"Invisible design": we design together with cars
Irina Cherepanova and Tatyana Zhukova from the uKit AI project, which teaches neural networks to redesign sites, translated the column of Airbnb product manager Amber Cartwright from product manager on how smart technologies can improve known products.
The car was and remains my constant partner. With it, I transform creative thoughts into a tangible product that I can share with the world. When I was about 20 and I came into design, leaving a career in modern dance, I could not think that the car would be my assistant in creating breakthrough products.
')
Nowadays, machine intelligence is developing rapidly, and after it, the methods and products that we design must evolve. Here is a story about designing in tandem with machines or, as I call it, about “invisible design”: working with artificial intelligence and machine learning. Tools that, I believe, create fertile ground for the future of product design.
Take an example from the past: an English gentleman walks in the garden at the beginning of the 18th century and watches an apple fall from a tree. He wonders why it did not fall to the side or bounce up from the ground. How is this possible? What forces are involved? What is their nature? Does this phenomenon apply equally to small objects like apples and large ones like carts? Sir Isaac Newton had dealt with these issues for over 20 years and brought out the law of world wideness. He was able to describe the invisible force, which has a tangible effect on our daily life.
Recently, looking through the Facebook feed, I noticed that several of my friends liked the company, which offered online to design and order individual frames for posters and canvases. I thought about all those unframed works that lay in my closet, and clicked to see what was being offered. Why am I interested in this recommendation? What caught my attention? What kind of information did they use to personalize this post? The invisible forces of science and mathematics involved here are social network algorithms. Adware is only the embryonic state of the opportunities that the power of machine learning brings in the next 5–10 years.
I began to comprehend the essence of this “invisible design” in Airbnb, when we launched products that require processing a large amount of data. And I would like to share some observations from my practice.
Last year, we discussed a machine learning model for a new pricing tool that the owners wanted to offer with a product team. We tried to create a model that would answer the question: “What will the rental price be at some point in the future?” Finding the answer was not a trivial task. I tried to keep up while my data processing partner described the regression model they were building.
He showed sketches and threw himself with clever words, and although I could get an idea, the terminology was unusual for me as a designer. After that, I asked him to sketch out a diagram of the model and discuss all the details again. This conversation was a revelation to me. When he spoke in a familiar language — by illustrating the data in the diagrams and diagrams — I instantly understood the model and its goals. At that moment I had a “light bulb on fire”. I realized what value the machine represents for the product and how to use the information obtained in designing user interaction.
We both were inspired by the fact that we came to this understanding, and when the language barrier was broken, we were able to speak freely about the prospects for the development of the product. The discussion has already started at a new level.
Smart pricing regression model and its visualization: the model consists of three parts, which vary depending on the host supply and market demand.
I realized that the experience gained from this conversation can be extrapolated to our teams. This discussion was only a tiny part of a great story; part, comparable to the moment when we make the first sketches of screens in a notebook. I realized that the history of the product is not limited to the screens that users can see and touch. She also describes what happens behind the scenes.
At the initial stages of product creation, a support story is usually built - it describes the end-user experience so that each team member has an understanding of the appearance and behavior of the product. Different forms are used - from storyboards to prototypes, strategy presentations and diagrams. These presentations are created for many reasons, but the most important is the formation of a common vision of the product.
Understanding the product increases the potential of the team. General knowledge allows innovations not just to move forward, but to take leaps and bounds.
When we have an understanding of what we design and how it should work, we begin to create a product, with each using their own arsenal of tools. The carpenter has a hammer. The photographer has a camera. A product designer has a sketch. The developer has the code. Interestingly, only the last one, our partner in the workshop, has a tool capable of learning, changing and developing with time.
“Invisible Design” adds data and algorithmic solutions to the initial design stages — wireframes and behavior scenarios — bringing multidimensionality to the usually flat and static product creation stage.
Take for example pricing tips for homeowners on Airbnb. This was the first iteration of our Smart Prices product: pricing recommendations for the New Year holidays. From the data on the last season of winter holidays, we knew that people usually travel less during the last two weeks of December, and observed a sharp jump in trips closer to the New Year. And they wanted to inform our landlord community that if you lower prices at a certain point, you can attract more guests.
From the data model, we found that markets differ from each other depending on the low seasons. They needed different promises and visualizations: for example, in Sydney, the idle period begins in November, and there is no dead season in Miami due to constant demand.
Our scripts and wireframes, initially implemented in a single module, after processing, were able to show how market trends and statistics will affect the product.
Example: market dynamics causes the display of the necessary variations of modules and messages
When we make a product for an international audience (which means a difference in needs), at the same time we are in constant search for typical solutions that will help us achieve simplicity and understanding of the product, despite the difficulties. Visualization of scenarios and demand allowed us to see not just a handful of modules, but the entire communication system.
I had the honor to work together with incredibly talented designers, the best masters of their work, but not only they are creative. The approach of data processing specialists to the matter itself is an art form. Creating models and hypotheses about human behavior based on sets of information opened my eyes to one important thing: human behavior is nontrivial, and it is impossible to design the experience of interacting with the product in isolation, within just one design team.
For many years, I worked in an agency where teams existed separately from each other: designers in one department, developers in another. When I came to Airbnb, things were the same. 100+ developers accounted for only ten designers. Colossal imbalance. But the staff expanded, and the vice president formed a steering group that consisted of a design manager (I), a product manager, a head of statistics, and technical and financial managers. My view of the world began to change. I participated in discussions to which I had not been involved before, and we made joint decisions, taking into account the tasks and needs of the teams that we represented. I learned how other “worlds” are arranged, and how it is possible to profitably use the knowledge and skills of my colleagues to improve the product.
It is important to adhere to this rule in order for the “invisible design” to work. In some companies, programmers are in charge of fate, in others, the king is the product manager, and in others, designers command the parade. But very rarely, leading departments work in conjunction, jointly determining strategy and respecting decisions that are within the competence of colleagues. Sometimes it is useful to “encounter foreheads” - it’s a damn good product.
The Smart Prices function, when an algorithmic model advised apartment owners on which rental price is best set for a particular day, is an example of how the team helped develop the product.
Initially, the product could automatically adjust the rates, and we thought that this would be an excellent solution for apartment owners, because you do not have to adjust prices every day. However, examining the results, we learned that some landlords want to set the rental ceiling regardless of market demand: it would seem that you can earn more money in the market, but one segment had its own idea of ​​how much reasonable it is to rent your home to guests.
When the model worked for some time, we had to adjust it, taking into account the qualitative information we collected - the views of the owners of the property - and enter them into the interface.
At the exit, we received a product that was very successful in terms of meeting user and information needs: landlords received more control mechanisms - they were able to set minimum and maximum prices, as well as choose the desired frequency of guests. Below you can follow the evolution of the first version of the product Pricing Tips to our current Smart Pricing feature.
"Price tips" were a simple interpretation of the model - you adjust the price and see how likely your apartment is rented. If the landlord activated the price hints for a month, it was impossible to flexibly adjust the upper threshold for the amounts, except to manually set your price for each particular day. Our latest version of Smart Prices allows you to set tips for up to four months and has more fine-tuning of the minimum and maximum rates, thereby providing controls and functionality that homeowners find useful.
Many landlords "Smart Prices" liked, and this is the result of the fact that representatives of different departments were included in the work, arranged the necessary discussions, when circumstances required.
These thoughts are only a prologue to conversations about “invisible design.” I will continue to explore and write about my discoveries - about understanding problems and ways to solve them, about using tools and analyzing results - as I move further along the path of product design. To familiarize yourself with how an “invisible design” develops in practice, see a detailed analysis of the user's Smart Functionality scenario, which I presented at the IxDA conference in Helsinki earlier this year.
1. Free UI / UX and design programs:
2. Program “Big Data: Basics of working with large data arrays”
For whom: engineers, programmers, analysts, marketers, designers - everyone who is just beginning to delve into the technology of Big Data.
Training format: online
Details on the link →
The car was and remains my constant partner. With it, I transform creative thoughts into a tangible product that I can share with the world. When I was about 20 and I came into design, leaving a career in modern dance, I could not think that the car would be my assistant in creating breakthrough products.
')
Nowadays, machine intelligence is developing rapidly, and after it, the methods and products that we design must evolve. Here is a story about designing in tandem with machines or, as I call it, about “invisible design”: working with artificial intelligence and machine learning. Tools that, I believe, create fertile ground for the future of product design.
Mathematics and science are invisible forces that increasingly reveal themselves and influence our lives.
Take an example from the past: an English gentleman walks in the garden at the beginning of the 18th century and watches an apple fall from a tree. He wonders why it did not fall to the side or bounce up from the ground. How is this possible? What forces are involved? What is their nature? Does this phenomenon apply equally to small objects like apples and large ones like carts? Sir Isaac Newton had dealt with these issues for over 20 years and brought out the law of world wideness. He was able to describe the invisible force, which has a tangible effect on our daily life.
Recently, looking through the Facebook feed, I noticed that several of my friends liked the company, which offered online to design and order individual frames for posters and canvases. I thought about all those unframed works that lay in my closet, and clicked to see what was being offered. Why am I interested in this recommendation? What caught my attention? What kind of information did they use to personalize this post? The invisible forces of science and mathematics involved here are social network algorithms. Adware is only the embryonic state of the opportunities that the power of machine learning brings in the next 5–10 years.
Machines will increasingly make decisions on the experience of user interaction, and designing in tandem with them is the key to the future of product design.
I began to comprehend the essence of this “invisible design” in Airbnb, when we launched products that require processing a large amount of data. And I would like to share some observations from my practice.
Last year, we discussed a machine learning model for a new pricing tool that the owners wanted to offer with a product team. We tried to create a model that would answer the question: “What will the rental price be at some point in the future?” Finding the answer was not a trivial task. I tried to keep up while my data processing partner described the regression model they were building.
He showed sketches and threw himself with clever words, and although I could get an idea, the terminology was unusual for me as a designer. After that, I asked him to sketch out a diagram of the model and discuss all the details again. This conversation was a revelation to me. When he spoke in a familiar language — by illustrating the data in the diagrams and diagrams — I instantly understood the model and its goals. At that moment I had a “light bulb on fire”. I realized what value the machine represents for the product and how to use the information obtained in designing user interaction.
We both were inspired by the fact that we came to this understanding, and when the language barrier was broken, we were able to speak freely about the prospects for the development of the product. The discussion has already started at a new level.
Smart pricing regression model and its visualization: the model consists of three parts, which vary depending on the host supply and market demand.
I realized that the experience gained from this conversation can be extrapolated to our teams. This discussion was only a tiny part of a great story; part, comparable to the moment when we make the first sketches of screens in a notebook. I realized that the history of the product is not limited to the screens that users can see and touch. She also describes what happens behind the scenes.
At the initial stages of product creation, a support story is usually built - it describes the end-user experience so that each team member has an understanding of the appearance and behavior of the product. Different forms are used - from storyboards to prototypes, strategy presentations and diagrams. These presentations are created for many reasons, but the most important is the formation of a common vision of the product.
Understanding the product increases the potential of the team. General knowledge allows innovations not just to move forward, but to take leaps and bounds.
Visualizing the roles that machines and statistics play in learning is the first part of “invisible design.”
When we have an understanding of what we design and how it should work, we begin to create a product, with each using their own arsenal of tools. The carpenter has a hammer. The photographer has a camera. A product designer has a sketch. The developer has the code. Interestingly, only the last one, our partner in the workshop, has a tool capable of learning, changing and developing with time.
“Invisible Design” adds data and algorithmic solutions to the initial design stages — wireframes and behavior scenarios — bringing multidimensionality to the usually flat and static product creation stage.
Take for example pricing tips for homeowners on Airbnb. This was the first iteration of our Smart Prices product: pricing recommendations for the New Year holidays. From the data on the last season of winter holidays, we knew that people usually travel less during the last two weeks of December, and observed a sharp jump in trips closer to the New Year. And they wanted to inform our landlord community that if you lower prices at a certain point, you can attract more guests.
From the data model, we found that markets differ from each other depending on the low seasons. They needed different promises and visualizations: for example, in Sydney, the idle period begins in November, and there is no dead season in Miami due to constant demand.
Our scripts and wireframes, initially implemented in a single module, after processing, were able to show how market trends and statistics will affect the product.
Example: market dynamics causes the display of the necessary variations of modules and messages
When we make a product for an international audience (which means a difference in needs), at the same time we are in constant search for typical solutions that will help us achieve simplicity and understanding of the product, despite the difficulties. Visualization of scenarios and demand allowed us to see not just a handful of modules, but the entire communication system.
True mastery comes with the experience and development of an individual style, and this quality is inherent exclusively to people. Not a single machine has yet learned to express individual creative intent and artistic views.
I had the honor to work together with incredibly talented designers, the best masters of their work, but not only they are creative. The approach of data processing specialists to the matter itself is an art form. Creating models and hypotheses about human behavior based on sets of information opened my eyes to one important thing: human behavior is nontrivial, and it is impossible to design the experience of interacting with the product in isolation, within just one design team.
For many years, I worked in an agency where teams existed separately from each other: designers in one department, developers in another. When I came to Airbnb, things were the same. 100+ developers accounted for only ten designers. Colossal imbalance. But the staff expanded, and the vice president formed a steering group that consisted of a design manager (I), a product manager, a head of statistics, and technical and financial managers. My view of the world began to change. I participated in discussions to which I had not been involved before, and we made joint decisions, taking into account the tasks and needs of the teams that we represented. I learned how other “worlds” are arranged, and how it is possible to profitably use the knowledge and skills of my colleagues to improve the product.
The product team must include specialists from each department in order to make key decisions together.
It is important to adhere to this rule in order for the “invisible design” to work. In some companies, programmers are in charge of fate, in others, the king is the product manager, and in others, designers command the parade. But very rarely, leading departments work in conjunction, jointly determining strategy and respecting decisions that are within the competence of colleagues. Sometimes it is useful to “encounter foreheads” - it’s a damn good product.
The Smart Prices function, when an algorithmic model advised apartment owners on which rental price is best set for a particular day, is an example of how the team helped develop the product.
Initially, the product could automatically adjust the rates, and we thought that this would be an excellent solution for apartment owners, because you do not have to adjust prices every day. However, examining the results, we learned that some landlords want to set the rental ceiling regardless of market demand: it would seem that you can earn more money in the market, but one segment had its own idea of ​​how much reasonable it is to rent your home to guests.
When the model worked for some time, we had to adjust it, taking into account the qualitative information we collected - the views of the owners of the property - and enter them into the interface.
Having gathered together - behavior researchers, designers, product managers, developers, data analysis specialists - we were able to quickly change the course of the product strategy.
At the exit, we received a product that was very successful in terms of meeting user and information needs: landlords received more control mechanisms - they were able to set minimum and maximum prices, as well as choose the desired frequency of guests. Below you can follow the evolution of the first version of the product Pricing Tips to our current Smart Pricing feature.
"Price tips" were a simple interpretation of the model - you adjust the price and see how likely your apartment is rented. If the landlord activated the price hints for a month, it was impossible to flexibly adjust the upper threshold for the amounts, except to manually set your price for each particular day. Our latest version of Smart Prices allows you to set tips for up to four months and has more fine-tuning of the minimum and maximum rates, thereby providing controls and functionality that homeowners find useful.
Many landlords "Smart Prices" liked, and this is the result of the fact that representatives of different departments were included in the work, arranged the necessary discussions, when circumstances required.
These thoughts are only a prologue to conversations about “invisible design.” I will continue to explore and write about my discoveries - about understanding problems and ways to solve them, about using tools and analyzing results - as I move further along the path of product design. To familiarize yourself with how an “invisible design” develops in practice, see a detailed analysis of the user's Smart Functionality scenario, which I presented at the IxDA conference in Helsinki earlier this year.
About Netology courses
1. Free UI / UX and design programs:
- "The Basics of Graphic Design: Composition, Color, Typography"
- “Adobe XD: Basics for an Interface Designer”
2. Program “Big Data: Basics of working with large data arrays”
For whom: engineers, programmers, analysts, marketers, designers - everyone who is just beginning to delve into the technology of Big Data.
Training format: online
Details on the link →
Source: https://habr.com/ru/post/338478/
All Articles