, child “sponsors” can play a big role too.
Compassion is a child development non-profit that works in 27 countries around world. They empower children to break free from the cycle of poverty through community lead education, health care and mentorship. Though the work is co-ordinated and carried out locally, child “sponsors” can play a big role too.
Compassion is a child development non-profit that works in 27 countries around world. They empower children to break free from the cycle of poverty through community lead education, health care and mentorship. Though the work is co-ordinated and carried out locally, child “sponsors” can play a big role too.
Beyond financial support, one of the most powerful ways they can encourage children is through writing letters to them. So many children and Compassion alumni talk about how life changing it is to know that someone so far away cares for and believes in them. On the other side of the experience, sponsors who write letters have a much stronger connection with the children they support and will sustain their sponsorship for a longer period than those that don’t write.
61%
Of sponsored children would rather receive a letter than a gift
61%
Of sponsored children would rather receive a letter than a gift
61%
Of sponsored children would rather receive a letter than a gift
32%
Decrease in supporter correspondence rates over 7 years
32%
Decrease in supporter correspondence rates over 7 years
32%
Decrease in supporter correspondence rates over 7 years
Beyond financial support, one of the most powerful ways they can encourage children is through writing letters to them. So many children and Compassion alumni talk about how life changing it is to know that someone so far away cares for and believes in them. On the other side of the experience, sponsors who write letters have a much stronger connection with the children they support and will sustain their sponsorship for a longer period than those that don’t write.
Beyond financial support, one of the most powerful ways they can encourage children is through writing letters to them. So many children and Compassion alumni talk about how life changing it is to know that someone so far away cares for and believes in them. On the other side of the experience, sponsors who write letters have a much stronger connection with the children they support and will sustain their sponsorship for a longer period than those that don’t write.
Beyond financial support, one of the most powerful ways they can encourage children is through writing letters to them. So many children and Compassion alumni talk about how life changing it is to know that someone so far away cares for and believes in them. On the other side of the experience, sponsors who write letters have a much stronger connection with the children they support and will sustain their sponsorship for a longer period than those that don’t write.
Beyond financial support, one of the most powerful ways they can encourage children is through writing letters to them. So many children and Compassion alumni talk about how life changing it is to know that someone so far away cares for and believes in them. On the other side of the experience, sponsors who write letters have a much stronger connection with the children they support and will sustain their sponsorship for a longer period than those that don’t write.
In Canada, however, convincing supporters to write letters has been increasingly difficult. Over the last 7 years their overall correspondence rate (writing at least one letter per year) has decreased by 32%, dropping from 71% to 48%.
In Canada, however, convincing supporters to write letters has been increasingly difficult. Over the last 7 years their overall correspondence rate (writing at least one letter per year) has decreased by 32%, dropping from 71% to 48%.
Convinced that letter writing can be an incredibly affirming experience for both sponsors and children, with a strong business benefit as well, I used my weekly, “innovation” time allotment to start to dig into what was keeping our supporters from writing.
In Canada, however, convincing supporters to write letters has been increasingly difficult. Over the last 7 years their overall correspondence rate (writing at least one letter per year) has decreased by 32%, dropping from 71% to 48%.
In Canada, however, convincing supporters to write letters has been increasingly difficult. Over the last 7 years their overall correspondence rate (writing at least one letter per year) has decreased by 32%, dropping from 71% to 48%.
In Canada, however, convincing supporters to write letters has been increasingly difficult. Over the last 7 years their overall correspondence rate (writing at least one letter per year) has decreased by 32%, dropping from 71% to 48%.
In Canada, however, convincing supporters to write letters has been increasingly difficult. Over the last 7 years their overall correspondence rate (writing at least one letter per year) has decreased by 32%, dropping from 71% to 48%.
Convinced that letter writing can be an incredibly affirming experience for both sponsors and children, with a strong business benefit as well, I used my weekly, “innovation” time allotment to start to dig into what was keeping our supporters from writing.
Convinced that letter writing can be an incredibly affirming experience for both sponsors and children, with a strong business benefit as well, I used my weekly, “innovation” time allotment, a long with time outside of work, to start to dig into what was keeping our supporters from writing.
Convinced that letter writing can be an incredibly affirming experience for both sponsors and children, with a strong business benefit as well, I used my weekly, “innovation” time allotment, a long with time outside of work, to start to dig into what was keeping our supporters from writing.
Convinced that letter writing can be an incredibly affirming experience for both sponsors and children, with a strong business benefit as well, I used my weekly, “innovation” time allotment, a long with time outside of work, to start to dig into what was keeping our supporters from writing.
Convinced that letter writing can be an incredibly affirming experience for both sponsors and children, with a strong business benefit as well, I used my weekly, “innovation” time allotment, a long with time outside of work, to start to dig into what was keeping our supporters from writing.
61%
Of sponsored children would rather receive a letter than a gift
61%
Of sponsored children would rather receive a letter than a gift
61%
Of sponsored children would rather receive a letter than a gift
61%
Of sponsored children would rather receive a letter than a gift
32%
Decrease in supporter correspondence rates over 7 years
32%
Decrease in supporter correspondence rates over 7 years
32%
Decrease in supporter correspondence rates over 7 years
32%
Decrease in supporter correspondence rates over 7 years
The process
Given the limited time frame and resources I had at my disposal, I took a lean approach to the "double diamond" process. When it came to finding the right problem, I gathered qualitative data from existing research and kept user research to surveys and short user interviews.
For the same reason, when it came to finding the right solution, I chose to test a single, functional prototype against our current letter writing process. The new approach I was proposing was radically different from our existing one and I wanted to prove that it would lead significant gains in key metrics. If the new design proved a success, I could bring this data to our leadership to get support for further, iterative testing and continued product development.
Given the limited time frame and resources I had at my disposal, I took a lean approach to the "double diamond" process. When it came to finding the right problem, I gathered qualitative data from existing research and kept user research to surveys and short user interviews.
For the same reason, when it came to finding the right solution, I chose to test a single, functional prototype against our current letter writing process. The new approach I was proposing was radically different from our existing one and I wanted to prove that it would lead significant gains in key metrics. If the new design proved a success, I could bring this data to our leadership to get support for further, iterative testing and continued product development.
Given the limited time frame and resources I had at my disposal, I took a lean approach to the "double diamond" process. When it came to finding the right problem, I gathered qualitative data from existing research and kept user research to surveys and short user interviews.
For the same reason, when it came to finding the right solution, I chose to test a single, functional prototype against our current letter writing process. The new approach I was proposing was radically different from our existing one and I wanted to prove that it would lead significant gains in key metrics. If the new design proved a success, I could bring this data to our leadership to get support for further, iterative testing and continued product development.
Given the limited time frame and resources I had at my disposal, I took a lean approach to the "double diamond" process. When it came to finding the right problem, I gathered qualitative data from existing research and kept user research to surveys and short user interviews.
For the same reason, when it came to finding the right solution, I chose to test a single, functional prototype against our current letter writing process. The new approach I was proposing was radically different from our existing one and I wanted to prove that it would lead significant gains in key metrics. If the new design proved a success, I could bring this data to our leadership to get support for further, iterative testing and continued product development.
Given the limited time frame and resources I had at my disposal, I took a lean approach to the "double diamond" process. When it came to finding the right problem, I gathered qualitative data from existing research and kept user research to surveys and short user interviews.
For the same reason, when it came to finding the right solution, I chose to test a single, functional prototype against our current letter writing process. The new approach I was proposing was radically different from our existing one and I wanted to prove that it would lead significant gains in key metrics. If the new design proved a success, I could bring this data to our leadership to get support for further, iterative testing and continued product development.
Given the limited time frame and resources I had at my disposal, I took a lean approach to the "double diamond" process. When it came to finding the right problem, I gathered qualitative data from existing research and kept user research to surveys and short user interviews.
For the same reason, when it came to finding the right solution, I chose to test a single, functional prototype against our current letter writing process. The new approach I was proposing was radically different from our existing one and I wanted to prove that it would lead significant gains in key metrics. If the new design proved a success, I could bring this data to our leadership to get support for further, iterative testing and continued product development.
Finding the right problem
Finding the right problem
Finding the right problem
Finding the right problem
Finding the right problem
Finding the right problem
Surveys
Interviews
Research
Hypothesis
Surveys
Interviews
Research
Hypothesis
Surveys
Interviews
Research
Hypothesis
Surveys
Interviews
Research
Hypothesis
Surveys
Interviews
Research
Hypothesis
Surveys
Interviews
Research
Hypothesis
Finding the right solution
Finding the right solution
Finding the right solution
Finding the right solution
Finding the right solution
Finding the right solution
Flows
Prototype
User tests
Analysis & Next steps
Flows
Prototype
User tests
Analysis & Next steps
Flows
Prototype
User tests
Analysis & Next steps
Flows
Prototype
User tests
Analysis & Next steps
Flows
Prototype
User tests
Analysis & Next steps
Flows
Prototype
User tests
Analysis & Next steps
Finding the right problem
I wanted to establish a better understanding as to why users were struggling to write letters, so I started off by conducting some user research through surveys and interviews. I also gathered some qualitative data about how Canadians prefer to communicate in their everyday lives.
I wanted to establish a better understanding as to why users were struggling to write letters, so I started off by conducting some user research through surveys and interviews. I also gathered some qualitative data about how Canadians prefer to communicate in their everyday lives.
I wanted to establish a better understanding as to why users were struggling to write letters, so I started off by conducting some user research through surveys and interviews. I also gathered some qualitative data about how Canadians prefer to communicate in their everyday lives.
I wanted to establish a better understanding as to why users were struggling to write letters, so I started off by conducting some user research through surveys and interviews. I also gathered some qualitative data about how Canadians prefer to communicate in their everyday lives.
I wanted to establish a better understanding as to why users were struggling to write letters, so I started off by conducting some user research through surveys and interviews. I also gathered some qualitative data about how Canadians prefer to communicate in their everyday lives.
I wanted to establish a better understanding as to why users were struggling to write letters, so I started off by conducting some user research through surveys and interviews. I also gathered some qualitative data about how Canadians prefer to communicate in their everyday lives.
9.5
Average number of years since respondents had last written a letter
9.5
Average number of years since respondents had last written a letter
9.5
Average number of years since respondents had last written a letter
I based my user survey on a single question: “Outside of Compassion, when was the last time you wrote a letter?” The results were telling; on average, respondents hadn’t written a letter of any kind in 9.5 years. I was able to follow-up with some participants which revealed key insights:
I based my user survey on a single question: “Outside of Compassion, when was the last time you wrote a letter?” The results were telling; on average, respondents hadn’t written a letter of any kind in 9.5 years. I was able to follow-up with some participants which revealed key insights:
I based my user survey on a single question: “Outside of Compassion, when was the last time you wrote a letter?” The results were telling; on average, respondents hadn’t written a letter of any kind in 9.5 years. I was able to follow-up with some participants which revealed key insights:
I based my user survey on a single question: “Outside of Compassion, when was the last time you wrote a letter?” The results were telling; on average, respondents hadn’t written a letter of any kind in 9.5 years. I was able to follow-up with some participants which revealed key insights:
I based my user survey on a single question: “Outside of Compassion, when was the last time you wrote a letter?” The results were telling; on average, respondents hadn’t written a letter of any kind in 9.5 years. I was able to follow-up with some participants which revealed key insights:
I based my user survey on a single question: “Outside of Compassion, when was the last time you wrote a letter?” The results were telling; on average, respondents hadn’t written a letter of any kind in 9.5 years. I was able to follow-up with some participants which revealed key insights:
9.5
Average number of years since respondents had last written a letter
9.5
Average number of years since respondents had last written a letter
9.5
Average number of years since respondents had last written a letter
9.5
Average number of years since respondents had last written a letter
“I don’t know what’s appropriate to talk about with a child in a different context than me.”
James, Current Sponsor
“I don’t know what’s appropriate to talk about with a child in a different context than me.”
James, Current Sponsor
“I don’t know what’s appropriate to talk about with a child in a different context than me.”
James, Current Sponsor
“I don’t know what’s appropriate to talk about with a child in a different context than me.”
James, Current Sponsor
“I don’t know what’s appropriate to talk about with a child in a different context than me.”
James, Current Sponsor
“I don’t know what’s appropriate to talk about with a child in a different context than me.”
James, Current Sponsor
“Writing a letter is physically painful for my brain.”
Theresa, Current Sponsor
“Writing a letter is physically painful for my brain.”
Theresa, Current Sponsor
“Writing a letter is physically painful for my brain.”
Theresa, Current Sponsor
“Writing a letter is physically painful for my brain.”
Theresa, Current Sponsor
“Writing a letter is physically painful for my brain.”
Theresa, Current Sponsor
“Writing a letter is physically painful for my brain.”
Theresa, Current Sponsor
When it came to qualitative research, I looked into the current, most popular platforms for personal communication in Canada. According to the CRTC, texting is the preferred mode of communication for most Canadians. On average, individuals sent 103 texts a day in 2020! Social media, email, instant messaging and video chat followed, respectively. Unsurprisingly, letters were nowhere to be found.
When it came to qualitative research, I looked into the current, most popular platforms for personal communication in Canada. According to the CRTC, texting is the preferred mode of communication for most Canadians. On average, individuals sent 103 texts a day in 2020! Social media, email, instant messaging and video chat followed, respectively. Unsurprisingly, letters were nowhere to be found.
When it came to qualitative research, I looked into the current, most popular platforms for personal communication in Canada. According to the CRTC, texting is the preferred mode of communication for most Canadians. On average, individuals sent 103 texts a day in 2020! Social media, email, instant messaging and video chat followed, respectively. Unsurprisingly, letters were nowhere to be found.
When it came to qualitative research, I looked into the current, most popular platforms for personal communication in Canada. According to the CRTC, texting is the preferred mode of communication for most Canadians. On average, individuals sent 103 texts a day in 2020! Social media, email, instant messaging and video chat followed, respectively. Unsurprisingly, letters were nowhere to be found.
When it came to qualitative research, I looked into the current, most popular platforms for personal communication in Canada. According to the CRTC, texting is the preferred mode of communication for most Canadians. On average, individuals sent 103 texts a day in 2020! Social media, email, instant messaging and video chat followed, respectively. Unsurprisingly, letters were nowhere to be found.
When it came to qualitative research, I looked into the current, most popular platforms for personal communication in Canada. According to the CRTC, texting is the preferred mode of communication for most Canadians. On average, individuals sent 103 texts a day in 2020! Social media, email, instant messaging and video chat followed, respectively. Unsurprisingly, letters were nowhere to be found.
103
Average number of texts individual Canadians send everyday
103
Average number of texts individual Canadians send everyday
103
Average number of texts individual Canadians send everyday
Looking at all the research, a few problems stuck out. First of all, the skill of letter writing has been lost as people move to faster, short-form communication platforms. This means that Compassion’s current mode of letter writing can be incredibly daunting for users. Especially if we don’t give them any guidance as they learn how to do it.
On top of this, users are used to spending a minute or two writing short form messages and communicating conversationally in small chunks. It’s a hard sell convincing them to dedicate 20-30 minutes to write a long-form letter in one go.
This lead me to my initial hypothesis: If we could make our letter writing process feel faster and more familiar, this could significantly reduce the intimidation factor of writing a first letter and improve our experience ratings across the board.
Looking at all the research, a few problems stuck out. First of all, the skill of letter writing has been lost as people move to faster, short-form communication platforms. This means that Compassion’s current mode of letter writing can be incredibly daunting for users. Especially if we don’t give them any guidance as they learn how to do it.
On top of this, users are used to spending a minute or two writing short form messages and communicating conversationally in small chunks. It’s a hard sell convincing them to dedicate 20-30 minutes to write a long-form letter in one go.
This lead me to my initial hypothesis: If we could make our letter writing process feel faster and more familiar, this could significantly reduce the intimidation factor of writing a first letter and improve our experience ratings across the board.
Looking at all the research, a few problems stuck out. First of all, the skill of letter writing has been lost as people move to faster, short-form communication platforms. This means that Compassion’s current mode of letter writing can be incredibly daunting for users. Especially if we don’t give them any guidance as they learn how to do it.
On top of this, users are used to spending a minute or two writing short form messages and communicating conversationally in small chunks. It’s a hard sell convincing them to dedicate 20-30 minutes to write a long-form letter in one go.
This lead me to my initial hypothesis: If we could make our letter writing process feel faster and more familiar, this could significantly reduce the intimidation factor of writing a first letter and improve our experience ratings across the board.
Looking at all the research, a few problems stuck out. First of all, the skill of letter writing has been lost as people move to faster, short-form communication platforms. This means that Compassion’s current mode of letter writing can be incredibly daunting for users. Especially if we don’t give them any guidance as they learn how to do it.
On top of this, users are used to spending a minute or two writing short form messages and communicating conversationally in small chunks. It’s a hard sell convincing them to dedicate 20-30 minutes to write a long-form letter in one go.
This lead me to my initial hypothesis: If we could make our letter writing process feel faster and more familiar, this could significantly reduce the intimidation factor of writing a first letter and improve our experience ratings across the board.
Looking at all the research, a few problems stuck out. First of all, the skill of letter writing has been lost as people move to faster, short-form communication platforms. This means that Compassion’s current mode of letter writing can be incredibly daunting for users. Especially if we don’t give them any guidance as they learn how to do it.
On top of this, users are used to spending a minute or two writing short form messages and communicating conversationally in small chunks. It’s a hard sell convincing them to dedicate 20-30 minutes to write a long-form letter in one go.
This lead me to my initial hypothesis: If we could make our letter writing process feel faster and more familiar, this could significantly reduce the intimidation factor of writing a first letter and improve our experience ratings across the board.
Looking at all the research, a few problems stuck out. First of all, the skill of letter writing has been lost as people move to faster, short-form communication platforms. This means that Compassion’s current mode of letter writing can be incredibly daunting for users. Especially if we don’t give them any guidance as they learn how to do it.
On top of this, users are used to spending a minute or two writing short form messages and communicating conversationally in small chunks. It’s a hard sell convincing them to dedicate 20-30 minutes to write a long-form letter in one go.
This lead me to my initial hypothesis: If we could make our letter writing process feel faster and more familiar, this could significantly reduce the intimidation factor of writing a first letter and improve our experience ratings across the board.
103
Average number of texts individual Canadians send everyday
103
Average number of texts individual Canadians send everyday
103
Average number of texts individual Canadians send everyday
103
Average number of texts individual Canadians send everyday
Hypothesis
Hypothesis
Hypothesis
Hypothesis
Hypothesis
Hypothesis
If we make our writing process feel faster and more like a modern, short-form conversation,
Then we could reduce intimidation factors and time-to-completion metrics, leading to improved experience ratings.
If we make our writing process feel faster and more like a modern, short-form conversation,
Then we could reduce intimidation factors and time-to-completion metrics, leading to improved experience ratings.
If we make our writing process feel faster and more like a modern, short-form conversation,
Then we could reduce intimidation factors and time-to-completion metrics, leading to improved experience ratings.
If we make our writing process feel faster and more like a modern, short-form conversation,
Then we could reduce intimidation factors and time-to-completion metrics, leading to improved experience ratings.
If we make our writing process feel faster and more like a modern, short-form conversation,
Then we could reduce intimidation factors and time-to-completion metrics, leading to improved experience ratings.
If we make our writing process feel faster and more like a modern, short-form conversation,
Then we could reduce intimidation factors and time-to-completion metrics, leading to improved experience ratings.
Finding the right solution
To begin with, I wanted to reframe the letter writing experience to be more like a text or IM conversation. A conversational UI seemed like the best way to accomplish this. As I was prepping some initial concepts, ChatGPT’s public beta was released. This was perfect timing and I eagerly incorporated ChatGPT into the new letter writing flow. The interface first guided users through the information gathering phase, then it used ChatGPT to synthesize and augment the content to make if feel like a real letter.
To begin with, I wanted to reframe the letter writing experience to be more like a text or IM conversation. A conversational UI seemed like the best way to accomplish this. As I was prepping some initial concepts, ChatGPT’s public beta was released. This was perfect timing and I eagerly incorporated ChatGPT into the new letter writing flow. The interface first guided users through the information gathering phase, then it used ChatGPT to synthesize and augment the content to make if feel like a real letter.
To begin with, I wanted to reframe the letter writing experience to be more like a text or IM conversation. A conversational UI seemed like the best way to accomplish this. As I was prepping some initial concepts, ChatGPT’s public beta was released. This was perfect timing and I eagerly incorporated ChatGPT into the new letter writing flow. The interface first guided users through the information gathering phase, then it used ChatGPT to synthesize and augment the content to make if feel like a real letter.
To begin with, I wanted to reframe the letter writing experience to be more like a text or IM conversation. A conversational UI seemed like the best way to accomplish this. As I was prepping some initial concepts, ChatGPT’s public beta was released. This was perfect timing and I eagerly incorporated ChatGPT into the new letter writing flow. The interface first guided users through the information gathering phase, then it used ChatGPT to synthesize and augment the content to make if feel like a real letter.
To begin with, I wanted to reframe the letter writing experience to be more like a text or IM conversation. A conversational UI seemed like the best way to accomplish this. As I was prepping some initial concepts, ChatGPT’s public beta was released. This was perfect timing and I eagerly incorporated ChatGPT into the new letter writing flow. The interface first guided users through the information gathering phase, then it used ChatGPT to synthesize and augment the content to make if feel like a real letter.
To begin with, I wanted to reframe the letter writing experience to be more like a text or IM conversation. A conversational UI seemed like the best way to accomplish this. As I was prepping some initial concepts, ChatGPT’s public beta was released. This was perfect timing and I eagerly incorporated ChatGPT into the new letter writing flow. The interface first guided users through the information gathering phase, then it used ChatGPT to synthesize and augment the content to make if feel like a real letter.
Flow Chart
Flow Chart
Flow Chart
I created a few quick flow charts to think through what content and functionality would strike the right balance between simplicity and flexibility. I wanted the process to be simple and quick, but I also wanted to give users the feeling that they were really participating in the process and that the letter would be unique to them. I then moved straight to designing the hifi prototype. This was only marginally slower than building a lofi wireframe and meant that a functional prototype was ready to test as quickly as possible.
I created a few quick flow charts to think through what content and functionality would strike the right balance between simplicity and flexibility. I wanted the process to be simple and quick, but I also wanted to give users the feeling that they were really participating in the process and that the letter would be unique to them. I then moved straight to designing the hifi prototype. This was only marginally slower than building a lofi wireframe and meant that a functional prototype was ready to test as quickly as possible.
I created a few quick flow charts to think through what content and functionality would strike the right balance between simplicity and flexibility. I wanted the process to be simple and quick, but I also wanted to give users the feeling that they were really participating in the process and that the letter would be unique to them. I then moved straight to designing the hifi prototype. This was only marginally slower than building a lofi wireframe and meant that a functional prototype was ready to test as quickly as possible.
I created a few quick flow charts to think through what content and functionality would strike the right balance between simplicity and flexibility. I wanted the process to be simple and quick, but I also wanted to give users the feeling that they were really participating in the process and that the letter would be unique to them. I then moved straight to designing the hifi prototype. This was only marginally slower than building a lofi wireframe and meant that a functional prototype was ready to test as quickly as possible.
I created a few quick flow charts to think through what content and functionality would strike the right balance between simplicity and flexibility. I wanted the process to be simple and quick, but I also wanted to give users the feeling that they were really participating in the process and that the letter would be unique to them. I then moved straight to designing the hifi prototype. This was only marginally slower than building a lofi wireframe and meant that a functional prototype was ready to test as quickly as possible.
I created a few quick flow charts to think through what content and functionality would strike the right balance between simplicity and flexibility. I wanted the process to be simple and quick, but I also wanted to give users the feeling that they were really participating in the process and that the letter would be unique to them. I then moved straight to designing the hifi prototype. This was only marginally slower than building a lofi wireframe and meant that a functional prototype was ready to test as quickly as possible.
Prototype
Prototype
I wanted the process to be simple and quick, but I also wanted to give users the feeling that they were really participating in the process.
I wanted the process to be simple and quick, but I also wanted to give users the feeling that they were really participating in the process.
I wanted the process to be simple and quick, but I also wanted to give users the feeling that they were really participating in the process.
I wanted the process to be simple and quick, but I also wanted to give users the feeling that they were really participating in the process.
I wanted the process to be simple and quick, but I also wanted to give users the feeling that they were really participating in the process.
I wanted the process to be simple and quick, but I also wanted to give users the feeling that they were really participating in the process.
Once the prototype was ready I kept things lean, running two unfacilitated user tests. First of all, I ran a semantic differential survey. I asked users to use a scale (eg. 1 being very complex and 6 being very simple) to rate the initial screen of both the new and original design. This would give us some good qualitative data as to whether we were moving the experience of the design in the right direction.
I also ran a unfacilitated usability test on the high fidelity prototype, asking users to walk through each step of the new process. This provided us with helpful quantitative data, specifically when it came to completion time and mis-clicks.
Once the prototype was ready I kept things lean, running two unfacilitated user tests. First of all, I ran a semantic differential survey. I asked users to use a scale (eg. 1 being very complex and 6 being very simple) to rate the initial screen of both the new and original design. This would give us some good qualitative data as to whether we were moving the experience of the design in the right direction.
I also ran a unfacilitated usability test on the high fidelity prototype, asking users to walk through each step of the new process. This provided us with helpful quantitative data, specifically when it came to completion time and mis-clicks.
Once the prototype was ready I kept things lean, running two unfacilitated user tests. First of all, I ran a semantic differential survey. I asked users to use a scale (eg. 1 being very complex and 6 being very simple) to rate the initial screen of both the new and original design. This would give us some good qualitative data as to whether we were moving the experience of the design in the right direction.
I also ran a unfacilitated usability test on the high fidelity prototype, asking users to walk through each step of the new process. This provided us with helpful quantitative data, specifically when it came to completion time and mis-clicks.
Once the prototype was ready I kept things lean, running two unfacilitated user tests. First of all, I ran a semantic differential survey. I asked users to use a scale (eg. 1 being very complex and 6 being very simple) to rate the initial screen of both the new and original design. This would give us some good qualitative data as to whether we were moving the experience of the design in the right direction.
I also ran a unfacilitated usability test on the high fidelity prototype, asking users to walk through each step of the new process. This provided us with helpful quantitative data, specifically when it came to completion time and mis-clicks.
Once the prototype was ready I kept things lean, running two unfacilitated user tests. First of all, I ran a semantic differential survey. I asked users to use a scale (eg. 1 being very complex and 6 being very simple) to rate the initial screen of both the new and original design. This would give us some good qualitative data as to whether we were moving the experience of the design in the right direction.
I also ran a unfacilitated usability test on the high fidelity prototype, asking users to walk through each step of the new process. This provided us with helpful quantitative data, specifically when it came to completion time and mis-clicks.
Once the prototype was ready I kept things lean, running two unfacilitated user tests. First of all, I ran a semantic differential survey. I asked users to use a scale (eg. 1 being very complex and 6 being very simple) to rate the initial screen of both the new and original design. This would give us some good qualitative data as to whether we were moving the experience of the design in the right direction.
I also ran a unfacilitated usability test on the high fidelity prototype, asking users to walk through each step of the new process. This provided us with helpful quantitative data, specifically when it came to completion time and mis-clicks.
Constraints and trade-offs
As mentioned early, the time and resources I had to commit to this project were quite limited. This meant that I really had to streamline research and design. I also had to tailor our next steps to work with a small team that had a lot of work to do. In end though, I actually really enjoyed embracing the challenge of remaining lean and I learned some great ways to design and test quickly.
As mentioned early, the time and resources I had to commit to this project were quite limited. This meant that I really had to streamline research and design. I also had to tailor our next steps to work with a small team that had a lot of work to do. In end though, I actually really enjoyed embracing the challenge of remaining lean and I learned some great ways to design and test quickly.
As mentioned early, the time and resources I had to commit to this project were quite limited. This meant that I really had to streamline research and design. I also had to tailor our next steps to work with a small team that had a lot of work to do. In end though, I actually really enjoyed embracing the challenge of remaining lean and I learned some great ways to design and test quickly.
As mentioned early, the time and resources I had to commit to this project were quite limited. This meant that I really had to streamline research and design. I also had to tailor our next steps to work with a small team that had a lot of work to do. In end though, I actually really enjoyed embracing the challenge of remaining lean and I learned some great ways to design and test quickly.
As mentioned early, the time and resources I had to commit to this project were quite limited. This meant that I really had to streamline research and design. I also had to tailor our next steps to work with a small team that had a lot of work to do. In end though, I actually really enjoyed embracing the challenge of remaining lean and I learned some great ways to design and test quickly.
As mentioned early, the time and resources I had to commit to this project were quite limited. This meant that I really had to streamline research and design. I also had to tailor our next steps to work with a small team that had a lot of work to do. In end though, I actually really enjoyed embracing the challenge of remaining lean and I learned some great ways to design and test quickly.
Flow Chart
Original
New
Results
The results from my prototype testing were inspiring. I wanted to make the new letter writing process feel more welcoming, fun and simple. Our semantic differential survey showed us that the conversational interface improved upon our original experience in all three categories: welcoming ( ⬆️ 11%), fun (⬆️ 26%), and simple (⬆️ 32%). On average, users rated the new experience 23% higher then the original!
The results from my prototype testing were inspiring. I wanted to make the new letter writing process feel more welcoming, fun and simple. Our semantic differential survey showed us that the conversational interface improved upon our original experience in all three categories: welcoming ( ⬆️ 11%), fun (⬆️ 26%), and simple (⬆️ 32%). On average, users rated the new experience 23% higher then the original!
The results from my prototype testing were inspiring. I wanted to make the new letter writing process feel more welcoming, fun and simple. Our semantic differential survey showed us that the conversational interface improved upon our original experience in all three categories: welcoming ( ⬆️ 11%), fun (⬆️ 26%), and simple (⬆️ 32%). On average, users rated the new experience 23% higher then the original!
The results from my prototype testing were inspiring. I wanted to make the new letter writing process feel more welcoming, fun and simple. Our semantic differential survey showed us that the conversational interface improved upon our original experience in all three categories: welcoming ( ⬆️ 11%), fun (⬆️ 26%), and simple (⬆️ 32%). On average, users rated the new experience 23% higher then the original!
The results from my prototype testing were inspiring. I wanted to make the new letter writing process feel more welcoming, fun and simple. Our semantic differential survey showed us that the conversational interface improved upon our original experience in all three categories: welcoming ( ⬆️ 11%), fun (⬆️ 26%), and simple (⬆️ 32%). On average, users rated the new experience 23% higher then the original!
11%
More welcoming
11%
More welcoming
11%
More welcoming
26%
More fun
26%
More fun
26%
More fun
32%
More simple
32%
More simple
32%
More simple
23%
Increase in overall experience ratings
23%
Increase in overall experience ratings
23%
Increase in overall experience ratings
“It was a very clean and simple interface... It looked like a typical messaging app ... so there wasn't anything too crazy or out of the ordinary to be accustomed to.”
Usability test participant
“It was a very clean and simple interface... It looked like a typical messaging app ... so there wasn't anything too crazy or out of the ordinary to be accustomed to.”
Usability test participant
“It was a very clean and simple interface... It looked like a typical messaging app ... so there wasn't anything too crazy or out of the ordinary to be accustomed to.”
“It was a very clean and simple interface... It looked like a typical messaging app ... so there wasn't anything too crazy or out of the ordinary to be accustomed to.”
“It was a very clean and simple interface... It looked like a typical messaging app ... so there wasn't anything too crazy or out of the ordinary to be accustomed to.”
Usability test participant
“It was a very clean and simple interface... It looked like a typical messaging app ... so there wasn't anything too crazy or out of the ordinary to be accustomed to.”
Usability test participant
Another massive win came in 91% percent decrease in time-to-completion. Looking at our historic analytics data, the average time it took users to write a letter using our original flow was 20m49s. Our usability tests for the new process showed us that the new process could bring that down to 1m45s, and the feedback users left showed us we were heading in the right direction.
Another massive win came in 91% percent decrease in time-to-completion. Looking at our historic analytics data, the average time it took users to write a letter using our original flow was 20m49s. Our usability tests for the new process showed us that the new process could bring that down to 1m45s, and the feedback users left showed us we were heading in the right direction.
Another massive win came in 91% percent decrease in time-to-completion. Looking at our historic analytics data, the average time it took users to write a letter using our original flow was 20m49s. Our usability tests for the new process showed us that the new process could bring that down to 1m45s, and the feedback users left showed us we were heading in the right direction.
Another massive win came in 91% percent decrease in time-to-completion. Looking at our historic analytics data, the average time it took users to write a letter using our original flow was 20m49s. Our usability tests for the new process showed us that the new process could bring that down to 1m45s, and the feedback users left showed us we were heading in the right direction.
Another massive win came in 91% percent decrease in time-to-completion. Looking at our historic analytics data, the average time it took users to write a letter using our original flow was 20m49s. Our usability tests for the new process showed us that the new process could bring that down to 1m45s, and the feedback users left showed us we were heading in the right direction.
20m49s
Original avg. time-to-completion
20m49s
Original avg. time-to-completion
20m49s
Original avg. time-to-completion
1m45s
New avg. time-to-completion
1m45s
New avg. time-to-completion
1m45s
New avg. time-to-completion
91%
Decrease in avg. time-to-completion
91%
Decrease in avg. time-to-completion
91%
Decrease in avg. time-to-completion
These metrics present a pretty strong case that, if we were to build and launch an MVP of this new process, it could go a long way to reversing the trend of decreased letter writing among supporters. And if that trend can be reversed, the effects on the children that read these letters could be, literally, be life changing.
These metrics present a pretty strong case that, if we were to build and launch an MVP of this new process, it could go a long way to reversing the trend of decreased letter writing among supporters. And if that trend can be reversed, the effects on the children that read these letters could be, literally, be life changing.
These metrics present a pretty strong case that, if we were to build and launch an MVP of this new process, it could go a long way to reversing the trend of decreased letter writing among supporters. And if that trend can be reversed, the effects on the children that read these letters could be, literally, be life changing.
These metrics present a pretty strong case that, if we were to build and launch an MVP of this new process, it could go a long way to reversing the trend of decreased letter writing among supporters. And if that trend can be reversed, the effects on the children that read these letters could be, literally, be life changing.
These metrics present a pretty strong case that, if we were to build and launch an MVP of this new process, it could go a long way to reversing the trend of decreased letter writing among supporters. And if that trend can be reversed, the effects on the children that read these letters could be, literally, be life changing.
Learnings
Adapting to user behaviour trends: The shift from long-form letters to a more conversational style of writing is reflective of broader behavioural trends. This adaptability is essential in UX design, and the project has reinforced the fact that I need to keep an eye on these shifts to inform future projects.
Feedback loops: Utilizing unfacilitated user testing platforms like maze.co and usabilityhub.com was an eye-opener. The quick turnaround times not only helped in maintaining the project's pace but also emphasized the importance of continuously incorporating user feedback for improvement.
Navigating constraints: Working within the limited time and resources provided by my "innovation" time allotment was both a challenge and a lesson in efficiency. It forced me to prioritize elements that had the most impact on user experience and business metrics. This constraint-driven design ultimately resulted in a lean but effective project, which was incredibly fulfilling.
Next steps
Compassion’s design/dev team is small and in demand, so I pitched a range of options to move forward with depending on available resources and the level of corporate buy-in:
Further testing: Refine the prototype through additional user testing and rapid prototyping to ensure we nailed down the correct MVP solution before digging into development.
Iterative integration: Gradually add conversational features and prompts to our current letter-writing process to ease users into the new system. Eg. 1. Add writing prompts 2. Simplify web app interface 3. Integrate AI assistant.
Utilize current vendor chatbot: Consider modifying our current, customer service chatbot to quickly standup a conversational letter-writing experience for limited release and testing.
Build a cross-platform app: Create a highly focused app that’s designed for future expansion, potentially replacing our existing platforms and becoming a one-stop solution for sponsors.
Compassion’s design/dev team is small and in demand, so I pitched a range of options to move forward with depending on available resources and the level of corporate buy-in:
Further testing: Refine the prototype through additional user testing and rapid prototyping to ensure we nailed down the correct MVP solution before digging into development.
Iterative integration: Gradually add conversational features and prompts to our current letter-writing process to ease users into the new system. Eg. 1. Add writing prompts 2. Simplify web app interface 3. Integrate AI assistant.
Utilize current vendor chatbot: Consider modifying our current, customer service chatbot to quickly standup a conversational letter-writing experience for limited release and testing.
Build a cross-platform app: Create a highly focused app that’s designed for future expansion, potentially replacing our existing platforms and becoming a one-stop solution for sponsors.
Compassion’s design/dev team is small and in demand, so I pitched a range of options to move forward with depending on available resources and the level of corporate buy-in:
Further testing: Refine the prototype through additional user testing and rapid prototyping to ensure we nailed down the correct MVP solution before digging into development.
Iterative integration: Gradually add conversational features and prompts to our current letter-writing process to ease users into the new system. Eg. 1. Add writing prompts 2. Simplify web app interface 3. Integrate AI assistant.
Utilize current vendor chatbot: Consider modifying our current, customer service chatbot to quickly standup a conversational letter-writing experience for limited release and testing.
Build a cross-platform app: Create a highly focused app that’s designed for future expansion, potentially replacing our existing platforms and becoming a one-stop solution for sponsors.
Compassion’s design/dev team is small and in demand, so I pitched a range of options to move forward with depending on available resources and the level of corporate buy-in:
Further testing: Refine the prototype through additional user testing and rapid prototyping to ensure we nailed down the correct MVP solution before digging into development.
Iterative integration: Gradually add conversational features and prompts to our current letter-writing process to ease users into the new system. Eg. 1. Add writing prompts 2. Simplify web app interface 3. Integrate AI assistant.
Utilize current vendor chatbot: Consider modifying our current, customer service chatbot to quickly standup a conversational letter-writing experience for limited release and testing.
Build a cross-platform app: Create a highly focused app that’s designed for future expansion, potentially replacing our existing platforms and becoming a one-stop solution for sponsors.
Compassion’s design/dev team is small and in demand, so I pitched a range of options to move forward with depending on available resources and the level of corporate buy-in:
Further testing: Refine the prototype through additional user testing and rapid prototyping to ensure we nailed down the correct MVP solution before digging into development.
Iterative integration: Gradually add conversational features and prompts to our current letter-writing process to ease users into the new system. Eg. 1. Add writing prompts 2. Simplify web app interface 3. Integrate AI assistant.
Utilize current vendor chatbot: Consider modifying our current, customer service chatbot to quickly standup a conversational letter-writing experience for limited release and testing.
Build a cross-platform app: Create a highly focused app that’s designed for future expansion, potentially replacing our existing platforms and becoming a one-stop solution for sponsors.
Compassion’s design/dev team is small and in demand, so I pitched a range of options to move forward with depending on available resources and the level of corporate buy-in:
Further testing: Refine the prototype through additional user testing and rapid prototyping to ensure we nailed down the correct MVP solution before digging into development.
Iterative integration: Gradually add conversational features and prompts to our current letter-writing process to ease users into the new system. Eg. 1. Add writing prompts 2. Simplify web app interface 3. Integrate AI assistant.
Utilize current vendor chatbot: Consider modifying our current, customer service chatbot to quickly standup a conversational letter-writing experience for limited release and testing.
Build a cross-platform app: Create a highly focused app that’s designed for future expansion, potentially replacing our existing platforms and becoming a one-stop solution for sponsors.
11%
More welcoming
11%
More welcoming
11%
More welcoming
11%
More welcoming
26%
More fun
26%
More fun
26%
More fun
26%
More fun
32%
More simple
32%
More simple
32%
More simple
32%
More simple
23%
Increase in overall experience ratings
23%
Increase in overall experience ratings
23%
Increase in overall experience ratings
23%
Increase in overall experience ratings
20m49s
Original avg. time-to-completion
20m49s
Original avg. time-to-completion
20m49s
Original avg. time-to-completion
20m49s
Original avg. time-to-completion
1m45s
New avg. time-to-completion
1m45s
New avg. time-to-completion
1m45s
New avg. time-to-completion
1m45s
New avg. time-to-completion
91%
Decrease in avg. time-to-completion
91%
Decrease in avg. time-to-completion
91%
Decrease in avg. time-to-completion
91%
Decrease in avg. time-to-completion