Mike Hudack
speaker
274 appearances
2 recordings
2 series
first heard Sep 2024
last heard Dec 2024
Mike Hudack’s voice in public audio — every appearance, attributed to the second.
Trend
recordings per month · last 12 monthsNo recordings in the last 12 months.Older appearances are listed below; set an alert to hear about the next one.
Appearances
I wrote the entire flight back. I wrote the entire cab ride home. I wrote for a couple hours after I got home. And then, you know, the next day I was like... we'd need to change everything.
Yeah, yeah. Yeah, well, I think one of the great insights, one of the things that I really learned here was that it's so important to allow people to make their own choices. It's just so important to let people make their own choices. Like if somebody wants a soggy burger and you've given them enough information, like they know how far away it is.
It's not like they have any illusions about what's going to happen to this burger over that period of time. Let them order the soggy burger. This extends to so many things that you build. There's this line you asked earlier about having more features or fewer features, you know, more buttons, fewer buttons, whatever.
You know, there's a line you can imagine a version of Spotify that only has a play button. And maybe you can thumbs up, thumbs down, and Spotify plays to you the best music that it can think to play to you at the moment based on the time of day, the weather, you know, what it knows of your likes and dislikes, whatever. You know, that
product actually existed pandora kind of worked that way maybe you would seed it with a song and it wasn't great to be honest like it never played you know i'm like i really want to listen to you know whatever right now i want to listen to pantera right now like i want to play that you have to give the user some level of choice i think it assumes that users are rational and actually a brilliant matthew mcconaughey commencement speech
Well, there are two limits, right? There's a limit where you expose the entire set of options and the entire set of buttons that you could possibly expose to a user. And on the other hand, you have only one button. The art is picking the plays. The science then is like evaluating when a controlled experiment looking at the data to understand whether or not you've
selected your point in this curve, this gradient, too far to the left or too far to the right, and then you just keep adjusting. And it might be different for different people, it might be different for different use cases.
It's a great question. I think that it's very hard to accurately interpret data, and it's very difficult to not lie to yourself with data. You can cut data any way that you want. I once shipped something into the... It's called the dive bar at the time at Facebook. It was like a social product, a social sharing surface. And the dive bar was this thing.
If you swiped the Facebook app to the left, it opened up on the right. It was originally used, I think, for messenger contacts. I think something like 75% of Facebook users opened the dive bar. And so there was an argument to be made that it was a great surface to ship into. The thing is that like 99% of those people opened it by accident.
and didn't want anything to do with it, and then immediately closed it. You know, so you can run all sorts of different analyses. You need to consider the entire set of things around you in order to understand it. And I've made those kind of mistakes thousands of times.
Yeah, yeah, for sure. And I think that the places where I've seen data abused the most is in customer service, where people will say, There's this famous, maybe apocryphal story of, I think it was like Amazon's customer service number, where they said, oh, well, you know, we answer all calls within, you know, some threshold of time. It was like in a WBR or something.
And I assume it was Bezos who was in the room and he was like, I don't believe you. And he dialed the number and, you know, he was on hold for 10 minutes before he was hung up on. And I think it turned out that like the analysis only included calls that were answered by Like all calls were answered within one minute.
One mistake I'll never make. again, that I made when I was young at my first startup is underestimating the competition or assuming that something that they are doing will not work or that there is this looming, impending thing which is going to prevent them from accomplishing that. No.
Like you have to have deep, deep respect for the people that you compete against and they are smart and good and they are trying to do the same thing that you are. And I think that that's like the first law of competing with anyone.
Yeah. You know, Deliveroo has one product, really, which is food. And then everything else is kind of a feature or a way of delivering that. So it's a weird way of thinking about it.
What is that? For sure. For sure. Well, the key is that you ship all of those things and carefully designed and controlled experiments. where you have a set of metrics that you're looking to improve. So you can imagine, for example, that driver chat is designed to reduce what's called rider experience time, the time that it takes a rider to get from the restaurant to handing you the order.
And there's often, you know, a couple of minutes at the end of RET where they're looking for, you know, looking for your apartment, you know, fumbling around or whatever. And you're sitting there being like, God, if the guy just hits 2B, you know, I'll have my food now. Do I really have to go outside? Whatever.
And so you can design an experiment which very clearly shows whether or not RET, like Writer Experience Time, decreases by the amount that you expect it to in the population that has messaging. What you do is you give messaging to, I don't know, somewhere between 20 and 50% of the users. And then you just look at the delta in RET between those two.
And you can do power analysis so that you know how big the experiment needs to be, how long it has to run. And then at the end of that, as long as there were no execution problems with the feature, and as long as you designed the experiment correctly, you're going to get an answer.
The answer might be that there's no statistically significant impact, in which case you should unship the thing, or maybe there's something wrong with it. It may increase RET, which would be a paradoxical outcome, and you then need to figure out why, or it may decrease RET, in which case you roll out the feature to 100%.
Showing 181–200 of 274 · page 10 of 14
← Previous
Next →