So this post is more public service announcement than an on topic finance take. However, you’ll see there is also a lesson here about how statistics can be misleading which is certainly applicable to many other areas such as corporate earnings, reserve estimates, valuation metrics, etc.
Are Uber Ratings “Fake Reviews”?
Yes, indeed, they are. Let me give you an example. I was trying to get home from an event Saturday night. A driver with a 4.93 rating took my fare. Long story short (more details below), he no showed because there was traffic. I had to get a second driver. This one had a 4.92 rating, just below the first ride… but he showed up.
So did I give the 2nd driver 5 stars and the 1st driver 0 stars? No, because I couldn’t rate the first driver because I didn’t “complete” a ride with him. This is the big, giant loophole!
How Often Does This Occur?
A lot. If this were an isolated incident, it wouldn’t really affect the ratings. In downtown settings or at airports, it happens all the time. I would say at least 20% of my rides get dropped. I am not the only one with this problem.
What really makes it painful is it becomes a game of chicken. You can see the estimated arrival time going up. You can see the driver not moving on the map. You text and get no answer. You know the person isn’t coming.
However, if you give up and drop them, you pay a cancellation penalty. The driver even gets part of this fee! So, you have to wait until they finally drop you and move on to another fare.
This can take a long time. On Saturday, it was close to 20 minutes (I knew fares were going to be much higher if I rebooked because I got out of the venue early while the fares were lower and, if I had to retry, I’d face surge prices).
Yes, my time value of money suggests I should have dropped him earlier and hoped Uber would refund the difference in fare but I was conducting an experiment so I waited it out!
Why Does It Occur?
Why does a driver accept a ride and then decide they don’t actually want it? Well, there are a number of possible reasons.
Some are trying to scam you for the cancellation fee. However, more often, they realize you are in an undesirable location (e.g. getting out of an arena where there is a lot of traffic) or that fares are surging and they can get more elsewhere or they cheat and look where you are going and decide it will be too hard to find a fare nearby after they finish your ride.
Basically, they act like NYC taxi drivers! However, at least the jerk cabbie tells you right away he won’t take you rather than thinking about it for 15 minutes before saying no!
Does Uber Care?
Apparently not. I’ve complained before and asked to be able to rate drivers who ditch you and was told that is not an option, as mentioned earlier. Why? They won’t say.
How Biased Are the Ratings?
Let’s do some examples. Take three drivers each rated 4.90. If one is honest and never ditches a fare, he deserves the 4.9. Our second driver is only partially unscrupulous. Let’s say she cancels 10% of her rides. If you have 90% 4.9 and 10% 0, she would drop to a 4.4. That’s pretty noticeable.
| Official Rating | % Ditched Rides | Pro Forma Rating | |
| Honest Abe | 4.9 | 0% | 4.9 |
| Shady Sally | 4.9 | 10% | 4.4 |
| Scamster Scott | 4.9 | 30% | 3.4 |
Next, we have the hard core abuser who drops 30% of his rides. At 70% 4.9 and 30% 0, he falls to a 3.4! Nobody wants a ride from a 3.4 so Scamster Scott is out of a job!
And he should be out of a job. He is abusing the system and hurting Uber’s reputation. I take more taxis and less Uber than in the past because I can’t rely on Uber consistently anymore. So yeah, that person should be unable to find rides. That is the whole point of having the rating system!
Recall, in my case, the dishonest driver had a 4.93 while the compliant one had a 4.92. The scoring system is clearly broken!
Don’t Online Product Reviews Have Similar Issues?
Absolutely. There are obviously plenty of fake five star reviews from people who never bought the product but either work for the company or are paid to write positive reviews.
The difference is I think there is a general skepticism of online product reviews. I don’t think the same wariness applies to Uber (and presumably) Lyft driver ratings.
So What’s the Fix?
The obvious fix is to let riders review anyone who accepts a ride rather than anyone who completes a ride. If the purpose of user ratings is for customers to be able to accurately assess the quality of the driver, then any information that affects quality should be included rather than suppressed. Otherwise, ratings are just a ruse like inflated grades or participation trophies.
I think the way this works is that Uber drivers ratings go down for cancelling too many rides. So you’re right – they are allowed to cancel up to a point, obviously they can’t cherry pick every transaction, but they should be allowed to cancel some rides.
Yes, and that’s why they try to make you cancel by ghosting you. They don’t get penalized if the passenger cancels because they don’t want to wait any longer for a no show.