Uber Drivers Forum banner

how do you think trips are assigned?

  • Straight forward: closest driver always

  • Balanced between drivers regardless if theyre the closest or not

  • Hybrid Mix of the two but definitely not straight forward

21 - 40 of 44 Posts
Lyft's PAX-to-Driver pairing (matching) algorithm is absolutely identical to Uber's in these essential ways:
  • It's kept secret, the public (both PAX and drivers) have no way of knowing any aspect for sure. We occasionally are given peeks at engineering articles, but they do not include "proprietary implementation schedules" or "effectiveness evaluations".
  • Both claim that it's "closest first then next closest if first declines", which we know empirically is not always true.
  • Both frequently make changes to the algorithm, testing ways to increase profits, without any prior notice.
  • Logically, the algorithm must be different for different services: standard, Premier, Lyft-line, etc. Or UberEats for the competition.
  • We suspect the following, but have no way to verify other than personal observations :(
    • New Drivers. Using the fishing analogy, to get them hooked with "Signup bonuses" and reel them in with extra pings.
    • Idle Drivers. To get these "under-utilized" resources profitable again.
    • "Better" Drivers: High Customer Rating, Low Cancellation rate or High Acceptance rate.
    • Use of a destination filter.
    • Vary in some respects from city-to-city.
The same can be said about Prime Time and Hot Zones (and Surge), secretive and changed without notice (based on internal engineering group's A-B testing), like the matching algorithm. We suspect:
  • Claims they're used to "draw" drivers into high demand areas (and areas of predicted high demand) are not always true.
  • Chasing after them is (almost always) useless because most PAX are smart enough to "wait it out". So few pings are available in the "later part" after PAX who will pay any price have already been picked up. There may be a small bump in pings as the "intensity" decreases to the point that "thrifty" PAX are willing to pay.
  • Best for drivers, who have learned prior patterns to already be in the areas they know are likely candidates for Prime Time and Hot Zones.
 
In regards to the "better drivers" category: as a new driver (only began driving at the end of last month) I've averaged a 100% acceptance rate. This is due to complete ignorance of how to avoid picking up pax who are over 5-6 min from me or need to take vacation length trips from SF to Oakland area or SFO. And yes while the latter pays well, it sucks when you're trying to hit the rides metric for that driver bonus to get your free car rental.

Anyways, of the 3 weeks I've been driving in SF so far, 1/3 weeks my acceptance rate was less than 100%. That was the week I had to cancelled couple pax due to poor network connection. Bad network connection only happened when I was outside pink zone areas regarless of PT or not. i.e. up in the SF hills, in coastal beach areas etc. However, i have verizon coverage in SF which gives me 4G 99% of time.

  • earned the $130 rental bonus and had 100% acceptance rate in week 1. Was offered AHG week 1 as n00b driver

  • earned the $130 rental bonus but failed to earn the $180 rental bonus AND had 98% acceptance rate in week 2. Did poorly in performing in this AHG week due to network issues. And especially those LONG @$$hat rides from SF to SFO and Oakland area. These rides killed my daily 25 riders metric because 50% of my rides that week were these long line AND one shuttle service. I realize they keep trying/testing different things on us drivers to see if these make money. But shuttle service. In my 4 car Hyundai Elantra? Seriously?? That shuttle service ride request took me from Alameda Coast Guard station to SFO. Good money but 45 mins long due to PM rush hour traffic from Oakland around 880 N to SFO. Completely gave up trying to return to SF that one afternoon due to start of rush hour & Giants game. Made $650 that week after expenses & tolls. Not offered AHG for week 3.

  • earned the $180 rental bonus with 100% acceptance rate in week 3. Averaged 35-40 rides per day for 12 hour work days over a 6 day work week. Made $850 after expenses & tolls. Was re-offered AHG for this upcoming week tomorrow.
Texted Lyft support on why I wasn't selected for AHG in week 2 despite meeting the acceptance rating & $130 rental bonus. Support said drivers were randomly selected by algorithm. The details by which this was done was only known to the Wizard of Oz, evil doppelganger twin of Lyft CEO.

Seems like your acceptance/cancellation stats as well as your performance hitting the power driver/rental metrics bonuses do count IMO.....
****************

Oh yeah, one more thing. This past Friday 4/21, San Francisco had a major citywide sweeping black out in the NW grid of the city. Had at least half dozen cancellations & forcibly rebooted my phone since the Lyft network (and Google maps) were FUBAR on Friday. Blackouts at major traffic lights extended several miles of city blocks from the epicenter. About the same time, some areas in NYC and LA also experienced blackouts. Complete coincidence, but several of my pax (couple whom were IT workers for cyber porn e companies the likes of tw@tter & Oracle) agreed with my conspiracy theory it was a DDos attack. lol. Anyhow, while the power grid was down in SF, the shortage of land lines helped get the wireless network flooded with calls of ppl being stuck in high rise elevators, being unable to work in offices/schools and/or just getting in to work (as Bart stations into SF were down as well) etc. etc.

Power first went out with one of my first pax trips at 9 am who was stressed out when he discovered he had no power to keep the ice cream safe for his bachelor party. Got several news updates the power would be restored by 2 pm. This ETC apparently slipped to 4 pm over the radio around 2:30p that day. Besides inoperable work/school buildings (which let out workers & students earlier that day), several gas stations were also inoperable due to network & power issues.

Despite that, several major concerts were going on in the Golden Gate Park for some insane reason. And the SFO bridge was nearing gridlock by that point. Good thing the Giants weren't playing in town that afternoon. That's when I decided to quit right then to avoid the crazy town that was about to go down come rush hour. Good thing too. Turns out power wasn't fully restored until about 5pm or so later that day. Being poor driving Ms. Daisy anywhere outside SF city limits was paradise to being stuck idling at 0 mph in traffic busing Oakland line pax across that SFO bridge >,<
 
In regards to the "better drivers" category: as a new driver (only began driving at the end of last month) I've averaged a 100% acceptance rate. This is due to complete ignorance of how to avoid picking up pax who are over 5-6 min from me or need to take vacation length trips from SF to Oakland area or SFO. And yes while the latter pays well, it sucks when you're trying to hit the rides metric for that driver bonus to get your free car rental.

Anyways, of the 3 weeks I've been driving in SF so far, 1/3 weeks my acceptance rate was less than 100%. That was the week I had to cancelled couple pax due to poor network connection. Bad network connection only happened when I was outside pink zone areas regarless of PT or not. i.e. up in the SF hills, in coastal beach areas etc. However, i have verizon coverage in SF which gives me 4G 99% of time.
  • earned the $130 rental bonus and had 100% acceptance rate in week 1. Was offered AHG week 1 as n00b driver
  • earned the $130 rental bonus but failed to earn the $180 rental bonus AND had 98% acceptance rate in week 2. Did poorly in performing in this AHG week due to network issues. And especially those LONG @$$hat rides from SF to SFO and Oakland area. These rides killed my daily 25 riders metric because 50% of my rides that week were these long line AND one shuttle service. I realize they keep trying/testing different things on us drivers to see if these make money. But shuttle service. In my 4 car Hyundai Elantra? Seriously?? That shuttle service ride request took me from Alameda Coast Guard station to SFO. Good money but 45 mins long due to PM rush hour traffic from Oakland around 880 N to SFO. Completely gave up trying to return to SF that one afternoon due to start of rush hour & Giants game. Made $650 that week after expenses & tolls. Not offered AHG for week 3.
  • earned the $180 rental bonus with 100% acceptance rate in week 3. Averaged 35-40 rides per day for 12 hour work days over a 6 day work week. Made $850 after expenses & tolls. Was re-offered AHG for this upcoming week tomorrow.
Texted Lyft support on why I wasn't selected for AHG in week 2 despite meeting the acceptance rating & $130 rental bonus. Support said drivers were randomly selected by algorithm. The details by which this was done was only known to the Wizard of Oz, evil doppelganger twin of Lyft CEO.

Seems like your acceptance/cancellation stats as well as your performance hitting the power driver/rental metrics bonuses do count IMO.....

Oh yeah, one more thing. This past Friday 4/21, San Francisco had a major citywide sweeping black out in the NW grid of the city. Blackouts at major traffic lights extended several miles of city blocks from the epicenter. About the same time, some areas in NYC and LA also experienced blackouts. Complete coincidence, but several of my pax (couple whom were IT workers for cyber porn e companies the likes of tw@tter & Oracle) agreed with my conspiracy theory it was a DDos attack. lol. Anyhow, while the power grid was down in SF, the wireless network got flooded with ppl being stuck in high rise elevators, being unable to work in offices/schools and/or just getting in to work (as Bart stations into SF were down as well).

Power went out with one of my first pax trips at 9am who was stressed out when he discovered he had no power to keep the ice cream safe for his bachelor party. Got several news updates the power would be restored by 2pm. This ETC apparently slipped to 4 pm over the radio around 2:30p that day. Besides inoperable work/school buildings (which let out workers & students earlier that day), several gas stations were also inoperable due to network & power issues. Despite that, several major concerts were going on in the Golden Gate Park for some insane reason. And the SFO bridge was nearing gridlock by that point. Good thing the Giants weren't playing in town that afternoon. That's when I decided to quit right then to avoid the crazy town that was about to go down come rush hour. Good thing too. Turns out power wasn't fully restored until about 5pm or so later that day. Being poor driving Ms. Daisy anywhere outside SF city limits was paradise to being stuck idling at 0 mph in traffic busing Oakland line pax across that SFO bridge >,<
I agree that you, unlike many of us, have good reasons to maintain a high acceptance percentage. More important is that in SF, it is more lucrative (not that difficult) compared to a newer market in the suburbs, where there are fewer pings per hour and they average pickup time/distance is double or triple what you see.

Further, Lyft has not offered a single AHG since I started. By contrast, Uber has offered a guaranteed bonus each week for the past month. And if "power driver" was offered in my location (it is not), I'm not sure that +10% would be worth the extra hours and aggravation. Even then, I would not qualify, driving a car older than the 2011 cutoff.

In your place, I'd feel the same way you do. I suspect many drivers in major cities like LA, DC, NYC, etc. feel the same way you do, too. However, my environment, outside a major city, is quite different. The result is a very different attitude to driving for Lyft, not just mine, but 90%+ of the Lyft drivers in my location. I have yet to meet a single driver in my location that does not use Lyft as a supplement to their primary, Uber.
 
Lyft purposefully, and I'll add typically intelligently routes you with the primary goal of staying near places likely to get pings and you'll drop off the rider when Lyft gets around to it. The customer is paying next to nothing, (e.g, $5.16 all inclusive for 6 miles ), and the driver is paid time and distance.

If they don't like it, they can pay for regular Lyft and as I telly riders, they own me, and customer service becomes my goal. With Lyft Line I'm a bus driver and Lyft dynamically changes everything, but the rider saves $.
I want to become a Lyft Bus Driver
when will Lyft Line expand? my market [Fresno] currently offers Lyft/Lyft Plus, I am dying and bugging for Lyft Line expansion. At least it will help the growth of Lyft ridership in Fresno, CA.
 
First two weeks driving Uber almost non stop pings, 3rd week less pings 4th week lesser pings 5th week same as 4th week, my ratings have gone UP each week as well as acceptance rate and cancellations way down over time. I think they salt the new drivers with extra pings. I also think the app determines "tolerance rates" for the drivers for example what is the longest wait time for driver before they log off. What is lowest daily earnings before driver doesn't log on again, what is the longest they will drive to pick up pax, pool acceptance rate etc. This way they can shape overall system performance for the pax and drive revenues higher. To think that they are not doing something like this is very naive.
 
I want to become a Lyft Bus Driver
when will Lyft Line expand? my market [Fresno] currently offers Lyft/Lyft Plus, I am dying and bugging for Lyft Line expansion. At least it will help the growth of Lyft ridership in Fresno, CA.
You may be joking, but Uber and Lyft buses are expected to roll out with self-driving cars.
First two weeks driving Uber almost non stop pings, 3rd week less pings 4th week lesser pings 5th week same as 4th week, my ratings have gone UP each week as well as acceptance rate and cancellations way down over time. I think they salt the new drivers with extra pings.
Yes, it has been long suspected that Uber and Lyft "salt the new drivers with extra pings" to reel in the new fish. Most do not last when they realize the truth.
 
years ago I worked in the construction industry where we got paid piece work, developers always sent new people to the easy money jobs first few weeks, determined what they could produce in a day, then adjusted rates to their production rate then sent them to a new project with different rates. This is labor force management 101. I have been around long enough to bet my last dollar the Apps manage for overall consumer (pax) satisfaction and maximum revenue generation. I learned back in the day, never listen to company BS, find the reality and then turn a lemon into lemon aide. For example with Uber locate the consistent profitable trips and eliminate the money losers. We just need to learn what will give us more and better pings after the first few weeks.

As new driver I have been in the beach area with many ants that were closer to pax and I watched on multiple occassions that I got the ping and not the others.
 
The reality is not what the TNC's app (driver or passenger) is showing us.

The logic for matching riders with drivers is certainly a closely guarded trade secret, and is going to put the company's business interests in front of those of drivers and riders. Don't expect transparency or simplicity in how they use big data to do that.
 
I agree, but when the green light hub says "it's the closest driver and it's all random" this is total BS, don't use this info to make driving decisions to survive in this industry. Learn how to make lemon aide from a lemon, learn as much as you can from observation and don't listen to company BS.
 
Lyft servers are laggy, folks

Click and zoom your way around screen to force the servers to update your location and stay in contact with you

And know spots with good strength low-latency LTE signals (ALL 3g and most non-LTE 4g is inherently high latency)...signal strength is important, latency us even more important tho

Try to avoid getting stuck in AT&T Band 17 though, its latency at any strength and speed seems to be made of pure suckage.

There is an interesting series of articles on Lyft's engineering blog that confirms they have been conducting all types of weird experiments on both drivers and passengers for years..

It's worth the read if you can stomach the graphic technical details and diagrams: https://eng.lyft.com/experimentation-in-a-ridesharing-marketplace-b39db027a66e
Well, cat's out of the bag: they fail math forever.

That's why it sucks now. Their chief data miner is a clueless charlatan
 
I agree that you, unlike many of us, have good reasons to maintain a high acceptance percentage. More important is that in SF, it is more lucrative (not that difficult) compared to a newer market in the suburbs, where there are fewer pings per hour and they average pickup time/distance is double or triple what you see.

Further, Lyft has not offered a single AHG since I started. By contrast, Uber has offered a guaranteed bonus each week for the past month. And if "power driver" was offered in my location (it is not), I'm not sure that +10% would be worth the extra hours and aggravation. Even then, I would not qualify, driving a car older than the 2011 cutoff.

In your place, I'd feel the same way you do. I suspect many drivers in major cities like LA, DC, NYC, etc. feel the same way you do, too. However, my environment, outside a major city, is quite different. The result is a very different attitude to driving for Lyft, not just mine, but 90%+ of the Lyft drivers in my location. I have yet to meet a single driver in my location that does not use Lyft as a supplement to their primary, Uber.
Thx for the feedback on that. I guess it is market centric then given my success with SF v. Sacramento. Uber drivers in the Sacramento forum have a similar driving experience as you and a less than stellar view of LyfUber o_O
 
Had a pax accidentally cancel themseleves some 5 min into their ride. Told them to re-initiate the ride and I'd immediately accept it. That's when I found out the network pairing system was FUBAR. Pax got picked up by another driver even though they were sitting right there in the back seat --- with no Lyft drivers within a visual radius. And I got another ping shortly afterward which really sucked. Because the pax was not happy having to leave the car with all those groceries they filled the trunk with.
I had that happen too
 
Just so you know, you had the option to continue the ride to the destination, point out the time of arrival and do a support ticket as a driver, issue with ride, ended ride early and give the details. They will adjust both sides of the fare.

You had a legitimate ride so the insurance would be active, and any possible law enforcement issues would be covered as this is a known possible scenario with a work around plan. Also, rerequesting charges another base and minimum ride. Not talking about much, but it over charges the customer. Doing the right thing here is a great customer experience and very likely to result in a tip.

On the cynical side, I will gladly pick people up at the stores and take them home and help take stuff to the ground level door threshold. These are the decent human beings just trying to accomplish a basic task. If they start driving themselves because we make this a pain, we also won't take them to work or to the concert.

It is the sorry pos people that insist on you taking them 0.3 miles to the store and demand you wait 45 minutes to take them back I have issues with. These situations I don't care about my rating. 1 star me and complain. It took me 2 minutes to get to you. Someone else can get you. Hell, I can get 2 more rides in with those fares and tips and still take you home, but don't disrespect my time for 0.075 a minute.
While I want to believe that the insurance would be there in this situation if you completed the drive, don't want to bet my house on it.

This did happen to me but it was just after I arrived and confirmed arrival. We were loading the groceries and got back inside and the Lyft app didn't have an active trip...then the ping from my text went off and the text said the customer had canceled. She was loading groceries and her phone was in her purse so I don't think she hit the cancel button since it takes several clicks to do that. In any event, she did rerequest and it came to me and I did a support ticket to get her cancellation fee refunded. Had she not gotten me, I would have had her cancel and rerequest until it came to me.

If by some crazy reason she was never paired with me, then only then, would I have driven without an active trip on the app.

I have some riders that have my number and text me for a ride. Multiple times I've picked them up, have them request a ride when they're in my car and I always get the trip. Have done this on both Lyft and Uber.
I do this too and it always works on the pickup at their residence, but when i go to get them, which is usually in a busier section of town with restaurants and many Uber/Lyft drivers, it's damn near impossible to get them paired with me.
 
Okay. Seriously. Its just programming and code. Weve all seen some stuff that makes us wonder how straight forward trip pairings are. Its been posited that it could be straight forward and days are just different and requests vary. Its Also been hinted at that trips are balanced out between drivers and long short requests are spread around the drivers nullifying right place. Right time.

WhAt do we actully know??

I heaitate to ask if any lyft people will post but i know atleast drivers may.

Does anyone have any hard evidence of one or the other or Atleast some solid empirical evidence?
Yawn. Every one of us has ended a trip, only to have the passenger say, "Wait! I have to go somewhere else!" Then they reorder and the call goes to someone else. My evidence is my eyes. You get the closest person or the farthest person based on your need for earnings + the amount of tips you've received.

https://uberpeople.net/threads/how-lyft-uses-your-tips.140748/
 
2k Lyft rides and never had this happen before. I am enroute to a pickup 10 minutes away with "last ride" selected. I am about 2 minutes into drive when gps turns me around and a new name appears as my rider. No notice from Lyft, nothing. It is a much closer drive for me and an airport run with 50% PT.

But here is the interesting part, I am on a $22 AHG with my one ride requirement, a $4.50 shorty, already met. I was just online to make my 50 minute requirement. While nefarious motives might be inferred, another Lyft car pulls up as I am leaving, so I question my riders carefully, and I was always the car they had ordered from their side. So I think a system glitch is more likely the cause, hope the other driver was compensated. Of course this busted my hour, but fortunately it was a single AHG hour that won't lower an average.

(No Line here yet, and if they implement without an opt out I am done.)
Happened to me the other day too. En route to a pickup one a 25% PT and suddenly hear a ding and see my pickup has been changed and new passenger. Get through the ride and notice they switched me to a non PT ride. I used to like Lyft but they've been doing crap recently that's just pissing me off
 
I think it's proximity based, driver/pax rating as well as Destination Filter users .

If you're using Destination Filter you get priority over rides going in your direction even if you're not the closet . I also think that they try to pair new riders with higher rated drivers . All of this is just speculation (except the DF thing )
 
I was in front of a bar, my Uber/Lyft sign was on and I had some guy asked me if they could get in the car and request me. It was drizzling a bit so I agreed, they told me where they was going and it was a decent size trip. So they got in they chose Uber first so I got Uber on the screen, they requested and the ping went to another driver like a block away. They cancelled. They tried Lyft I logged of Uber and left Lyft standing they requested and the ping went to another drive on the street behind us. They cancelled again and re-requested, same results different driver until the 3rd. attempt they got me. I had pings non-stop that night before this happened so I figure Lyft was trying to balance the average with all the drivers, regardless of who is closer.
Just empirical, nothing scientific but I have noticed that several times when we are chatting with other drivers.
If you are too close to pax the system will not see you. I will either stick my phone out window or ask pax to stand a few feet from my car...otherwise high chance request will go to other drivers blocks away. If you get pax on your side it is very easy to ask them to cancel and request again so you get the ride. Most pax tell me they prefer to stay with me.
 
21 - 40 of 44 Posts
This is an older thread, you may not receive a response, and could be reviving an old thread. Please consider creating a new thread.
Top