<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Jay Cox-Chapman on Medium]]></title>
        <description><![CDATA[Stories by Jay Cox-Chapman on Medium]]></description>
        <link>https://medium.com/@coxchapman?source=rss-232e42b312de------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*xD81TkzL3nXty4M6c-TMfw.png</url>
            <title>Stories by Jay Cox-Chapman on Medium</title>
            <link>https://medium.com/@coxchapman?source=rss-232e42b312de------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Fri, 20 Nov 2026 04:47:51 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@coxchapman/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[A ride for everyone: how Zoba helps cities meet transportation equity goals]]></title>
            <link>https://medium.com/zoba-blog/a-ride-for-everyone-how-zoba-helps-cities-meet-transportation-equity-goals-e27675d5ca0f?source=rss-232e42b312de------2</link>
            <guid isPermaLink="false">https://medium.com/p/e27675d5ca0f</guid>
            <category><![CDATA[micromobility]]></category>
            <category><![CDATA[cities]]></category>
            <category><![CDATA[optimization]]></category>
            <category><![CDATA[equity]]></category>
            <category><![CDATA[transportation]]></category>
            <dc:creator><![CDATA[Jay Cox-Chapman]]></dc:creator>
            <pubDate>Thu, 16 Sep 2021 15:09:24 GMT</pubDate>
            <atom:updated>2021-09-16T15:09:24.798Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/924/0*dz8IC4Gs5cqZNHs5" /></figure><p>Fixed transportation networks are hard to build. Highways and mass transit — even dedicated bus lanes — all take significant public investment and political will. As a result, transportation infrastructure in many cities is profoundly unequal, with richer areas of the city enjoying greater access to transportation.</p><p>Shared mobility fleets dramatically lower the cost of changing old and unequal patterns since they can be deployed with little cost to the city and without fixed infrastructure. Once in place, mobility fleets help planners uncover demand for new routes and infrastructure as users reveal previously hidden demand patterns.</p><p>By deploying vehicles to a historically underserved neighborhood, residents get instant access to modern, convenient transportation, no fixed infrastructure required. Forward-thinking cities therefore require operators to manage special zones in their markets as a part of their operations and as a condition of operating permits.</p><p>Zoba’s decision automation product, Zoba Move, provides recommendations to operators about where to deploy, rebalance, or swap vehicles to capture the greatest number of rides. Operators can configure their zone requirements in Zoba Move, and we will automatically make recommendations that both capture ridership and meet regulatory requirements.</p><p>Zones can be <strong>minimum </strong>vehicle levels, as with equity or opportunity zones, or <strong>maximum</strong> vehicle levels, as with central business districts. For example, a city may require that a neighborhood contain at least 10% of the deployed fleet, or stipulate that the central business district contains no more than 1,000 vehicles. The goal is to stimulate demand in underserved neighborhoods and to ease sidewalk clutter in dense downtowns.</p><p>Zoba manages zones by optimizing for tasks that satisfy the constraints in task lists while still maximizing the amount of rides and revenue a service captures. Market managers no longer need to manually add extra tasks for zones on top of their deployment, pickup, and battery swap tasks. During optimization, Zoba will review the position of a live fleet and assess how many vehicles are inside or outside of the zones. Zoba will then recommend whether to move vehicles in or out of a zone in a way that preserves as much ridership as possible.</p><p>Below are three simple scenarios for a fleet with a <strong>minimum zone requirement of 25%</strong>. In the first, the constraint is met. There are 3 vehicles inside the zone and a fleet size of 12.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*zP5mE2itaAJQVoea" /></figure><p>Here, the fleet has more vehicles in the zone than strictly required. The constraint is met, but Zoba will recommend rebalancing one vehicle away from the zone if demand is low and the vehicle could be more productive elsewhere.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*j9aFj-6rSgHDmhFY" /></figure><p>In our final scenario, the fleet is not meeting the minimum zone constraint. There are only 11 vehicles deployed, and only two inside the zone. Zoba will recommend deploying an additional vehicle into the zone to meet the constraint.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*E8-UNrKojHLGRTLP" /></figure><p>Managing shared mobility fleets is a complex task: market staff need to juggle business goals, fleet operations, and regulatory requirements. Zoba’s zone feature makes market managers’ lives easier while minimizing the impact on ridership. Zoba is proud to support our customers in meeting their regulatory commitments in a way that maximizes ridership.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=e27675d5ca0f" width="1" height="1" alt=""><hr><p><a href="https://medium.com/zoba-blog/a-ride-for-everyone-how-zoba-helps-cities-meet-transportation-equity-goals-e27675d5ca0f">A ride for everyone: how Zoba helps cities meet transportation equity goals</a> was originally published in <a href="https://medium.com/zoba-blog">Zoba Blog</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Announcing swap tasks in Zoba Move]]></title>
            <link>https://medium.com/zoba-blog/announcing-swap-tasks-in-zoba-move-b5b860687d6f?source=rss-232e42b312de------2</link>
            <guid isPermaLink="false">https://medium.com/p/b5b860687d6f</guid>
            <category><![CDATA[transportation]]></category>
            <category><![CDATA[battery]]></category>
            <category><![CDATA[micromobility]]></category>
            <category><![CDATA[scooters]]></category>
            <category><![CDATA[electric-vehicles]]></category>
            <dc:creator><![CDATA[Jay Cox-Chapman]]></dc:creator>
            <pubDate>Thu, 29 Apr 2021 18:49:18 GMT</pubDate>
            <atom:updated>2021-04-29T18:49:18.566Z</atom:updated>
            <content:encoded><![CDATA[<p>Today, Zoba is announcing support for battery swap tasks in the output of Zoba Move, our supply repositioning product. The addition of swaps increases labor productivity up to 20% in terms of tasks per hour, on top of the 20–50% gains in ridership that customers typically see from the deployment and rebalancing features of Move.</p><p>Now, market managers simply enter a desired volume of deployment, rebalance, and swap tasks and Move chooses the best vehicles, locations, and tasks for the market. A shift that previously required hundreds of individual decisions is simplified into a single task list, batched and sequenced for easy execution.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/636/0*qV4IjlC7DsckuvX_" /><figcaption>Without Zoba, fleet tasks can be overwhelming</figcaption></figure><p>Where possible, Zoba will combine rebalance and swap tasks together for higher operational efficiency. The gains can be significant, <a href="https://medium.com/zoba-blog/optimal-operations-how-decision-automation-can-streamline-micromobility-operations-aa217e989d5b?source=friends_link&amp;sk=6b9f14fd574e42088dd511a595e20375">as my colleague Nick detailed in a previous post</a>.</p><p>Using swaps is easy. Before each vehicle leaves the warehouse, the market manager inputs the number of swaps to complete, along with van capacities. In the example below, we see this market planning to rebalance 10 vehicles and swap 35 batteries. They’re doing it with two vans, each of which can carry 30 vehicles and 33 batteries.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/583/0*Bocl6m56yTexvS_Q" /><figcaption>Market managers enter task volumes and vehicle capacities in the Zoba portal</figcaption></figure><p>The output then includes pickups (for rebalancing), swaps that occur in-place, and a combination swap and pickup, batched by van and sequenced in an optimal order. The output below can be integrated into your operational tools (via CSV or JSON) or used directly as a task list.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/712/0*BVobw0QkJUWcZzbP" /><figcaption>Sample task list</figcaption></figure><p>Swap tasks can be executed as standalone tasks (a “battery_swap” in the output) or can be combined with rebalancing tasks (pickup_swap_battery). In the example below, assume each circle contains 50 vehicles. The day’s tasks will be 50 rebalances and 50 swaps, for a total of 100 tasks. However, 25 of the rebalances coincide with swaps on the same vehicle, so the market only performs 75 vehicle visits. Those 25 tasks represent productivity gains for you and your team because they don’t have to revisit those vehicles separately.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/743/0*bggmpqRTzbcUKmPX" /></figure><p>Behind the scenes, Zoba Move is making decisions about battery level and demand. If a vehicle is in a high-demand location but has a low battery, all it needs is a new battery to continue capturing rides. If it’s in a low-demand location and has a low battery, we want to move it to a better location with a fresh battery. As Zoba cofounder Dan Brennan pointed out last month, <a href="https://medium.com/zoba-blog/operations-planning-for-swappable-battery-micromobility-80b4c0138109">combining tasks is the optimal way to transition to swappable batteries.</a></p><p>We also batch tasks by van and sequence them in an optimal order to minimize driving time and respect van capacities, both in terms of vehicles and batteries:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*Avblv-lt4NKYd1yf" /><figcaption>Zoba’s batching and sequencing features give operations staff a clear path through their tasks.</figcaption></figure><p>Adding swap tasks to Zoba deployment and rebalancing lists makes task management more integrated and convenient for operational teams, and raises overall fleet performance. To learn more and start using swaps today, <a href="https://www.zoba.com/contact-sales">contact Zoba sales</a>.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=b5b860687d6f" width="1" height="1" alt=""><hr><p><a href="https://medium.com/zoba-blog/announcing-swap-tasks-in-zoba-move-b5b860687d6f">Announcing swap tasks in Zoba Move</a> was originally published in <a href="https://medium.com/zoba-blog">Zoba Blog</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Joseph Brennan and Oliver Bruce talk operations and the future of micromobility]]></title>
            <link>https://medium.com/zoba-blog/joseph-brennan-and-oliver-bruce-talk-operations-and-the-future-of-micromobility-9c71f7f12e08?source=rss-232e42b312de------2</link>
            <guid isPermaLink="false">https://medium.com/p/9c71f7f12e08</guid>
            <category><![CDATA[micromobility]]></category>
            <category><![CDATA[operations]]></category>
            <category><![CDATA[autonomous-vehicles]]></category>
            <category><![CDATA[pricing]]></category>
            <category><![CDATA[transportation]]></category>
            <dc:creator><![CDATA[Jay Cox-Chapman]]></dc:creator>
            <pubDate>Tue, 30 Mar 2021 20:02:31 GMT</pubDate>
            <atom:updated>2021-03-30T20:02:31.577Z</atom:updated>
            <content:encoded><![CDATA[<p><em>Last week, Zoba cofounder Joseph Brennan sat down with Oliver Bruce of Micromobility Industries to talk about Zoba’s approach to micromobility operations. Listen below as they discuss decision automation, the future of micromobility, and why operations may be operators’ best bet for improvements to unit economics. To learn more about Zoba and get in touch, visit us at </em><a href="http://www.zoba.com/"><em>zoba.com</em></a><em>.</em></p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2F79ipNkpCEN4&amp;display_name=YouTube&amp;url=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3D79ipNkpCEN4&amp;image=http%3A%2F%2Fi.ytimg.com%2Fvi%2F79ipNkpCEN4%2Fhqdefault.jpg&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;type=text%2Fhtml&amp;schema=youtube" width="854" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/c049c9a4bd059ff7eb88491c68a7da9a/href">https://medium.com/media/c049c9a4bd059ff7eb88491c68a7da9a/href</a></iframe><p><a href="https://podcasts.apple.com/us/podcast/magic-operational-optimization-more-rides-more-money/id1434457337?i=1000514803672">Apple podcasts</a>, <a href="https://open.spotify.com/episode/3c0LUzHfLX2njCbtgelYDm?si=rF7cVNUMSqi-NYKQvTUDHA">Spotify</a>, <a href="https://www.youtube.com/watch?v=79ipNkpCEN4">Youtube</a>.</p><p><em>The following is a transcript of Joseph and Oliver’s conversation, lightly edited for clarity.</em></p><p><strong>Oliver Bruce: </strong>I’m really excited to have you on. I’ve been following your work probably for about a year or so.</p><p>I’ve been really excited to learn a bit more, mainly because when I was at Uber, I picked up an understanding of why software-enabled operations and analytics were so powerful. And it’s been really exciting to see that getting applied to the micromobility space.</p><p>Maybe what we could do is start for the audience right back at the beginning. What’s your background? Where did you come from? And then we can work through how you started Zoba.</p><p><strong>Joseph Brennan: </strong>The Uber background on this topic is interesting. Uber in a lot of ways got everyone thinking about these problems. Oftentimes that’s the first thing people go to because they made these problems cool: these geospatial analytics issues.</p><p>Zoba is a micromobility optimization platform. We are optimizing operations. What that tends to look like is vehicle placement tasks, decision automation. So you’ve got swaps, pickups, drops, all these different things. We’re creating logic to automate some of those decisions. Additionally, we’re getting into the dynamic pricing game in a big way here as well, which is where we think a lot of the market is going eventually.</p><p>As far as my pathway into micromobility, it was a bit of stumbling into it. I started Zoba with my brother Daniel in 2016, we were both grad students at the time; he was in Boston, I was in Beijing. He had the original idea. Daniel was fresh out of the military, where he had been amazed by the geospatial tech they were using there. And so he had this idea that we could start a company that was trying to elevate the geospatial analytics space in a pretty horizontal way. This was pre-Kepler and a lot of the open source innovation that happened at Uber. And we felt in general, people were not taking advantage of all the data that they were building.</p><p>Our path to micromobility was working on a general spatial analytics platform, and we’re having conversations in a lot of spaces, and then just felt this immediate pull from many of the micromobility operators we talked to. This was super early. There were people operating municipal bike shares and stuff. This was 2017, so early on, even before Bird was really becoming super popular.</p><p>Around the same time we met up with a guy who now leads our data science team, Evan Fields, who was finishing a PhD at MIT focused entirely on novel demand-forecasting techniques for mobility systems with hard assets. So not two-sided marketplaces, but micromobility or car share, and so the combination of what we were already working on, in spatial analytics, and meeting some key people at the right time. It all just converged. And so we dug into the micromobility space, went pretty vertical there and have been doing that for a few years now.</p><p>And it’s been tons of fun. To grow up with the space, which I never expected to work on, urban tech. My brother and I grew up in a really small town in Texas. So it’s fun now playing this role in city life.</p><p><strong>Oliver Bruce: </strong>Totally. Talk us through the geospatial thing, for someone who doesn’t have any reference point. What are you doing? All of this stuff is pretty complex to me.</p><p><strong>Joseph Brennan: </strong>This is a really interesting question. So the way I would answer this now is pretty different than the way I would answer this previously.</p><p>What Zoba does now is automate a lot of operational interventions. But they’re all operational interventions that are time- and space-sensitive. We’re deciding “hey, you’ve got a thousand vehicles, this is the 10% you should be doing a price intervention”, or “you should place 50 more out in the market. Here’s where they should go.”</p><p>When you think geospatial, you think maps or map-making and there are certainly maps involved, but the real core thing is decision automation as it relates to very dynamic questions of space and time. So I’d say it’s really more about very concrete things: We placed the vehicle there and not there. We priced the vehicle differently on that corner versus that corner.</p><p><strong>Oliver Bruce: </strong>Wow. I’m really excited about the pricing thing manually. We can come back to that though. Is it just for placement?</p><p>A company — let’s give them a name that has four letters and starts maybe with some letter like a T or an O or A — when you go and start working with them, is it just placement of the vehicle in the beginning of the day? Are you starting to intervene at multiple stages throughout the day in terms of moving a scooter or an e-bike around on a shared system, or is it mainly just where it gets placed in the morning?</p><p><strong>Joseph Brennan: </strong>It depends if it’s a swappable, and we get into this later, but this is a bigger conversation. There has been a quiet ground shift in terms of how operations are happening in micromobility. If it is a non swappable fleet, that’s the first iteration.</p><p>The charge cycle on non-swappable fleets is very basic, right? You go get the 10% or 20% of the vehicles that are dead, maybe more. If you’re getting very high utilization, you take them back to your warehouse. You charge them, you place them out at 4:00 AM or 5:00 AM or whatever.</p><p>In a swappable context, this is dramatically more complex. There’s no natural signal to get all the vehicles at night and move them. These things are happening throughout the course of the day. And there are many more decision factors going on. So you are rebalancing vehicles, you are having to decide how many to rebalance.</p><p>Should you rebalance any? We don’t have to move them anymore because we’re not taking them back to the warehouse. They’re just on the street. We’re swapping batteries. We’re doing point maintenance. We might have to pick some up and these things are happening much more dynamically all day long. So I would say if it is a swappable fleet, this is an all-day thing.</p><p>It’s not just about whether this vehicle is poorly allocated and this one is well allocated. It’s the entire operations logic. So should I be swapping on route, or not?. Do I have regulatory issues I have to solve on route to my rebalances?</p><p><strong>Oliver Bruce: </strong>Totally. Okay. I know you’ve got some beautiful images and things like that. And I was just wondering, can you pull those up?</p><p><strong>Joseph Brennan: </strong>One thing I want to provide context for someone who is new to spatial analytics, so you know how to think about it. The first-generation of spatial analytics was mostly to inform. So you imagine ArcGIS or whatever, or Kepler, you’re pulling up data. You’re looking at it. You’re seeing “I get rides over there. I get rides over here.” Whatever you’re looking at, right.</p><p>The way Zoba approaches it is much more like your Uber example, where a backend decision-making process relates to spatial problems. When you open the Uber app and they serve your price and they serve your ETA, there is a tremendous amount of complex data processing and data analytics happening behind the scenes to serve you that. A lot of it relates to space and time, but it’s not just showing you a map. It’s getting to that final decision. Zoba is more on that end of the spectrum.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*6KJZrPb6i6Nnd7iYiQegnQ.png" /></figure><p>I’ll show you a couple of things quickly. So if you imagine a micro fleet that’s moving, there’s a lot of really interesting supply dynamics with micromobility that aren’t the same with an Uber. One of those is that the users create the supply network primarily.</p><p>If we have a thousand vehicle fleet, and we are rebalancing 10 to 20% of the fleet on any given day, the users are doing two rides per vehicle per day. The users are an order of magnitude more frequently creating the supply network. They’re choosing where the vehicles are left.</p><p>And what vehicles will be available to the next cohort of users? They’re really creating the system and Zoba is just operating within the system they create. And so just show you what this looks like this, this little circulation visual, these areas in blue are on net attractors of vehicles.</p><p>So vehicles flow in there more frequently than they leave. Areas in red are the exact opposite. They’re leaving more frequently than they show up. And so, as you see, as in most markets, this is pre-COVID. COVID data looks very different just because there’s less circulation, there are fewer commuters, but if we look pre-COVID, you see people flow into downtown Austin in the morning.</p><blockquote>Ultimately what we’re trying to generate are task lists that go out to an operator and then they implement it. And if everything’s working right, it should really change the downstream readiness of the fleet to meet demand where it is.</blockquote><p>And then here in the afternoon, they flow back out. And you have some nice circulation, but anywhere you look in the market, there are asymmetries. If we look here at Zilker Park, people are dropping vehicles on-net here, both in the morning and in the afternoon, over the course of the day supply starts to accumulate here.</p><p>There’s less supply available in the whole network. And as a result, the network as a system really suffers. The way the Zoba system works is that we’re trying to understand, not just the next set of user behaviors, but how the whole system will evolve over the course of a day.</p><p>So just to give you an example in Austin, again. So this visual is a demand evolution throughout a generic day in Austin. So where do we expect users?</p><p><strong>Oliver Bruce:</strong> So when it, when it’s going up, what does that mean? Does that mean that there’s demand there or that there’s new supply that’s being deposited?</p><p><strong>Joseph Brennan: </strong>Higher levels are higher demand. And when we say demand at Zoba, we actually mean demand. We mean where users want rides. Not trip starts, not app opens. This is a Zoba generated proxy for demand. We’re trying to measure it as best we can.</p><p>And it’s a lot of what makes our system special is that we’re quite good at that. So this is measured in rides wanted per hour. It’s beautiful. We’re looking at how demand is flowing throughout Austin throughout the course of the day, sensitive to weather and all the different things that might be going on in the market.</p><p>This is one of the core building blocks. The other core building block is O-D matrices. I mentioned the supply fact earlier — that users are dropping the vehicles where they want to end their trip, right? The users don’t care about how well-positioned your fleet is. It’s not material to them. What this looks like is for any given point, you want to understand what the probability distribution is of places that users will drop the vehicle.</p><p>So if we just look at the University of Texas here, with tens of thousands of students, this is for a weekday morning. The thickness of the line here is suggesting the relative probability that a vehicle will be taken to some destination, given a trip start. So if the trip starts where I have my cursor, where is it likely to go? It’s most likely to go somewhere else in the University of Texas, the students are taking it to another part of campus or infrequently, south towards downtown.</p><p>And then if those users take it, you know, let’s say just west here, then we can say, okay, so then if the vehicle ends up west, where’s it likely going to go next? And if you have these pieces together with the demand, you can basically model how the market will evolve over the whole day and then make whatever limited interventions you’re making in such a way that they have an asymmetrically high impact on the market. And another way to think about is chaining together rides or placing vehicles, such that they float downstream to demand. And then what this looks like in a very concrete way, it’s pretty simple, right?</p><p>Let’s look at it in a rebalancing context. For a swappable fleet, you’re saying pick up this vehicle ID and reallocate it to one of these points. I had mentioned earlier that we’ve got a bit more full stack in terms of task unification today. In practice, what this looks like is three different vanloads of about 50 vehicles each.</p><p>We’re saying, “look, you’re going to start with this van. And these are all the stops you’re going to go through.” In a non-swappable context that would just be dropoffs. In a swappable context, that could be “pick up a vehicle, drop a different vehicle, swap a battery here.” There’s a lot more on route that might be changing. But ultimately what we’re trying to generate are task lists that go out to an operator and then they implement it. And if everything’s working right, it should really change the downstream readiness of the fleet to meet demand where it is.</p><p><strong>Oliver Bruce: </strong>Totally. Oh, dude, this is fascinating. That level of routing — I mean, I remember talking to some of the early operators in the scooter game. The early days it was “we know where the scooter is. We think there’s demand here. We can pull some of the basic analytics of knowing that someone’s tried to open the scooter app to see if there was availability there.”</p><p>They’ve gone and — I can see the granularity of it. You’re getting to the point where you route it and drop the scooters, often these places I take it, all of that has been done. Those calculations are largely done. How do I put it? You’re getting to the point where you’re optimizing for revenue. Right? You’re looking for ways to drop the scooter to somewhere that the probability is high that someone’s going to go take a trip. Yeah. But headed somewhere where someone will take a trip after and after and after.</p><p><strong>Joseph Brennan: </strong>Totally.</p><p><strong>Oliver Bruce: </strong>The thing is at the moment, we haven’t really even talked about pricing and, and, and at the moment, I mean, I certainly don’t know of anybody who has done integrated planning. There’s no scooter app that I’ve opened where I’ve said, “I want to go from here to here and then I actually walk to the scooter and it’s said, this trip is going to cost you $3 for that trip or $1 for that trip,” because you’re actually taking it to a place that we want you to go to.</p><p>Ryan from Jump was interviewed by Jason Calacanis and he was talking at the time, but they had done incentive programs for people riding into high-demand areas.</p><p>And that was about the only thing that I’ve ever seen. We haven’t really seen any of the operators actually get fully integrated yet. And you know, where’s that going? I assume pressed price responsive demand is just going to be next.</p><p><strong>Joseph Brennan: </strong>Right. Yeah. Dynamic pricing is definitely on the horizon. We’re gonna see more and more of it. One last thing on the routing: it’s challenging because there’s enough different about the micromobility use case that frustrates a lot of current routing engines.</p><p>Like battery levels, there are certain heuristics that are relevant to the micromobility space where, you know, if you need to know what the battery level is for a certain threshold to pick up on route, that’s hard for most systems to manage outside of parcels and the like.</p><p>On the pricing piece, if I could show you something else. I’m pretty excited about this. I think this is going to be a big way the space moves. Right now, you’re saying there’s very little dynamic pricing work being done, even though it’s so commonplace in a lot of other adjacent spaces. In addition to that, where you do see dynamic pricing, you can’t see this as the user, but on the backend.</p><p>What’s typically happening is there’s either a time threshold. The vehicle’s been idle for two days at that point, it’s already a huge problem, but only then it gets a price intervention. The other thing that’s happening is static geo-fences, so if I drop in this specific area, it doesn’t change. I get an incentive. It’s not actually really dynamic in any meaningful way. It’s just altering the user experience a bit. We think that the main blocker is that it’s just a really hard problem in micro, because you need to be right, many hours in advance.</p><p>If you think about Uber, Uber has a price-responsive supply. If they need more drivers on the road, they get more drivers on the road, they’re doing a matching problem where, you know, they raise your price a little bit. They raised the drivers price a little bit or pay a little bit, and there’s a nice matching problem.</p><p>The supply moves. You can get the driver wherever you want to get the driver. With micro, the user is not telling you where they’re going. They’re not telling you how long the trip is going to be. You have to anticipate a lot about the user than other systems. In ridehailing, they would just signal with their, with their search or whatever.</p><p>The way we’ve approached this is, and the main thing they’re missing, is to accurately model the demand over many hours. You just, you can’t do anything, right? There’s so many missing factors from what the user’s telling you. We’re using a lot of the same tech that I showed you earlier.</p><p>So understanding how the demand is going to evolve, understanding what those chains of rides are going to be, to basically offer the user incentives for almost user rebalancing, right? We flag say 5–10% of the fleet for a price decrease on trip start.</p><p>We don’t have to make it a drop-off constraint. We don’t have to say you have to drop the vehicle here, which is fairly user-hostile because we know the probability distribution of where the users are likely to go, if I’m pretty certain you’re going to take the vehicle to downtown from where you are.</p><p>In the grand scheme of things, it’s not that beneficial for me to require you to go downtown. I just have to continue to model that and do my best. It’s estimating the probabilities.</p><p><strong>Oliver Bruce: </strong>Totally. I get that, when you say that there are just so many variables that you can’t know about a customer, it just really drives that home to me — the times I’ve been walking into town and passed a scooter.</p><p>And the scooter is in a terrible place and so you should be paying me to ride this into the demand area.</p><p><strong>Joseph Brennan: </strong>Especially with mopeds where it costs so much to move. With kick scooters, the rebalancing can be challenging, but I mean more and more operators getting into mopeds, they should definitely be paying you to move it just because it’s so costly to move them.</p><blockquote>We’re early on in this industry is what I’m trying to say. I think people don’t appreciate that. No one’s using the first generation of scooters anymore. Like scooters that lasted two, three weeks are gone. No one does that anymore, but a lot of operators are still operating the similar way than they were early. I think that the big inflection point is anyone who tries to transition to swappables with an old model gets nailed. It’s really hard.</blockquote><p><strong>Oliver Bruce: </strong>Totally. And yet it’s very hard to be able to surface that intent from me, and I wouldn’t necessarily ride it. I’ve decided to walk into town, so I’m going past and I’m not in another transport mode. I bet I would only make that decision if I didn’t need it very much. It’s this weird thing of how do you service that demand. And then that’s where the overall idea of a marketplace for mobility comes in and, and this is, you know, I just love to hear your thoughts on you know, the, the idea of an infrastructure and going back in history a little bit.</p><p>Like that’s what I worked on a little bit when I was at Uber. What’s this idea of what would happen if there was a strategic threat, if Lyft combined all of their supply with all the taxis and with every other player in the world, right?</p><p>Say there was one person who said, I want a ride and then everybody could bid for that ride. You know that there was an open marketplace for this, and I can see that that’s maybe going to come. You knew you couldn’t do it as it exists on existing infrastructure at the moment, but you might see it coming along with some of the open marketplaces that have been built on crypto and that stuff.</p><p>That there would be an open marketplace for supply and an open marketplace for demand. And so you would be able to see, for example, the rider that’s saying I’m going from here to here and it can be serviced by a variety of modes.</p><p>That’s how we actually started getting more open, flexible marketplaces, but that’s 5–10 years down the line. There’s a lot of stuff that needs to happen. What do you think would be the steps that you’d take or that will be taken to be able to get us to that world?</p><p><strong>Joseph Brennan: </strong>Where Zoba ends up so dialed in on the operations today with the signals we see. You’re going to see cities get smarter with how they’re wielding their power in terms of some of these rudimentary regulations where they just tax you intensely.</p><p>Over time, instead of a per vehicle basis, which you’re going to pay the city, it’s going to be “if you’re in these zones, these demand zones or whatever else, this is the price ascribed to these in terms of a more unified system.”</p><p>One big shift that people under appreciate is that as with micromobility, also with autonomous vehicles. Micromobility looks a lot more like AVs than you would think. And the reason why is that the supply is fixed. These Uber-like models, DoorDash or whatever, they’re two-sided marketplaces where the operator doesn’t pay for the trip unless the trip is matched.</p><p>And then they pay out to the driver. They take a cut and the price is an incentive for the supply. You can get more supply on the street. With micromobility, like AVs, it’s going to be interesting because you’ve got some fixed amount of supply in the near term for the city.</p><p>And then to your point, you’re trying to figure out across all these things: buses, AVs, transit, micromobility, whatever it is, how to best service this whole ecosystem of demand with an amount of supply that’s relatively locked in the near term. It’s a very challenging problem. And so some cross functional communication there could definitely work out.</p><p>I mean, I don’t, you know, you would know better than I what that might evolve to look like,</p><p><strong>Oliver Bruce: </strong>It’s like MDS and GBFS, but on steroids: pricing and the ability to do dynamic pricing and all that stuff. That’s conceptually, that’s what MDS is, right?</p><p>Like that’s why DOT came up with it and then subsequently put it into the OMF — sorry — OMF is the Open Mobility Foundation. Gee, I should stop with the acronyms. But yeah, the idea of mobility data specifications is because LA could see that there was a future coming where they were going to be autonomous vehicles on the road and that they wanted to be able to do real-time pricing for them.</p><p>And you need to know where the vehicle is and you need to know how to price it and you need to know how to do it, that stuff. Real-time pricing for our cars on our roads. And that actually, you know, that is the biggest thing. If governments were really serious about trying to solve congestion, they would actually do that.</p><p>They would pursue that as a mechanism rather than trying to build more infrastructure, because it just comes down to the fact that we have mispriced all of our streets and their space allocation. That’s a whole other discussion for another day. I do want to dig in a little bit in terms of the stuff that you see that’s coming down the pipe from operators.</p><p>So you talked about swappable batteries earlier. It’s not a silver bullet, but what are the operational efficiencies that you can see getting unlocked by operators?</p><p><strong>Joseph Brennan: </strong>Obviously I have a pretty Zoba-forward take here. There is, rightfully so, a tremendous amount of focus on hardware. Everyone’s been thinking about the hardware and iteration loops and you guys have done a ton of great work on this and that makes sense. It is probably the biggest thing to change your economics and make the businesses work.</p><p>I think that people — just because they can’t see it — underestimate how far behind we are on the operations side. The way you were describing operators earlier in terms of it being an intuition-driven, challenging process, which is super inefficient. That’s still mostly how it works or if it doesn’t work that way, it’s very early innings.</p><p>I think people don’t realize that. We’re solving these problems. They’re hard to solve. You’re dealing with real-world hardware, very fast changing markets. Whether or not it’s raining completely changes the dynamic and the whole market.</p><p>It’s not uncommon for us to launch in markets and see 30, 40, 50, 60% increases as reasonable on ridership, which, similar to a scooter lasting two, three, four times as long, it changes the fundamentals of the business. And so we’re seeing more and more of that.</p><p><strong>Oliver Bruce: </strong>I just want to stress this — you’re seeing 50 to 60% improvements in some markets. In other words, you’re really changing it from a business that loses money to a business that actually makes money.</p><p><strong>Joseph Brennan: </strong>Yes. Yeah. That’s just on the gains in terms of ridership. Then you think about consolidating tasks.</p><p>Task consolidation is a real nerdy micromobility topic, but this audience might enjoy it. If you’re not doing it, you end up with people crisscrossing the city all day. There’s a battery swap and then there’s a pickup and there’s a drop-off and they’re driving a ton of miles.</p><p>If it’s planned in a way where you could trade-off between all of these decision factors, which you can’t do with intuition. Well, you can consolidate so much of this down and just really get a lot more out of each labor hour. So that is a big improvement on the opex side of the business. 30, 40, 50% increases in per asset ridership are feasible; on the operations side, the changes that are possible are as dramatic on the hardware side. I just think most operators haven’t been at that level of the business yet.</p><p>So the first thing you do when you start a micromobility company is get scooters on the road. Then the students break them and users throw them in the river and it’s really challenging. You’re losing scooters. And you do that for a while, and then you get to a point where you start tuning.</p><p>We’re early on in this industry is what I’m trying to say. I think people don’t appreciate that. No one’s using the first generation of scooters anymore. Like scooters that lasted two, three weeks are gone. No one does that anymore, but a lot of operators are still operating the similar way than they were early. I think that the big inflection point is anyone who tries to transition to swappables with an old model gets nailed. It’s really hard.</p><p><strong>Oliver Bruce:</strong> Really interesting. If we’re just talking logistics, how does it work for companies to bring Zoba online? How does that work? I’m going to do your sales pitch for you here.</p><p>How much access to data can you get and how do you make it make sense? Everybody’s running slightly different systems and stuff.</p><p><strong>Joseph Brennan:</strong> The basic way it works is that we’re building this optimization platform that needs to feed on data to learn about where there are gaps, how the operations run.</p><p>All of our scooter companies want to test this out, and we run it in simulation first. You start with a cut of your data. Historically, we look at the simulation and say “you are under performing your potential by 40–50%.” We actually do this before going live and we wouldn’t enter engagements where there aren’t meaningful gains, where it’s not worth it for the customer. But usually we say, okay, that looks great. And we then start generating task lists for your operational cadence. We start generating those tasks lists.</p><p>We are an API first company, so that means integrating into some other existing part of your stack. When you start to execute on these tasks, we start to measure the impact of those interventions, and that’s basically how it works on the data side. Most of the data we use relates to trip histories and vehicle history.</p><p>We want to know where the vehicles are at all times and what has been available and when it’s been under maintenance. We can actually get just about all of this from MDS. MDS wasn’t designed for this purpose, but Zoba can get an MDS token and get most of what we need there.</p><p>One surprising issue you run into with MDS is that there’s a lot of inconsistency in how it gets implemented. The Mobility Data Specification for anyone who’s not tracking the data part of this. You end up sometimes seeing that the actual IOT is fine, it’s just that it was mapped wrong internally or something. So that can be an issue.</p><p>But generally that’s how it works. And let’s say we get a month of data — it depends on the market. The length of time depends on how many rides you have and so on.</p><p><strong>Oliver Bruce: </strong>Pricing for you works as — do you do it as a fixed rate or how does that work?</p><p><strong>Joseph Brennan: </strong>Typically by vehicle coverage. The way we typically do it is we look at simulation. We often start out with a pilot, not always, but often. And then we start to go from there. And if you’re in very many markets, you know, let’s say you have 30,000 vehicles or something. Maybe we cover 8,000 initially, and then you ramp over time.</p><p>I will say that Zoba tends to be more relevant later in a company’s life cycle. If you’re very early, you have to be pretty good at operations to really get the most out of a very intense decision automation engine.</p><p>If you’re just trying to get the basics down it’s just challenging. That took us some time to realize. And so but if you can execute on this and we monitor how well these tasks are executed, the combination of the hardware, and the — I don’t want to call it low-hanging fruit cause it’s hard — but gains in operations are possible.</p><p>From our vantage point, these businesses are going to be very strong and people discount the hardware — sorry — they discount the operations side and only focus on hardware.</p><p><strong>Oliver Bruce: </strong>I had a conversation with Alan from Beam a little while back and his big thing was — I mean, he’s ex Uber, then Mobike — sorry — Ofo. His whole point was, “This is an ops game. That’s where we’ve found the most value. We didn’t even bother trying to do our own custom hardware, just standard diagnostic stuff. We didn’t really care. It’s the operations where we win or lose.” They’ve done very well. They’re the largest micromobility operator in Asia Pacific.</p><p>Cool. The business itself you started in 2016. This has obviously been a boom and bust sector. We’ve seen billions of dollars poured into scooter operators.</p><p>How has it been for you as a business to grow through that? And did you go through the boom and bust yourselves and what’s it been like for fundraising and things like that?</p><p><strong>Joseph Brennan: </strong>On the boom and bust, last year was obviously hard for anyone exposed to this space. For Zoba, because we’re a pretty deep tech company and we’re not bearing the brunt of these operations, which are so expensive, it was easier for us. We actually grew in head count and retained everyone. For us, 2020 was a heavy development year and we’ll be better for it in 2021. But we were fortunate to keep everyone in and grow.</p><p>In terms of the boom and bust of what we’re seeing, we also get quite a bit of visibility into the data, obviously, on what’s happening in ridership. And it has us pretty optimistic. What we’re seeing now differs depending on the region. Even last year, we were coming into what should have been peak season. Even though the lockdowns were pretty intense in a lot of places and public transit, for example, might be off 20, 30, 40% from where the ridership should be, micromobility in a lot of the markets is up a bit or only off very slightly, but, but basically where you’d expect it to be.</p><p>What that tells us as a business this year, is that as people start to move around, we’ll see a shift in modal share. We’ve obviously been hugely influenced by you and the “market for miles.” The market for miles has gotten very small, right? People just aren’t aren’t moving. But the modal share in that market from what we’ve seen was clearly healthy or growing. We’re very confident in this year.</p><p>On the fundraising side, the fundraising is hard. It’s always hard. Zoba, as a company, we feel pretty fortunate. We’re backed by some of the best VCs out there, including CRV, which is an early backer in DoorDash and Uber — sorry, DoorDash and Bird.</p><p>Founder Collective; Mark Cuban was involved since the very beginning; Aaron Schildkrout is a former head of data at Uber; Anthony Goldbloom, Founder and CEO of Kaggle; it’s a really great crop of investors who are helping us build Zoba, we’re really trying to build a category.</p><p>In terms of the general environment now when micromobility picks up, for companies raising in the space, it’s going to pick up. But you know, it’s a good time to be a remote work startup. It’s not a great time to be a transportation startup, but things will pick up.</p><p><strong>Oliver Bruce:</strong> Totally. The thing that’s interesting for me about what you guys have done is to have taken the analytic intelligence aspect and just really honed down on that and then worked out a way to monetize that aspect to it. One of the things that we’ve talked about in the micromobility world is that it’s an innovation, not an invention.</p><p>Over time, this whole sector is going to modularlize, The shared operators — shared operations makes the most sense as a franchise. You bring in the intelligence. cobbled together and you do a McDonald’s equivalent and someone buys the franchise and runs it in a city and that’s a business and you can see it happening with the Bird platform and others — but that hasn’t actually played out so far.</p><p>We thought that was going to be the case in terms of how the market would move. But I do think that over time, there’s no way that one large company can own and have assets on every corner in every city. That’s just not the way that it’s going to play out. So this idea that you’ll end up with lots and lots and lots of players, and then how do we build and ensure that the operational efficiencies that are maybe unlocked in Bogota, Colombia can actually get shared across to somebody who’s operating in some other places.</p><p>That’s where a company like yours is really interesting because it allows those learnings to filter up and then to filter down and then actually the overall industry itself learns and grows faster. So I’m excited. It’s an amazing time to be building scooter companies, because we wouldn’t have been able to build this even, you know, five years ago when we were doing Uber and Lyft, because they just wouldn’t have been anybody to buy your products.</p><p><strong>Joseph Brennan: </strong>That’s the way the world’s going. If you think about Zoba and this gets back to the fundraising and when you think about what a very big Zoba looks like down the road, the kinds of problems Zoba is solving are pretty hard and pretty specific.</p><p>Will you have some large companies, maybe a food delivery or Uber type. Companies of that scale, you will have good capabilities and what Uber has done is amazing. The economy is generally becoming more urban and more on-demand generally.</p><p>I think that these are some of the more fundamental problems and some of the harder problems. Zoba is a super specialized firm — this is all we do. We pour all of our resources into it. We have engineering resources just building simulation tooling to test certain parts of our system.</p><p>The focus you get as a startup like ours on this problem is hard to match even with a lot of resources. We’re excited. Our hope is that we can enable this transition in much the same way that it would be hard to build a lot of companies today without the modern API-centric tech stack — you’d have to do so much core.</p><p>We’re thinking that we’ll play a lot of the same role for sharing. As far as how the market plays out again, this is a topic where I wouldn’t dare throw my opinion in the ring here as you’re much sharper on the topic, but there’s a lot of incredible entrepreneurs building companies in the micromobility space. However, they end up evolving, whatever iteration it is, we’ll be able to serve it, which is exciting.</p><p><strong>Oliver Bruce:</strong> Well look, man, I’m conscious we’re up against time. So I just want to say thank you for this. This has been super interesting and, and I’m sure that the audience themselves will really appreciate being able to follow up.</p><p>We’re on YouTube, we’ll be able to follow up with a bunch of links. But is there any way for folks who want to track you down, the easiest way to do that would be, are you on Twitter?</p><p><strong>Joseph Brennan: </strong>Yeah, I’m not, I’m not a big Twitter person. You can find us at zoba.com and it’s easy to get in touch there. Linkedin as well.</p><p><strong>Oliver Bruce:</strong> Awesome. Well, Hey, thank you so much, Joseph. Really appreciate it.</p><p><strong>Joseph Brennan: </strong>Thank you.</p><p><em>To learn more about Zoba and our approach to operations and decision automation, get in touch at </em><a href="http://www.zoba.com/"><em>zoba.com</em></a><em>.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=9c71f7f12e08" width="1" height="1" alt=""><hr><p><a href="https://medium.com/zoba-blog/joseph-brennan-and-oliver-bruce-talk-operations-and-the-future-of-micromobility-9c71f7f12e08">Joseph Brennan and Oliver Bruce talk operations and the future of micromobility</a> was originally published in <a href="https://medium.com/zoba-blog">Zoba Blog</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Why micromobility will make a big comeback this summer, in one chart]]></title>
            <link>https://medium.com/zoba-blog/why-micromobility-will-make-a-big-comeback-this-summer-in-one-chart-8a21a3a6d568?source=rss-232e42b312de------2</link>
            <guid isPermaLink="false">https://medium.com/p/8a21a3a6d568</guid>
            <category><![CDATA[covid19]]></category>
            <category><![CDATA[transportation]]></category>
            <category><![CDATA[transit]]></category>
            <category><![CDATA[micromobility]]></category>
            <dc:creator><![CDATA[Jay Cox-Chapman]]></dc:creator>
            <pubDate>Wed, 10 Mar 2021 21:49:14 GMT</pubDate>
            <atom:updated>2021-03-10T21:52:23.584Z</atom:updated>
            <content:encoded><![CDATA[<p>On Monday, the New York <em>Times</em> published <a href="https://www.nytimes.com/interactive/2021/03/08/climate/nyc-transit-covid.html">data</a> showing the impact of COVID-19 on the New York City transit system. The news was dire: the system is still operating 75% below its 2019 capacity a year after the pandemic began.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*VeTKGg__0DsuR4pvps3uoQ.png" /><figcaption>Credit: Veronica Penney / New York Times</figcaption></figure><p>At Zoba, we care about cities and want to see transit succeed. After all, micromobility and transit are <a href="https://www.sciencedirect.com/science/article/pii/S1361920920308130">deeply complementary</a>. Still, the trends that the New York <em>Times</em> identified are not what we have been seeing in the micromobility industry. Thanks to an <a href="https://www.citibikenyc.com/system-data">open dataset from Citibike</a>, a bikeshare system operated by the NYDOT and Motivate, a subsidiary of Lyft, we can see that trips have not only recovered, but are growing again — even in their normal winter doldrums and even in the United States, which has had an uneven and unpredictable response to the pandemic. It turns out customers value the fresh-air experience of riding a bike to get around the city, as their normal patterns of commuting and travel have been disrupted.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*U-Ns11QwsrCS2TlY" /><figcaption>Citibike ridership has recovered and resumed growth</figcaption></figure><p>The Citibike program posted positive growth by early June 2020, though with some volatility likely caused by curfews around the summer racial justice protests. Ridership remained at 2019 levels throughout the summer. As the weather cooled, more riders stuck with Citibike to avoid taxis and the subway. The system posted double-digit year-over-year growth compared to last winter.</p><p>The variability in Citibike data when compared to the transit data is mostly explained by weather. Transit ridership is much more resilient to weather fluctuations than micromobility. Weather also explains the outlier spike on the right hand side of the graph, which is the result of very low ridership on those days in 2019: there was a major snowstorm <a href="https://www.nytimes.com/2019/12/03/us/winter-storm-snowfall.html">in New York City that week.</a></p><p>So, what does the astonishing recovery of Citibike tell us about the future of micromobility? At Zoba, we’ve been seeing similar trends emerge across the world and believe these trends are early indicators of a significant modal shift. In September 2019, Citibike served about 1.5% of total trips in New York; in September 2020, that share was 5%. Riders are foregoing traditional means of transportation that put them in close proximity to others; they are choosing the open air and social distance of micromobility instead. The significant decrease in transit ridership shows us that human movement in cities is still very low: micromobility has captured a larger share of a smaller pie. As movement resumes going into the summer, and as more and more people are vaccinated, we can expect to see a surge of demand for micromobility in New York and around the world.</p><p>Summer 2021 will be the summer of micromobility. If you’re operating a micromobility system and want help coping with increased operational volume, get in touch!</p><p><em>Notes</em></p><ul><li><em>Link to MTA </em><a href="https://qri.cloud/nyc-transit-data"><em>data</em></a><em>. Thanks to Chris Whong and the team at Qri for their work!</em></li><li><a href="https://github.com/qri-io/data-stories-scripts/tree/master/citibike-dailies"><em>Script</em></a><em> for assembling daily Citibike volumes.</em></li><li>Link to <a href="https://www.citibikenyc.com/system-data">Citibike data</a>.</li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=8a21a3a6d568" width="1" height="1" alt=""><hr><p><a href="https://medium.com/zoba-blog/why-micromobility-will-make-a-big-comeback-this-summer-in-one-chart-8a21a3a6d568">Why micromobility will make a big comeback this summer, in one chart</a> was originally published in <a href="https://medium.com/zoba-blog">Zoba Blog</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[How we measure at Zoba]]></title>
            <link>https://medium.com/zoba-blog/how-we-measure-at-zoba-db259555f1e4?source=rss-232e42b312de------2</link>
            <guid isPermaLink="false">https://medium.com/p/db259555f1e4</guid>
            <category><![CDATA[micromobility]]></category>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[transportation]]></category>
            <category><![CDATA[optimization]]></category>
            <dc:creator><![CDATA[Jay Cox-Chapman]]></dc:creator>
            <pubDate>Thu, 12 Nov 2020 22:02:00 GMT</pubDate>
            <atom:updated>2020-11-12T22:02:00.138Z</atom:updated>
            <content:encoded><![CDATA[<p><em>This piece is co-authored by </em><a href="https://medium.com/u/3295e61196f4"><em>Evan Fields</em></a><em>, Head of Data Science at Zoba, and </em><a href="https://medium.com/u/232e42b312de"><em>Jay Cox-Chapman</em></a><em>, Director of Product at Zoba.</em></p><p>At Zoba, we fervently believe in measuring how well our products work. In a previous post, we explored why spatial optimization is so important to mobility operators: if supply fails to keep up with demand, a fleet will forfeit up to a third of possible rides. Zoba Move, our deployment and rebalancing product, recommends optimal supply patterns for mobility operators.</p><p>Therefore, we are always asking — and helping our customers ask — how can we tell if a supply strategy was successful and what value it provided? How can we tell whether we’re deploying to ideal locations or rebalancing the right vehicles? We care about demonstrating the incremental rides and revenue that result from following Zoba’s recommendations because it means we’re helping our customers succeed. Today we’re going to discuss some of the metrics that we use to capture how well our recommendations are working.</p><h3>The real world is too complex for A/B testing</h3><p>As a first principle, the goal of any supply strategy is to improve some high level objective such as total trips served or total revenue generated. This holds whether you’re moving micromobility scooters, delivering packages, managing an autonomous vehicle fleet. Therefore, measuring the effect of a new supply strategy should be easy: just look at the change in the preferred key objective. This answer is both philosophically correct and borderline useless in practice. In practice, weather, seasonality, day of week, variations in fleet size, <a href="https://medium.com/zoba-blog/lost-rides-the-nature-of-shared-mobility-competition-b4d3f219445">competition</a>, regulation, marketing, hardware changes, etc. can hopelessly confound any analysis of a change in supply strategy. With so many factors, it’s impossible to determine to what extent a change in supply strategy caused a change in a high level objective. For example, if total rides increased when a new supply strategy started, was that increase caused by the new supply strategy or because it was the first warm week of spring?</p><p>Most software companies, when they want to test the effect of a change, set up an A/B test to check whether the change has a statistically significant effect on an outcome they care about. Zoba has to contend with messy real-world effects that are not present in a software-only user experience. Even switchback testing, a specialized form of A/B testing that tries to control for time-based effects, is difficult to use because of the difficulty of quickly pivoting real-world operations and seasonal effects.</p><h3>What about regression?</h3><p>To properly understand the effects of supply interventions, we need a way to control confounding factors. Many modelers will naturally look to regression as a way of looking at a market as a whole, or “top-down”. At Zoba, we’ve found that regression <em>can </em>work well enough for measuring the effects of operational interventions, but it’s an unreliable tool for two key reasons. Firstly, key mobility objectives are determined by an array of factors typically too numerous to enumerate — let alone quantify well enough for use in regression. Consequently, parsimonious regression models with available data tend to have large unexplained variance. It may come as a surprise that simply gathering more days worth of data does not solve the problem. Regression models covering long time periods have to account for slow-but-impactful market evolutions due to seasonality, changing travel behavior, hardware upgrades, new pricing models, and so forth¹. Secondly, mobility exhibits significant temporal dependencies. Think about how the rides that happen today depend on how the fleet was arranged this morning, which in turn depends on the rides that happened yesterday. These time dynamics have complex nonlinear structure and are difficult to capture in an interpretable regression model.</p><p>While top-down approaches like A/B tests or regression can be very useful, we find detailed, granular trip-by-trip and vehicle-by-vehicle data more practical to conduct natural experiments. For example, we can control for important day-by-day effects such as weather and day of week by comparing sets of deployments which occurred on the same day. We might compare ride or revenue outcomes for vehicles deployed according to Zoba’s recommendations on July 4th to outcomes for vehicles deployed <em>outside </em>Zoba’s recommendations on the same day. Since both sets of deployments happen on the same day and share the same weather, differences in outcomes between the groups are unlikely to be driven by day of week effects, the holiday, or weather conditions, and we get closer to isolating the effects of our supply strategy change.</p><h3>Not all metrics are equal</h3><p>Even with granular data, developing effective metrics requires subtlety due to the spatio-temporal interactions inherent in shared mobility. Users (and potential users) of shared mobility have vehicle preferences, typically for vehicles that are nearby, highly charged, and undamaged. Each user may choose the vehicle they deem best. Consequently, an idle shared vehicle awaiting a user competes with all nearby vehicles (from both the same and <a href="https://medium.com/zoba-blog/lost-rides-the-nature-of-shared-mobility-competition-b4d3f219445">competing </a>services) for rides. That means the next event that happens on a given vehicle depends on the locations and states of nearby vehicles. Those vehicles’ next events in turn depend on <em>their </em>nearby vehicles. Trace this network of inter-vehicle dependencies out, and you’ll realize that in general, the next event on any vehicle depends on the locations and states of <em>all </em>other vehicles in the market. To illustrate vehicle dependencies, we plotted the locations of vehicles in a live market (blue dots), with lines connecting vehicles within 200 meters of each other — close enough that they’re competing for rides. Notice how each vehicle competes with only a few neighbors, but almost the entire fleet is joined together by this “competition network:”</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*rOBnFcDhrmi1ub5x" /><figcaption>Each dot is a vehicle. The lines connect vehicles within 200m of one another.</figcaption></figure><p>Vehicle locations and states are in turn dependent on previous events: most obviously, a vehicle usually ends up somewhere <a href="https://medium.com/zoba-blog/supply-demand-imbalance-in-shared-mobility-cost-and-consequence-72a50007831b">because a customer put it there</a>. In short, we have competition creating spatial dependencies and rides creating temporal dependencies, and thus every vehicle’s next event depends on all previous events from all vehicles in the market.</p><p>Some common industry metrics are confounded by these spatio-temporal interactions. One such metric is the “time to first ride after deployment.” At face value, this seems like a useful metric: if I’ve placed a vehicle in a good spot, it should see a ride relatively quickly. However, “time to first ride” does a poor job of describing whether a supply strategy is optimal or not, for three reasons. First, next events on vehicles are subject to vehicle interactions, so other vehicles in the neighborhood of the deployed vehicle will have a significant effect on the outcome. If a vehicle is placed next to four other vehicles, the other four may get a ride first, even if there is strong demand. Second, a short time-to-first-ride is not always optimal; operators will often rightly deploy in the morning to capture demand for the afternoon commute. Finally, time-to-first-ride will be heavily skewed by the time of day that the deployment was done.</p><p>The web of dependencies between vehicles suggests caution — or at least epistemic modesty — when designing metrics. Because vehicles interact, no metric can measure how “good” a single deployment or rebalance action is. Indeed, the very notion of the quality of a <em>single</em> supply adjustment is ill-defined exactly because all contemporaneous supply adjustments jointly affect all vehicles. Supply interactions are complex — mobility is complex! — and no metric can capture a practically useful notion of how successful a supply intervention was while accounting for these interactions. This is doubly the case for simple, intuitive metrics. Instead, the best we can do is develop a bouquet of metrics, each of which uniquely captures something important about how well supply interventions work. And when analyzing metric data, we must remain aware of the inevitable biases of each metric. In general, we believe that these metrics actually <em>understate</em> Zoba’s impact, but we feel the tradeoff is worth it for making our impact easier to understand.</p><h3>How we measure at Zoba</h3><p>When a new customer begins using Zoba to inform their deployment and rebalancing optimizations, we keep track of exactly which supply interventions followed Zoba’s recommendations. This is essential; you can’t measure how effective an operational intervention was without knowing whether (or to what extent) the intervention was actually implemented. We then present the customer with a set of “per-vehicle” metrics that track exactly what happens to an individual vehicle after that vehicle is deployed or rebalanced.</p><blockquote>When a new customer begins using Zoba to inform their deployment and rebalancing optimizations, we keep track of exactly which supply interventions followed Zoba’s recommendations.</blockquote><p>Zoba’s effect can be estimated by comparing metrics for vehicles touched according to Zoba’s recommendations to metrics for vehicles rebalanced or deployed outside of our recommendations. Whenever possible, we perform a <em>paired </em>comparison between comparable sets of deployments or rebalances, such as deployments with and without Zoba that happened on the same day. For a given day, we look at the average metric for all Zoba deployments and all legacy deployments. So we can say something like “On November 3rd, Zoba deployments had a 90% ride probability, and legacy deployments had a 60% ride probability. Thus on this day, Zoba deployments had a 50% better ride probability.”</p><p>At Zoba, we’re continually working to develop principled metrics which capture some of the <a href="https://medium.com/zoba-blog/certifiably-optimal-scooter-and-car-sharing-307b6d050d2d">competing goals</a> when performing a supply adjustment. The remainder of this piece describes three such metrics,<strong> rides after deployment, 24-hour ride probability</strong>, and <strong>wasted deployments.</strong></p><h4><strong>Rides after deployment</strong></h4><p>Ideally, a user rides a vehicle from one high demand location to another, so that the vehicle can quickly serve another ride. This doesn’t always happen. Consider a commuter riding a vehicle from a train station to their suburban home at the end of the workday — the vehicle is unlikely to serve another ride for the rest of the day. To capture the notion that vehicles should be deployed such that users keep them in good circulation, we measure each vehicle’s <strong>rides after deployment, </strong>which is simply the number of rides the vehicle serves within the 48 hours after it’s deployed or until it’s re-deployed, whichever happens first. Why 48 hours? We find that for most markets, 48 hours is a long enough time horizon to give vehicles the chance to capture multiple downstream rides, but it’s also short enough to reflect that vehicles should get rides soon after deployment.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*Gmc1Zg-ce5Tut7i2" /></figure><h4><strong>24-hour ride probability</strong></h4><p>This metric measures the likelihood that a vehicle will receive a ride within 24 hours after deployment. After all, it’s bad if vehicles sit idle. We use 24 hours because it smooths over cycles of high and low demand and avoids time-of-day effects. The ride probability is a powerful and highly interpretable metric, but it can be subject to some confounding. In particular, recall that whether or not a vehicle gets a ride depends on the locations of nearby vehicles. An individual vehicle’s ride probability can be depressed by over-deploying in the same area. As an intuitive example, suppose Zoba recommends a single deployment at a street corner, but instead 20 vehicles are placed at the corner. Then Zoba would consider the <em>first </em>vehicle deployed as following Zoba’s recommendations, but with so many vehicles present, the Zoba-approved deployed vehicle might not actually capture any rides in the first 24 hours after deployment.</p><h4><strong>Wasted deployments</strong></h4><p>The problem with deploying 20 vehicles to the corner when a single deployment is optimal brings us to our next metric: <strong>wasted deployments.</strong> Deploying too many vehicles doesn’t lead to more rides than deploying just a few. Zoba captures this intuition by counting wasted deployments<em>, </em>calculated as follows. A market is divided into small spatial cells. If the number of vehicles in a given cell never falls below <em>k &gt; 0 </em>vehicles over a given day, then the same set of rides could have been served with <em>k </em>fewer deployments to that cell, and thus the last <em>k </em>deployments to the cell are wasted. The following figure shows two cases at the same location, one where the 2 scooter deployments were valuable, and one where they were wasted.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*glD0-6iUQrwv3doe9UFMrw.png" /></figure><p>By jointly analyzing the rides after deployment, ride probability, and wasted deployment metrics, we can form a holistic picture of immediate and downstream effects on deployed or rebalanced vehicles as well as how the other vehicles in the neighborhood are affected. When a suite of metrics tells a similar story of increased ride probability, increased downstream rides, and decreased waste — typically the case when Zoba begins informing market supply operations — we can be confident the corresponding supply strategy is working.</p><p>If you’re interested in optimizing your fleet and digging deeper into fleet performance, drop us a line at info@zoba.com.</p><p><em>[1] In our experience, the regression models of shared mobility supply interventions work best with about 60 days of data. With much less data, estimated coefficients have untenably large uncertainty. With much more data, seasonal effects and other long term changes add too many additional confounding factors. But even near that 60 day sweet spot, regression-estimated effects still have frustratingly wide uncertainty and have a habit of suggesting physically implausible results.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=db259555f1e4" width="1" height="1" alt=""><hr><p><a href="https://medium.com/zoba-blog/how-we-measure-at-zoba-db259555f1e4">How we measure at Zoba</a> was originally published in <a href="https://medium.com/zoba-blog">Zoba Blog</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Announcing Dynamic Pricing from Zoba]]></title>
            <link>https://medium.com/zoba-blog/announcing-dynamic-pricing-from-zoba-dddee5df34bf?source=rss-232e42b312de------2</link>
            <guid isPermaLink="false">https://medium.com/p/dddee5df34bf</guid>
            <category><![CDATA[transportation]]></category>
            <category><![CDATA[startup]]></category>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[scooters]]></category>
            <category><![CDATA[mobility]]></category>
            <dc:creator><![CDATA[Jay Cox-Chapman]]></dc:creator>
            <pubDate>Wed, 07 Oct 2020 06:02:32 GMT</pubDate>
            <atom:updated>2020-10-07T06:02:32.778Z</atom:updated>
            <content:encoded><![CDATA[<p>If you operate a shared mobility fleet, you spend a lot of time thinking about moving vehicles. Your drivers are on the road for hours a day, deploying or rebalancing vehicles, and these days, they might have added sanitization to their daily tasks. If you’re in moped or car sharing, rebalancing is especially expensive. However, there’s another group that moves your vehicles ten times as frequently: your customers.</p><p>Today, Zoba is launching our <a href="http://zoba.com/dynamic-pricing">Dynamic Pricing</a> product into beta. Dynamic Pricing recommends per-vehicle prices in order to shape demand in a mobility market. By using price signals to encourage users to circulate vehicles, operators enjoy additional revenue, reduced operating costs, and improved fleet performance.</p><p>Dynamic Pricing relies on the same underlying models as Zoba Move, our spatial optimization product used by some of the biggest operators in the world. You send us trip events, and we return pricing recommendations via an API call.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*EwkVcx_5V1sqO2nlZlH90g.png" /><figcaption>A Dynamic Pricing view of a demo fleet in Austin, TX, as seen in the Zoba Portal. Orange dots are vehicles we’ve identified for discounting to encourage circulation; blue dots are full-price vehicles.</figcaption></figure><p>Dynamic Pricing works by identifying vehicles that are in low-demand areas and are likely to be ridden to high-demand areas. Our platform then automatically recommends vehicles to be discounted based on their future potential for circulation. Users still get where they need to go, but enjoy a discount. My colleagues Joseph and Evan detailed our approach to our pricing models in a previous post <a href="https://medium.com/zoba-blog/dynamic-pricing-for-shared-mobility-zobas-approach-3c59ec461712">here</a>.</p><p>We’re looking for a handful of forward-looking operators to partner closely with us during the beta. Drop us a line at <a href="/service/mailto:info@zoba.com">info@zoba.com</a> or sign up at <a href="http://zoba.com/dynamic-pricing">zoba.com/dynamic-pricing</a>.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=dddee5df34bf" width="1" height="1" alt=""><hr><p><a href="https://medium.com/zoba-blog/announcing-dynamic-pricing-from-zoba-dddee5df34bf">Announcing Dynamic Pricing from Zoba</a> was originally published in <a href="https://medium.com/zoba-blog">Zoba Blog</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
    </channel>
</rss>