“Designing Products Scaling Human Experiences”: Samihah Azim with Lyft (Video + Transcript)

October 22, 2018

At Girl Geek X “Elevate 2018” conference, Samihah Azim (Product Designer, Lyft) discusses designing products at One Medical and Lyft by considering all sides of the user journey, and balancing their needs with business/product metrics. Startups aim to transform the world, and designers create artifacts visualizing that North Star vision with the users at top of mind. Monitoring the health of the business with the user in mind helps achieve successful, human-centered products along the journey to the North Star.


Samihah Azim / Product Designer / Lyft
Sukrutha Bhadouria / CTO & Co-Founder / Girl Geek X


Sukrutha Bhadouria: [No audio 0:00-0:16] Hi, welcome everyone. I hope you’re having a good time with our conference so far. Welcome to Samihah, who is going to be our next speaker. She’s a Product Designer at Lyft, where she’s also the intersection of business calls and designing experiences impacting local communities. Prior to Lyft, she actually designed for a local commerce at Postmates and crafting high quality patient healthcare experiences at One Medical. Outside of design, she does a lot of cool things. She mentors for the State Department TechWomen program, enjoys power lifting, and also loves to cook. So, without further ado, I do want to say that these talks are going to be recorded and you will have access to the videos later after the conference, so go ahead, Samihah, thank you.

Samihah Azim: Thank you. Hello everyone, my name is Samihah as Sukrutha has said and I’m a product designer at Lyft. Thank you all for being here and Girl Geek X for having me and also, speaking of Lyft, shameless plug, we are hiring across the board. It is my favorite job. Prior to Lyft I was designing at Postmates, prior to that One Medical and today I’m going to talk about design as a powerful tool to scale products that have a core human to human experience. And, it can often be slower to scale these experiences and to get to that North Star vision for the products, but technology can be used to scale strategically and augment the human to human experience that’s happening outside of the software.

Samihah Azim: So, how does design add value when a product is scaling? Well, most products have a longer term North Star vision and if not the product, then certainly nearly every organization has a long-term North Star vision of what they imagine the future to be. But, it’s nearly impossible to get to that future today. You can’t go from zero to 100 unless you’re Drake and you’re on the catch up, but most of us are not Drake.

Samihah Azim: Designers, we’re very good at creating artifacts of what we imagine the future could look like. Where we can add value here is to really bridge that gap between today’s world here and our ideal world tomorrow. We can phase out what we have and what we need today, where we also add a tremendous amount of value is in using qualitative research that helps inform what our users need in each phase in order to come along the journey to our North Star vision. If you imagine that this the world here today and the North Star is up here. What is it that our users need in order to come up to this journey with us?

Samihah Azim: With design thinking, we’re really distilling a problem to understand how to solve it. Data helps to tell us and inform us in a lot of our decision making by telling us what’s happening. It doesn’t necessarily tell us why something is happening and that’s really where design adds a tremendous amount of value. We can help guide what we should test, if what we’re testing is the right thing to test and highlight if there are confounding factors that are potentially affecting the results.

Samihah Azim: When I worked at One Medical, we knew that a longer term goal was to make high quality affordable healthcare accessible to more people, but healthcare is a business, where that human to human or that human interaction component is still very important and technology isn’t something that will likely replace it but rather it would augment the experience. Brick and mortar is core to One Medical and technology augments the healthcare experience, so how do you scale a business where that human to human interaction is so core to everything but it also requires more operational resources that are often harder to scale. In short, it’s not easy.

Samihah Azim: When your product has a service that’s core, where humans are interacting with other humans, these are experiences that are happening outside of software but that will be associated with the product experience. This is where designers, we need to look at everything on a systems level and when I say systems level, I mean on the entire ecosystem and especially the business model and revenue streams. How does a company make money to further scale? I’m going to tell you a story from when I was working at One Medical and how I used design to learn how to scale human experiences.

Samihah Azim: In healthcare, medical practices make money through insurance billing codes and appointments are seen as inventories. Inventory is limited because doctors are limited, so how do you scale? Well, this is also where it’s important to look at what users need. Users that need to see a medical provider are booking appointments but not everyone that’s booking an appointment needs to be seen in person. There are acute issues. Issues such as flu or yeast infection, cough, nausea even getting an STD panel ordered that don’t require a physical visit to the doctor.

Samihah Azim: At One Medical, we knew that we wanted more patients to get the help that they needed by using virtual care products that we had built. I ran a design sprint with cross functional colleagues to understand the problem of why more people weren’t using care channels outside of the office visit. What we found is that people’s mental models today is that virtual care is something that, and specifically video visits, is something that is associated with travel or 2:00 a.m. emergencies. If it’s a 2:00 a.m. emergency, you should probably call 911.

Samihah Azim: One of the three projects that came out of the sprint that I had facilitated is a project that we called Integrated Booking Flow. Essentially, we wanted to test if educating users on virtual care and giving them what they needed today would get healthier patients to use virtual care in order for that to free up inventory for more sick patients that actually need it, and would that be something that the business could then scale and make high quality affordable healthcare accessible tomorrow.

Samihah Azim: The product manager on this project and I, we had really tight feedback loops, where we would meet regularly, multiple times a day and frequently we would also pair on both product management, as well as design. There’s a clinic next to the headquarters, so what we did is we would hop in there multiple times a day over a two-day period and test a bunch of paper prototypes as well as InVision prototypes. And, you can see on my screen, on the slides at the top, are the many prototypes … a sampling of the many sampling of the many prototypes that we had tested, the Guerrilla Usability Testing and what this helped us learn was what would work and what wouldn’t.

Samihah Azim: It was invaluable in not only helping us to learn that but also managing the many opinions of stakeholders and people who weren’t necessarily stakeholders but were involved in the projects. Essentially, this project was a test to see if integrating traditional equipment booking with video visits was further investing in and then, you can see on the screen on the bottom. That’s a sample of all of the feedback that we managed to capture and the action or inaction that we may or may not have taken, and the reasoning to that based on the usability testing that we had done.

Samihah Azim: When we rolled out the test, what ended up happening was that … Well, it was too successful. How often can you say that, right? We ended up having to turn it off because the virtual care providers were getting far too many requests and the SLA that we were communicating ended up being incorrect. We did validate that when mental models were gently guided towards this new shift in thinking, when users learned that video visits were for more than just travel and more than just emergencies, they adopted it for minor issues.

Samihah Azim: We didn’t spend weeks trying to perfect that most perfect V1, we shipped something that was good enough in order to learn and, in fact ,the screen on the slides, you’ll see the flow of integrated booking. Where a user goes to book an appointment and that middle screen was actually something that I had come up with by pairing with the data team to understand. We knew that a lot of the data was unstructured and so with their help, we were able to pull the top seven reasons why users are coming in for an appointment visit, which actually don’t require a physical visit but can be treated virtually.

Samihah Azim: Then, in that second to last screen is the alternate booking screen that users would see, patients would see if they chose one of those reasons, and we were gently guiding them towards and educating them that they could get care virtually for those issues and it would be much faster. If there’s one takeaway from this learning, it’s that it’s okay to move fast and ship an imperfect V1 in order to test and learn, so that you can iterate and ship that perfect V2.

Samihah Azim: This doesn’t mean shipping shoddy visual design. You can absolutely have pixel perfection without the V1 being in that ideal state. On the note of imperfect V1: sometimes when a product is scaling and especially with products that require that human to interact with another human, some team members might get a little too scope happy, scope cutting happy. I am that designer that cares a lot about things like client side load time and I’m also cutting scope or finding another way to solve a problem but there are times where you do have to introduce scope in order to have a viable product to test.

Samihah Azim: There was one company that I worked at, where we were working on an experiment and if successful, we would have further developed it, turned it into a core part of the feature in our products. But the product manager on the project was on this huge scope cutting spree. That by the end of it, it was barely a functioning test that made many of us question why anyone would use our product when competitors had far more basic functionality? And, we likely would have gotten a bad signal had we built it and tested it, where the … It came to a point where the term minimal viable product no longer applied. It was, the viable part got lost.

Samihah Azim: It’s important to ask questions because running experiments requires time and resources depending on your user set or what you’re testing, you have to wait until you get statistical significance, which can take a couple of weeks to a couple of months. You want to make sure you’re testing the right thing because you’re also using up engineering time, as well as design time and often times marketing time as well and you want to get a good signal.

Samihah Azim: That doesn’t necessarily mean it has to be a positive signal or a positive performance rather since you can learn a lot from negative performance but rather, what you want to learn are what is the metrics, what are they telling you? Is it something that you can learn from and does it fit into the overall product on a systems level?

Samihah Azim: If someone wants to test a popup flow at the point of user conversion, maybe there’s a good reason, ask why? But, it could also be that users are tapping on it because it’s a popup and it’s there and not necessarily because they find any value from it, so some questions … On that note, ask some questions. Some questions that I like to keep in my back pocket and ask are, “Why are we testing this, after this experiment what’s next, how do we know that the metric move this way because of this variant or XYZ confounding factor, did we even reach stat-sig, bro?” Well, maybe ask it a little less broy than that.

Samihah Azim: Why, and this is my favorite, why does it look like we’re p-hacking the data? Luckily at Lyft there is no p-hacking of any data. In a lot of organizations moving fast, growing, and scaling has the perception of being incompatible with staying true to your values. Designers, we want to feel like our work has a positive impact on the world. Sometimes when we’re so close to the data, it’s hard to have that perspective that work that grows an organization or a company that aims to do good with good intentions, is having an impact on the world.

Samihah Azim: Two of my favorite Lyft core values are uplift others and make it happen. At Lyft there’s actually an entire team that’s dedicated to growing the business, I am on that team. It is called the Growth Team, some creative naming there, but when we think about growth, we’re really looking to grow with intentionality so that we can continue to make a positive impact on the world. In fact, at the core of Lyft is tipping. It’s been a core part of the business and part of growing the business means initiatives that uplift our drivers.

Samihah Azim: Actually, recently this week, we actually got half a billion dollars in tips. Oops my apologies, I have the wrong data on there, it should be 2018. We actually raised … So, there were 250 million in tips in 2017 and just this week we hit half a billion. It took five years to get to quarter of a billion and then a matter of months to half a billion, and that really goes to show how growing a business can also help do good in the world as well. So, drivers are an important part of Lyft, so designing experiences that make a Lyft driver’s life easier, helps them earn more and that’s a design challenge where we’re doing good by one set of users and by doing good we can further scale, which helps us do even more good.

Samihah Azim: We also think about important causes and when we think about growing the business, it is also in the context of, how can we better benefit the society and without scaling, it’s hard to grow social giving. Last year in 2017, the team introduced a feature called Round Up & Donate, where Lyft passengers could round up their ride to the nearest whole dollar and that change would then go to a cause of their choosing, so 3.7 million dollars were donated to 14 causes. A lot of these causes were standing up for civil liberties, supporting service members, and investing in teaching members from underprivileged communities to code.

Samihah Azim: Now, I want to end on a couple of key points on what can designers do to be the most valuable team member possible? Well one, we can show value. The best way for design to show our value is to really to start caring about business goals and team metrics and then, being comfortable with an imperfect V1, so that we can test and learn to build that perfect V2 and move faster to that ideal North Star. Then also being really cognizant of what’s being tested when designing and what effect it has of other metrics, as well as what effect it has on other users or on users. I am not sure if we have time for Q&A but if you have any question, feel free to reach out to me on Twitter.

Sukrutha Bhadouria: Yeah, thank you so much, Samihah, this is great. We have time for one question and so, the one that’s gotten the most votes, “Could you share with us about a technical challenge that you are currently working on and what are the technologies, tools, and concepts that you are using?”

Samihah Azim: Unfortunately, I can’t talk about any products that have not been released yet, but I guess I could talk about it on a very high level, so there’s a project that I’m on, where I designed this, what we would consider the ideal V1 and there are some technical limitations on the engineering side, where we could get to that point but it would take about six weeks.

Samihah Azim: When I heard that I was like, “That is a really long time, how do we get there faster?” Because I think it’s also important to learn as quickly as you can. If we spend six weeks building that ideal V1 maybe it doesn’t perform that well, maybe there’s a lot of assumptions that we’ve made. What I then did is paired really closely with engineering, as well as the product manager on the project and we broke the different pieces down, and really understood what the scope of each part would be and how we can get to shipping the test faster.

Samihah Azim: We went from six weeks to three weeks, which is great because that means we can then … Once we ship that we can … once we hit statistical significance, we now have enough data to learn from that to then make adjustments to the idealized version. The benefit of that, the current, what we’re calling V1 now, which would take three weeks as well as having that idealized V1, which would now be V2, is that the backend engineers now understand what direction we want to go in, so that way they’re able to build in a way that’s scalable to the future where the client side engineers they … to them what’s more relevant is the immediate V1 that would take three weeks to build.

Sukrutha Bhadouria: Thank you so much, Samihah, we’re actually out of time. Thank you everyone.

Samihah Azim: Thank you and thank you for having me.

Sukrutha Bhadouria: Thank you.

Share this