Top 5 things we learned from our first million RideAustin rideshare trips
By Andy Tryba & Marisa Goldenberg
We never intended to go into the ridesharing business. But after Uber and Lyft ceased operations in Austin on May 9, 2016, we in the Austin tech industry felt passionate that we shouldn’t just sit around and complain — we should look forward and solve our own problem.
Lost in the heated debate of Prop 1 and the two market leaders abruptly leaving was the fact that there were thousands of Austin community members (drivers) that lost some or all of their current income source. Additionally, there were 125K trips a week that Austinites were taking to live their lives. These riders aren’t just 2 a.m. bar hoppers — these are Austinites that depend on rideshare to go to the doctor’s office, to see their families, to go to work, etc.
This is why we started RideAustin. Built by Austin for Austin. This is also why it’s a non profit.
Contrary to popular belief, nonprofits are not all poorly-run-socialists-with-crying-baby-videos-to-raise-wasted-money. There are great nonprofits that operate day in and day out that are better run than most companies (Seton Healthcare here in Austin is a standout nonprofit example). We intended to be one of them. As a nonprofit, we believed that as we got more efficient in our operations, we could pass along those savings directly to the riders (via lower fares) while paying the drivers a larger percentage of the trip fare — instead of maximizing profit margin. This doesn’t mean we don’t care about margin — it simply means we put it all back into the business. We are doing this today by taking 0% commissions out of the drivers earnings on our ‘Standard’ class vehicles.
We also believed that RideAustin should not only help riders and drivers, but also become a community asset. This is why we developed the ‘Round-up’ feature where riders can optionally donate the remaining change on their bill to a local charity they choose in the app.
How have we done so far? Well, just 8 months after launch, we will to cross our 1 millionth trip this week. With 45–50K trips completed every week, we believe we are Austin’s #1 volume rideshare. RideAustin drivers have earned over $13 million since our launch. And over 60K RideAustin riders have also elected to ‘Round-up’, raising over $100,000 for local charities.
Our journey has had plenty of ups and downs, and we are grateful for the support of the Austin community. There is still much to learn and improve upon. We also believe that the ridesharing industry today lacks transparency and openness of data. Ridesharing is an important part of the future of transportation — why shouldn’t we all share data to make it better and better? We intended to do our small part to add visibility into this space, so we openly publish all our ride data weekly and have started this blog to share even more data. It’s through open collaboration that we can all get better…
Our top 5 learnings
- Following local laws doesn’t ruin rideshare (at least in Austin)
The center of Austin’s Prop 1 rideshare debate was around fingerprint background checks — something Uber & Lyft say breaks the ridesharing model. Uber & Lyft are so passionate about this topic that they walk away from hundreds of millions of dollars by leaving cities that pass laws around it. Jury is still out in Austin on what type of impact fingerprinting created. But we’ve fingerprinted and approved over 4,000 drivers to date, and we approve another ~500 per month. 99% of our driver miles since November are from fingerprint cleared drivers.
A fingerprint background check costs about $40. Someone in the State of Texas got a sweetheart deal so that only one provider of fingerprint background checks was approved (MorphoTrust). The limited locations caused a heck of a delay just after Prop 1 passed, but after the initial rush died down, a typical visit for drivers takes about 10 minutes. After getting fingerprinted, the driver and the TNC company can receive the results back from the city transportation office 3–5 business days later — and the driver is approved (or not).
As of 2 months ago, there were 5,235 TNC drivers permitted by Austin Transportation. A majority drive for multiple apps though. According to a recent survey of active RideAustin drivers, 66% of respondents had driven for other TNC apps in the past 30 days. In retrospect — since the city fingerprint check works for all TNCs, we probably should have teamed up with the other companies in Austin to split the cost (we spent over $65K to cover fingerprinting costs in the first 2 months alone).
Regardless of the debate around impact — it’s true that every additional requirement in the driver application process adds friction. We know now that many of the former Uber super part-time, but active, drivers (1–5 hours / week) no longer do rideshare. There were rumors that there were 10,000 drivers under Uber & Lyft — and with only 5,253 today — that would imply a reduction in supply for Austin by 47% (unfortunately — Uber & Lyft don’t share their data so the 10,000 number has never been verified). But what we have seen is the friction in our signup system means only 60–70% of applicants meet all the requirements in a typical month. To date in Austin, these reductions haven’t been a barrier to having enough drivers for 3–5 mins pickups in the highest demand locations in most time windows — but in outside suburbs where supply is less — their pickup times are substantially higher.
To be clear — we are not for regulation that hinders innovation. But in this case, it’s unclear if the local regulations hurt us on our accelerated volume growth. It is clear that the fingerprinting didn’t ‘ruin’ ridesharing (for now at least). Our first 100K rides took 94 days and our latest 100K took 15 days. We continue to accelerate as we onboard more drivers.
RideAustin continues to gain market share through seasonal fluctuations
2. Building rideshare technology is challenging, but not impossible
It only took a team of 6 full time engineers about 4 weeks to get our initial version of RideAustin up and launched. We didn’t sleep much during that time, and our initial beta customers will tell you that the first versions were rough around the edges. But we dramatically limited the scope of the product (iOS only, 1 car type, limited service area, etc.) to reduce our time to market. We were also incredibly fortunate to have had various experts in the Austin tech community pitch in to help us and donate some of the initial technology we built on.
The technology stack consists of 4 interdependent pieces — rider apps (iOS and Android), driver apps (iOS and Android), the dispatch server & database and the management console. For speed of deployment and scalability, we rely heavily on several third party services such as AWS, Google, Twilio, Stripe and a host of others. At the beginning, we knew we had no idea how to build a rideshare stack. To overcome our lack of knowledge, we’ve been keen on fast releases to experiment and learn from each feature. As a result, in 8 months, we’ve had 58 releases of the rider apps, 59 releases of the driver apps, 65 server releases and 25 management console releases. As we’re sure with any consumer app, there is never a shortage of very direct feedback (which sometimes comes in form of screaming at you at 3am) and infinite feature requests. We’d like to think we get better with every release — but sometimes you take 2 steps forward, 2 steps backwards and 2 steps in some other random direction.
We will go into more detail on the building of the technology stack (and spectacular fails along the way) in another posting. But our overall conclusion is that with the right team (we now have 15 engineers) it’s not impossible to build a ridesharing application. The complexity comes in optimizing it for scale as well as systematically reducing costs of each component to fit into the narrow margins of ridesharing.
3. There’s a reason Uber lost nearly $3 billion in 2016
The rideshare business is extremely expensive and low margin. RideAustin would not exist if it wasn’t for the generosity of the Austin tech community that has donated over $7M so far in the 9 months. And this, by the way, doesn’t include the value of the generous in-kind or discounted services and technology that has been donated from the Austin community.
Most people are unaware of the costs associated with every trip — we’ve lost $4M on just the trips alone (we’ll come back to operating expenses later), mostly from driver and rider incentives, and with insurance and Google API calls. To illustrate the magnitude, below is a snapshot of our peak losses per trip (September) and our transition to break-even gross margin.
As we accepted more rides — we lost more money. What a great business… Oh yeah, non-profit…
Unfortunately, that’s the way scale marketplaces work. You have to over-invest on customer and supplier acquisition to ensure the marketplace functions. Only after you reach ‘scale’, your costs (sometimes) reduce dramatically. It is well known that Uber and Lyft have had to subsidize driver pay and ride costs in order to gain market momentum. And the same is required of any new TNC. Even though Uber and Lyft were no longer operating in Austin, there were numerous other companies who entered the market — who all had a substantial head start on us in technology development and operations infrastructure. If we wanted to be #1 and start experiencing the cost reductions, we had to pay to get there.
Similar to the 1849 gold rush where it was the really the ‘suppliers’ that made the money (like Levis Strauss jeans) — the current real margin winners in the TNC space are the folks selling to all TNCs. For example, there is only one insurance provider in the market who will cover TNC start-ups (we’ll keep them unnamed). And as a brand new company and no history of success in the rideshare space, we had ZERO leverage to negotiate the rates. In September (before we renegotiated — thank goodness), we averaged a crazy $2.50 per trip on insurance. This was basically all of the profit per ride and then some. Good lesson for entrepreneurs from the insurance provider — be the horizontal player and don’t care about who wins…
Over time, we got better and were able to focus on reducing per-trip costs. We did this both on the tech side (such as making substantial reductions to our Google API calls by modifying our dispatch algorithm and reducing our AWS instance sizes and moving to Aurora) as well as on the operations side (more intelligently managed rider and driver acquisition costs and negotiated insurance again at the 5 million mile mark). But even with these cost savings, we would still lose $1.50 on each trip — certainly not economics that lead to a sustainable business.
This week, we’ve increased our booking fee from $1.50 to $2.00 on all trips. And we’ve also added a processing fee of $0.30 + 2.3%, with a minimum fee of $1.00 to all trips. These changes get us to break-even gross margin.
However, we still have substantial operating expenses — over $3M to date. On the operating expense side, the bulk of our expenses have come from headcount costs across our engineering, support, and driver/rider acquisition teams (~$1.5M). The next largest expense ($700K+) has been driver acquisition, with $300K+ spent on driver recruiting incentives and $300K+ on background checks (3rd party online checks and our fingerprint reimbursements). Rider acquisition has cost us $400K+, with $100K in airport advertising and $90K in driver bonuses for rider referrals. The rest of the rider acquisition expense has been spread across digital advertising, traditional marketing, and street teaming.
And there are still plenty of other expenses — rent (downtown onboarding location with a large surface parking lot) and other outsourced tools and services (e.g. legal work).
We will continue to work on additional cost reductions and increasing our mix on SUV, Premium, and Luxury rides to generate enough gross margin to cover our operating expenses. Until then — thank you Austin community for the donations — please keep them coming :) …
4. Supply & demand are hard to match perfectly
Riders want to be picked up in 3 minutes or less. Drivers want a passenger in their car 100% of the time. This is not easy to pull off across 300+ square miles, every hour of every day.
Right after Uber and Lyft pulled out of Austin, 10 minute wait times were the norm. Most riders didn’t mind waiting — they were just happy to have ANY way to get around town. But as the rideshare apps in Austin scaled supply, wait times came back down to near the 3–5 mins of Uber and Lyft levels.
You learn in Econ101 about the concept of supply/demand — they meet somewhere — and rainbows and butterflies emerge. In the rideshare world — these lines meet for 1/1000th of a second (at best) — and then you’re either over/under supplied or over/under demand. The result of that is that at any moment — you’ve got either riders complaining they can’t get a car or drivers complaining that there are not enough rides for them. Throw the complication that ride demand can come from anywhere in the city and drivers can be anywhere in the city, hourly fluctuations can change demand 10x, different car types have different demand curves — and the intersection of those two lines really feels like finding a leprechaun.
This issue was particularly challenging in our first real ‘test’ — the University of Texas home football game vs. Notre Dame on Labor Day weekend. Unfortunately, we were only 83 days into our service being live, and we didn’t have nearly enough drivers to accommodate the demand. As a result, we disappointed the community and left over 2,500 riders without rides — despite doing a (then) record 17K rides that week.
What we’ve learned is that we have to continue to onboard drivers for those peak demand times. We have over 4,000 drivers today, but over 90% of them are part time. On recent weekdays, only 250 or so are working at the same time, and about 1,100 drivers log on at least once. On a typical Saturday, we have around 1,600 who log on, and peak events (like New Year’s Eve) can draw around 1,000 concurrent drivers. Since 60–65% of all rides occur over the weekends, we need as many drivers during that time, especially when the downtown bars let out at 2 a.m. Heat maps by hour help us analyze, but also being proactive on communicating with the drivers on when and where we expect demand helps everything run a bit more smoothly.
Several weeks after our UT-Notre Dame flop, we launched our own version of Uber surge. Increasing prices definitely brings out more drivers while suppressing demand — and that capability has been important to continue improving our balance while we scale. But instead of an army of PhD data scientists and advanced automated algorithms, it’s still up to our small operations team to manage our dynamic pricing. We have a long way to go.
Occasionally, our demand and supply line up perfectly — and drivers, riders, and the RideAustin staff love it. But fulfilling Austin’s intensely concentrated pockets of high volumes (e.g. weekend downtown bar close time, and major event exits like home football games and ACL Music Festival) is still a challenge. It makes us feel a little better to hear that the market leaders faced the similar peak demand challenges when they were a year ago. But we still need to keep improving our supply — while increasing our non-peak window demand. If drivers aren’t kept busy, they get disenfranchised and leave your network (or just alternate between multiple TNC apps).
5. Enormous social good can come through small change
The ‘Round-up’ feature was something we had in mind since the initial inception of RideAustin in May. It turns out this was a great feature — and one we’re probably the most proud of. Small change can add up, and RideAustin passengers have rounded up over $100K for local charities so far. Almost 25% of our weekly trips have a round-up included (average round-up is $0.50). At our current run-rate, we could easily surpass $300K in charitable contributions in 2017.
We knew Austinites love animals — but we were surprised to see that 25% of total contributions have gone towards animal-related organizations. The $12K in Austin Pets Alive! Round-ups roughly covers their entire annual budget for specialized medical equipment. The APA! clinic treats more than 7,500 animals per year.
Central Texas Food Bank was another big beneficiary from our round-ups. The food bank can provide 4 meals for every dollar raised, which means RideAustin passengers covered over 30,000 meals in just 8 months.
Below are the round-up contributions to-date. Currently, we have nearly 50 Austin area charities to choose from in the app, and will be rotating every 3 months to provide opportunities for more non-profits, big and small.
In the next update, one of our new charities will be in honor of one of our drivers whose son recently passed away. When we discovered that her family has fostered and adopted several children with special needs, we checked our round-up wait list and sure enough, we found a great match — a nonprofit that promotes adoption of children with special needs, Hands Anonymous.
What’s next for RideAustin?
If RideAustin were a human, we’d be getting close to crawling for the first time. And that seems just about right. We still have a lot to learn, but we’re proud of the accomplishments to date and are incredibly honored to have been embraced by the Austin community. You have our commitment wake up everyday and continue making RideAustin better and better. It won’t always be perfectly smooth, but we’ll be open and honest with everything we do.
There are, however, some uncertainties in the air around several proposed statewide regulations on TNCs that would effectively overturn the Austin vote (as well as Houston and San Antonio’s city regulations). It’s yet to be seen if these proposals will pass the State Legislature — but we have seen donations to RideAustin slow down substantially with this uncertainty looming. Regulatory uncertainty always cause folks to pause — and in our case — the pull back dramatically impacts what we can do for the Austin community.
We will be publishing this blog on a monthly basis to share even more. If there are specific questions you have, we welcome suggestions for future posts.
