How Fast Should Your Site Be?

[00:00:00]
>> Todd Gardner: So we just got done talking about testing different ways we can test the various tools. Now we're gonna talk about setting goals. We're gonna figure out how fast does your website actually need to be? There's a couple of things we need to talk about. One is, how fast is fast enough?

[00:00:17]
Who gets to decide that? And then you need understand some things about their users, about your users. So how fast is enough? Fast, what fast is is very subjective to who your users are? Different user groups are gonna think fast means different things has largely determined based on what kind of app are you building, and for whom?

[00:00:43]
If you're building something that is showing developer stickers, it's got to be pretty fast cuz your audience is developers who are impatient, and the value of the thing I'm delivering is relatively low. It's not super important to their day to day life, whereas, if you're doing something super important like applying for a loan or filing your taxes, people are willing to wait a lot longer for those higher value things.

[00:01:11]
When people think about performance, it's not the absolute numbers that always matter, because users tend to remember things slower than how they actually happened. When something happens, if they feel like it interrupted them, even if the actual is technically Okay, often they'll perceive it. A human will perceive it as slightly slower than the objective number, and going back the next day, they'll remember it even worse than it was.

[00:01:38]
And so the raw numbers don't always completely matter depending on who your users are. Now this comes back to psychology. So web performance is not just about the tech, it's about the people. And people are willing to wait different amounts of times for different things. There's a couple of things that came out.

[00:01:59]
This was some psychology studies that happened in the 60s. And there was a couple of just general learnings. This isn't specific to computers, this is general. This was a study of people waiting in lines. And what people felt and reactions to it. And there are some key learnings.

[00:02:16]
One, people want to start. If you're going to do something, they're anxious to get started with whatever it is they came to do. They want to see the sticker, they want to fill out the application, they want to do the search, they wanna do the thing that they came here for and they're anxious to get going.

[00:02:33]
If people are waiting and they're bored, they'll remember that wait longer Better than it actually was. So a lot of times you'll see people like throwing up fun animations or like little games or something to distract the user when they're waiting, because they're trying to get around this because if they're bored while they're waiting, like we were sitting there waiting for web page tests to start that, even though that probably wasn't objectively that long.

[00:02:59]
It felt like forever to me, because I was bored waiting for it to be done. I wanted to get on to the next thing. Anxious waits feel slower. If I have a lot of anxiety about this action that's happening, like showing a live performance result in front of a live audience.

[00:03:18]
I wanna I feel like that's slower than it actually was. I'm anxious about the results and that it doesn't break. If I don't understand why I'm waiting, it feels longer. Again, why did that take so long? It never took that long in any test or rehearsal that I ever did.

[00:03:35]
I didn't understand why it was slow, and so it felt slower. This was actually a great case study that I could call back here. So unexplained waits feel slower if I don't understand why I'm waiting. Uncertain waits if I don't understand how long it's gonna take, like I have no context.

[00:03:52]
If this is gonna be a minute, an hour, a day or a week, that's going to feel longer than it actually is. And people will wait longer for higher value things. So a developer sticker, maybe not willing to wait very long. A insurance application, willing to wait for that, might want to wait for that.

[00:04:14]
In fact, sometimes you have to be intentionally slow if you're a particularly high value. The most notable case of this is Turbotax. If any of you use Turbotax, there is an intentional slowdown in Turbotax. When you are all done with your taxes and you hit submit. There's this animation banner called looking over every detail where it's like shuffling through papers and the Atlantic went through the source code, or the decompiled the source code.

[00:04:43]
There's nothing actually, it's doing it is a fixed time weight. It knows everything it needs at that point. TurboTax in their studies believe the user will trust their system more if they slow it down like it's actually doing something. Like it's not just a thin wrapper around IRS APIs that have existed and there's no reason for TurboTax to exist anymore.

[00:05:08]
I hope Intuit isn't a subscriber. [LAUGH] so this is a case that people not only will wait for value, but they want to wait longer if something is perceived as high value.