Episode 10

Headshot of Tammy Everts

with Tammy Everts

Our guest this week is Tammy Everts. Tammy is a Senior Researcher and Evangelist at SOASTA. In this episode, we discuss getting started with performance budgets, why you should focus on article and product page load times instead of homepage load times, how to build a case for improved web performance to your company or team, the real impact of performance on the bottom line, and more!

Direct Download

Transcript:

Katie Kovalcin:

Welcome! You’re listening to The Path to Performance, the podcast for everyone dedicated to making websites faster. I’m your host, Katie Kovalcin.

Tim Kadlec:

And I’m your other host, Tim Kadlec. This is Episode 10, Katie!

Katie:

It’s a big milestone for us; we’re in the double digits now.

Tim:

Yeah, it’s kind of a big deal so we should have baked a cake or something; that’s the only problem. When you have a celebration like this, you have to have baked goods and we didn’t do it.

Katie:

We need a soundboard that has kazoos and confetti and stuff, although I guess confetti doesn’t make a sound but you know, fireworks…

Tim:

Only if it’s like really heavy confetti! Little rocks or something, I don’t know.

Katie:

Cookies?

Tim:

Could be potentially dangerous…see, I like that, yeah. Combine the baked goods and the confetti.

Katie:

Like Cookie Crisp. That’s literally cookie confetti.

Tim:

That would be amazing, actually. So this celebration is going to be the best when we actually do it because that’s what we’ll do, we’ll take a box of Cookie Crisps and just throw it up in the air and then just try to catch it with our mouths and it’s going to be great and glorious!

Katie:

Yeah, so….

Tim:

I like it.

Katie:

We have a very special episode beyond the Cookie Crisp, I guess, just because it’s Episode 10 and we have an awesome guest today, Tammy Everts is going to talk to us about tons of awesome good stats but before we talk to her, let’s catch up on some of the things that have been happening lately with performance. So, Facebook has started to do 2G Tuesdays. Have you heard about that, Tim?

Tim:

I have, yeah. 2G Tuesdays. I don’t know, for some reason as you were saying that it reminded me of TGIFs or something. But yeah. 2G Tuesday. So the idea is that basically when you come into the thing on Tuesday mornings, the employees get to switch to a 2G connection for an hour which sounds terribly painful.

Katie:

Yeah, but definitely really cool that a big company like Facebook, which we’ve already talked about several episodes ago about they’re moving to…what was that thing?

Tim:

The Instant Articles thing? Yeah.

Katie:

Yeah, the thing that like I never paid attention to again but it was cool that they cared about performance!

Tim:

Right, no! So yeah, it is cool, so it is cool to see them doing; I mean obviously they’ve got a huge focus on performance inside the organization. We’re seeing a bunch of evidence of that. And it is cool. It’s very cool to have somebody like this large of an organization with this sort of an engineering, like the level of respect people have for that engineering team and have them come out and publicly say that they’re doing this, this 2G thing so that their employees can kind of actually feel what it’s like in some of these markets they’re trying to hit like India or Thailand or places like that. I mean, that’s a big deal.

Katie:

Yeah, they said they were doing it to help build that empathy for what the actual users feel using their product, which I think is a really big hurdle to overcome and I think that’s awesome that they’re taking steps toward that.

Tim:

Yeah, absolutely. I mean, so it sounds like it’s an optional thing, like the employee can choose not to maybe switch to this 2G thing, but I’m sure they’re going to be encouraging it or something, internally. I mean, they’re not just going to slap it there and have nobody use it; they’re going to find ways to make people do that, which is cool. So it’s nice and it’s not completely revolutionary: we’ve had…there was an attempt to get a Throttle Thursday thing to take off, I know, which we can link up to and I think Etsy had talked maybe even on our podcast, Lara had mentioned maybe the idea of slowing things down a little bit internally, so there’s all sorts of things where this has kind of been expressed that maybe this is something people should consider doing but it’s really awesome to see it actually happen.

Katie:

Yeah, definitely.

Tim:

And kind of in a related note, while we’re on the sort of global theme is at Velocity Amsterdam they announced that their WebPageTest has thanks to headspin.io which is this new company that’s kind of in stealth mode so it’s not entirely one hundred per cent clear what they’re doing, but they have hooked up test agents on web pages so you can now actually test your site on a real mobile network in India, for example, so you can see how CNN.com loads in Bangalore over a 3G connection and I ran a few through there and it’s absolutely, it’s terrifying; really it’s remarkable how slow it is. So that’s another really eye-opening thing, I guess, so if you’re looking at the global audience or trying to really start to feel what it’s like to load your site across different networks, both the throttling and the real testing now that they’re on web page tests are great ways to do it.

Katie:

Yeah, definitely. And speaking of Velocity Amsterdam, after I presented someone came up and asked me a question, which is a question that I’ve often got after talking about performance budgets and stuff is, people always want to know, should I have a different budget for mobile connections? Should I have different budgets for different viewports and responsive images and all of that stuff and Tim, as we’ve talked about performance budgets on the past here and you’ve fully explained those, what are your thoughts about incorporating all of these different devices into your performance budgets?

Tim:

Yeah, I think if you’ve got the bandwidth for it, it’s a cool idea. If you pick up a copy of Lara’s book and inside of it she’s got a section where she talks about what an example performance budget might be and she breaks it up like that, where she’s got a specific budget. Actually, I think her budget might be for a 3G connection versus a wifi or something like that, but she breaks it out based on the network type and she has different metrics and things like that and I’ve seen companies that have done that; a performance budget for mobile, a performance budget for desktop; probably networks, probably a better thing than say, mobile versus desktop because that’s kind of a lot blurrier. So, you can break it out that way if you want. I typically never really recommend that out the gate though. If you’re just starting with the performance budget idea, I think it’s probably better to pick one, maybe two, maybe a 3G versus a wifi or something like that at most and stick to those budgets and kind of get that incorporated into the process first, because it’s not like you can just throw twenty different budgets at it and there’s no trade-off in terms of complexity; every other budget you’re adding and then enforcing, hopefully, means that that’s one extra scenario that you have to be testing, in terms of your automation that it’s going to be running through the build process and making sure you’re adhering to the budget and one more aspect of the conversation. I worry that if you get too granular with it, it’s going to start to actually slow down your entire process and become a little bit more prohibitive, which is not the point; it’s supposed to be there to help you along the way and I think if you got a little bit too intense with it, it could maybe have a negative impact.

Katie:

Right, that’s kind of our whole focus with the podcast and everything that we talk about is just making performance as easy to access as part of your process as possible, so things can get as layered or complex as we want them to be, as detailed as we want to get about performance but your analogy of starting to exercise and can’t just sprint for miles, you’re going to get tired; those baby steps are a really important part of this whole process and definitely start small and don’t want to bite off more than you can chew. But they are useful, if that works for your team.

Tim:

Yeah, totally; if you can scale up to that, absolutely. And I still don’t exercise, by the way. It’s not gotten any better! I installed some little app that’s supposed to remind me every day and all that happens is every day, I get this little reminder on my phone that says, “Hey, you’re due for your next workout” and then I just kind of sadly swipe it away.

Katie:

I hate those really judge-y apps. I tried one of those that was just like it was really conversational; it was supposed to be like your friend and it even looked like texts and it was this judge-y text from this robot friend every day just like, “Hey, have you worked out yet?” and I’m like, “Back off, dude!”.

Tim:

Right, I mean, look: and it happens at the worst time too; it pops that up and you’re sitting there, you’ve got a pile of Oreos on your desk. I mean, this might be just me but you have a pile of Oreos on your desk as it pops up and you’re like, come on, leave me alone. No, I hear you. There is a thing with apps a little too eager to help sometimes.

Katie:

Yeah, you’re just…

Tim:

They’re quietly judging you.

Katie:

Yeah, you’re just a judge-y friend that I didn’t need. And I can delete it!

Tim:

Right, well that’s an advantage of the app versus a real life friend; it’s harder to delete. There’s no…it gets a little messier. Anyway!

Katie:

Anyway! Back to performance.

Tim:

Yeah. So, Episode 10. So, we’ve got Tammy.

Katie:

We do. Tammy is really great, super-smart; she just has a wealth of knowledge of stats and cool metrics and stuff that she will explain way better than I just did!

Tim:

Perfect. All right, so we’ll come back in just a second with Tammy.

Tim:

And we’re back with Tammy Everts from SOASTA. Tammy—can you tell us a little bit about SOASTA and what you’re doing there, I guess?

Tammy Everts:

Sure, yeah, I’d love to. SOASTA develops and sells real user monitoring, performance testing, load testing and performance analytics solutions to a lot of different markets; so there’s e-commerce and media and enterprise; basically anybody you could think of that cares about performance and monitoring their performance. And what I do at SOASTA is I kind of wear two hats: I am a Senior Researcher and Evangelist, so I have the awesome fun of going into SOASTA’s massive, massive, billions and billions of beacons-worth of real user data and slicing and dicing it and seeing how performance correlates to a ton of really cool user experience and business metrics and I get to write about it and go to lots of great conferences so it’s kind of a dream job, really. And then the other hat that I wear is Director of Content, so just helping SOASTA get the message out on pretty much any channel that we can figure out how to get our message out on.

Tim:

Yeah, that sounds pretty sweet and I know, Katie, we’re kind of interested in the performance data a little bit, huh?

Katie:

Yeah! A little bit!

Tim:

Yeah, that’s great. And you’ve been at SOASTA…I’m sorry, how long have you been SOASTA now, because you were at Radware before that, right?

Tammy:

Yeah, I was at Strangeloop, which was kind of a pioneer in the FEO space, front end optimization, developing solutions and then they were acquired by Radware which is another great company that develops FEO solutions and I made the jump over to SOASTA in the spring, so I’ve been here about six months, something like that; six or seven months.

Tim:

Nice. OK. So, we really talk a lot on the podcast about this side of things: how do companies get to a culture performance? How do they build the case for it and stuff. Which I think lines up perfectly, which is why we wanted to get you in here to talk about some of these metrics. What are some of the things that you have been seeing? I guess that’s a really broad question, but what are some of the metrics that you have been finding while you’re mining the data that could help somebody? I guess, let’s back up a sec: if somebody’s going to make the case to their organization that they need to build a culture performance, how do they start to build that case up?

Tammy:

That’s a really good question. I actually am just wrapping up at shop.org, the shop.org summit this week, and yesterday I spent the afternoon doing these things they call Doctor is In sessions where people from different companies can book one-on-one time with various people to talk about whatever their business’s pain is. So it could be SEO, could be CRO, in my case people came and talked to me about performance and that was the big question that people had; these are people who are directors of e-commerce or similar roles to that who cared about performance but just weren’t really sure how to go about even just getting a toe-in. These are for big household names sort of companies and so the advice that I was giving them was to start by using the tools that are available freely.

It’s really hard to go from doing nothing to do with performance or having very little visibility into it to just, OK, we’re going to jump in the deep end and set ourselves up with real user monitoring and do all these really sophisticated analytics, when you haven’t even done the groundwork in creating a performance culture yet, so I was recommending that people use simple, free tools like Mark Zeman’s PerfMap, for example; free Google plug-in, you know about it, and you download it for Chrome and it just gives you that instant visibility heat-map where you can look at your pages or look at anybody’s pages for that matter, and see how much latency is involved or how much bandwidth is involved in seeing the visual resources on the page.

Or I was also telling them to use WebPagetest or to trial SpeedCurve which is built on the backbone of WebPagetest… Sorry, URLs for those who need them: speedcurve.com, webpagetest.org and just plug in the URLs for their pages and begin to see…and I was also telling them, giving them some high level pointers on how to read a waterfall chart, because I think everyone should know how to read a waterfall chart, not just people in dev ops, to just get some visibility into how fast are your pages for real users? Where might the bottlenecks be? And to just begin developing a bit of a vocabulary around performance. And then I was showing them really cool features like on WebPagetest, the fact that you can see a filmstrip view and see how your pages loads frame by frame in one hundred millisecond increments, or creating videos and side-by-side videos showing how your pages load alongside your competitors, so just lots of free, easy to use tools out there that a lot of people aren’t aware of, that’s kind of my number one thing that I think people should start with. That was a really long answer!

Tim:

No, that’s great. I think that we’ve already probably…we do a list of show links for every show. I think this show link list is probably going to be our longest one already! It’s impressive.

Tammy:

Watch you toes, I’m dropping links!

Tim:

That’s great, so you basically recommend just diving head first into the freely available tools and do you think that they have to…do they have to go through some sort of process and build up some sort of base level of knowledge before they can even start to make their case for it or is it basically just pull one of them up, do some testing and immediately start trying to jump in and fix whatever you find?

Tammy:

I think…I want to say both. One of the things that I was talking about with people yesterday; people were frustrated by the fact that they would go to their…often they don’t even have performance engineering teams; just teams of developers who do everything and these are marketing and e-comm folks, but going to developers and saying, instinctively, anecdotally, that their site is slow. They know their site’s slow or they’re getting emails from users saying their site is slow and when I was talking to them, I was reminding them about, from a developer’s perspective, the web as we know it’s been around for twenty years in terms of popular use and the cry that the site is slow has been basically around for just as long and I feel like it’s just turned into white noise. If you build a site, people are always telling you it’s too slow and unless you can actually use a tool that gives you real data you could point to it and say no: here empirically I can tell you that our site is slow, if you’re just saying your site feels slow, it’s just adding to that white noise. It’s like the whining of children: you just block it out after a while, so I really recommend that people start there.

Katie:

So, how do you kind of…I know that you do a lot of human-centered kind of research and that ties into user experience and stuff. How do you talk to people about that where there’s more abstract maybe even white noise kind of conversations about performance?

Tammy:

So, it’s a good question because to me, something that kind of clouds the issue is the fact that there are so many surveys; good surveys, and I like surveys, that companies survey users and ask them, “How fast do you think a website should load?” and “When will you bounce?” and “What are you likely to do next if you bounce? Will you go to a competitor site? Will you tell your friends that the site sucked?” and I think that surveys are interesting; it’s a good barometer of what people think they want, but I like actual studies of real people where you actually see how they behave, so some of the cool stuff that I’ve been able to do in the past, for example at Radware, is actually getting people in lab settings and seeing how they actually perform when you task them with using sites on desktop, especially on mobile: I think that’s really fascinating, and seeing what people actually do versus what they say they’ll do, so just as a for-instance, last year at Velocity, I presented some research that we did at Radware which was about mobile stress where we got people actually in a lab and it was a lot of work on the technical side to do this, basically synched up their mobile devices with…oh, the word’s escaping me…oh, with EEG headsets; I’m not going to try to pronounce the long version of EEG because I only get it right about once out of every four times; I’ve embarrassed myself enough in the past.

So, we hooked them up with those and did some neat stuff with eye tracking and introducing artificial delays so we could make the delays pretty consistent for some users and then not for others, and we found that sure enough, engagement dropped and stress spiked, you could see it on a graph; it’s really, really compelling, stress spiked by up to twenty six per cent during the browsing phase of a transaction and what’s interesting about that is, and later, correlating that with, we had different settings so it’s a little apples and oranges, but looking at real user data from SOASTA where we’re looking at how real users behave during different phases of a transaction and sure enough, what we see is that people are most sensitive to performance during the browse phase of a transaction, i.e. when they’re on category or product pages, when they’re browsing: that’s when positive or negative changes to load time will actually have the most impact on user behavior. For example, we see more people bounce or we see more people convert.

Tim:

Oh, interesting. So while they’re doing the searching for it is when it’s actually most impactful?

Tammy:

Exactly. So at SOASTA we came up with something we call Conversion Impact Score and it’s a really, really complex algorithm that I won’t do justice to if I try to explain it; when I wrote a blogpost about it, I just cut and pasted one of our mathematician’s for dummies version of it and just pasted it into my blogpost, but what it does is it’s a metric that calculates, kind of what I said earlier, the impact of performance on particular pages, so, how impactful would it be if we made performance better or worse on some pages than on others, and what we found in the top four pages or groupings of pages on sites, and this is pretty consistent for every site that I calculated the Conversion Impact Score for, is it’s not the home page. Home page ranks number four or number three depending on which grouping, but by far the most important two groupings of pages are product pages and category pages and then the second pair, distantly third and fourth are search and home page. So it’s funny because yesterday, when I was at shop.org’s summit talking to people, they told me, when I asked them what pages are you focused on making faster, everybody said, oh our home page, because that’s the most important page. So it’s really great to have data that lets you know, if you’re using a real user monitoring solution and you have access to your data in concise and cool ways, you could actually calculate something like the Conversion Impact Score to let you know, well you know what? Actually maybe we shouldn’t focus so much on our home page. Maybe there’s the case we made for really focusing on product and category pages.

Katie:

How do you think that designers who might not have awesome mathematicians writing algorithms and the really fancy set-ups can test and figure out those things on their real projects, whether that’s client or internal?

Tammy:

Well, at the risk of sounding like an ad, I’m a total convert: I think that every site, if you care about user experience and you really take optimizing your pages and optimizing that experience seriously, then you should use some kind of real user monitoring solution. So, do your due diligence, go out, find a solution that works for your particular set-up because really, I read a lot of case studies; not exactly case studies, but where I scrutinized individual companies or different mini-verticals within e-commerce or media and I look at their data and I share it with people to show, this is what we learn about this company or this type of company or this group of companies. But you need to have access to your own data and you need to have, I think, access to all of it to really pick up on all the patterns that show up and anti-patterns that show up in the data.

Tim:

I really like this …I really like this idea of digging in deeper and the Conversion Impact Scores. I was just talking with somebody, with Scott Jehl the other week about the idea that we spend so much time talking about performance in terms of the home page: that’s where everybody tends to start, that’s where if you look at medians from Alexa top whatever sites, they’re always home pages, right? Because they generate the most traffic on a per page basis. But I love the idea of getting a little bit more nuanced about that because if you look at things like even a news site or anything like that, generally speaking, the article template’s probably going to be your biggest bang for your buck in terms of visits, at least, if you add them all together and it’s really fascinating to me that the performance, it reminded me a little bit of Yesenia Perez-Cruz, what she calls, I think she was calling it “journey weight”; the performance as you move from page to page and how that…we tend to think about the experience on a single page basis, but the conversion impact is viewing it more as an ebb and flow, as you’re moving throughout the site and performing different interactions, which I think is a really mature way of looking at it, actually.

Tammy:

Yeah, and I sort of cut my teeth on that way of thinking, even way back at Strangeloop where we were really focused on looking at user paths and analyzing user paths for the purposes of front end optimization because if you know where people are coming from and you know based on past behavior of previous users where they’re likely to go next; you can do some really cool stuff from an FEO perspective with regard to predictive analytics. For example, with Site Optimizer, which is now a Radware fastview, they have an analytics engine that can actually anticipate where somebody is likely to go based on the page that they’re on right now and can pre-load resources into the user’s browser based on that and make pages hugely faster so I think that looking at page flows and, I mean that’s kind of the way to go and I think that’s got to be the way forward for everybody. And one of the other great benefits of real user monitoring is that it does track those user paths for you. One of the really interesting things that I learned when I first started out at SOASTA, looking at user path data for one retailer is the fact that actually the most common user path through the site was actually people coming in on the shopping cart page and that really opened my eyes to a use case which in hindsight seems so obvious: people frequently will walk away from their shopping card because they fill it up and then they want to think about it and then they come back later and from your site’s perspective, that’s a brand new session, the session times out and it’s the start of a new session, but just realizing that that’s a user path kind of opens you up to realizing that people do experience your site in really different ways.

Tim:

So, all of this talk about the RUM data and the user path; this is all fantastic, but obviously this requires resources. If you’re going to have…there’s potentially financial resources depending on your RUM solution that you’re using; there’s time and people because you’re going to need somebody to actually do something with the RUM data, so when you’re talking about going down this path which obviously has great benefits like analyzing the user path and figuring things out from that perspective, but you have to build up the case internally within your organization that it’s worth our time to have somebody focused on this, or it’s worth our time to invest in one of these RUM products. Do you have recommendations for people on that perspective of things? They’re listening to this and they’re like, you know what? Tammy is super genius and she makes a lot of sense here. Now, how do I actually get my company to say we’re going to allocate time, energy and money towards making this a reality?

Tammy:

That’s another very good question. And again this came up a lot when I was at the shop.org summit yesterday. What I recommended to people, and I’m sure you’ve been down this path many times as well, you can add to this is, using again, one of the free online tools like WebPagetest; identify…I think it’s pretty easy to read a waterfall chart from a really high level and I wrote a blogpost about it on Web Performance Today, Radware’s blog, back in the day just called Waterfalls 101 and it talks people through how to interpret it and most sites can pluck some low-hanging fruit. There’s no perfect site out there and definitely most sites are not optimized to their fullest so, using something like WebPagetest, go figure out, OK, we’ve got these awful third party scripts that are blocking the rest of the page from rendering or images are out of control; we’ve got png files that are six megabytes in size: true story, I saw that yesterday, that could be a jpeg; find that low-hanging fruit, make the page faster and if you’re tracking the versions which you should be, if you’re a retailer; basically release all of your optimizations on the same day, benchmark that it’s faster today than it was yesterday, again using WebPagetest, grab a before and after, and then watch your analytics and see if this affects conversions, if you can, refrain from making any big changes and just track that over a week or whatever’s realistic for you and use that to help build your case. That’s just a really easy example.

From a high level perspective there’s actually one of our VP is James Urquhart wrote a post on SOASTA’s blog just this week, just our blog, you’re just going to link-drop again, is performancebeacon.com, and it’s basically just about how to bring tech and business closer and the growing role of business intelligence and there’s an article that we cited where we talked about the fact that every dollar invested in business intelligence, and that includes performance analytics, yields slightly more than ten dollar ROI, so it’s just having that confidence that getting good data that’s actionable and your RUM data, the whole point of RUM data as it’s delivered in 2015 is that it should be highly actionable, it will pay for itself. And I think also just seeking out good case studies; find other companies in your space and seeing what they do and why they’re successful. I say Etsy a lot: they’re a great company, they have a performance oriented culture, they have really aggressive performance goals because they want people to be able to flip through pages on their site as if they’re flipping through pages in a catalog. And again when I was at shop.org yesterday talking to retailers and I was talking about Etsy; people were just nodding knowingly in agreement. Everybody knows Etsy and everybody knows, oh yeah, that is what it feels like and that’s a big part of the reason why it’s such a successful e-commerce site, so I think that’s a way to start hashing together a case for performance.

Katie:

What’s been either your most favorite or most shocking data that you’ve uncovered and researched?

Tammy:

My favorite’s always the most recent, so the latest piece that I wrote on The Performance Beacon was, I found it really interesting. We were interested in tracking performance as it relates to conversions, bounce rate, in the month leading up to back to school and seeing, because that’s a big shopping month, so what we did was we chose twelve retailers who really focus on the whole back to school market in that month and we looked at their data for 2015, so the month leading up to Labor Day. But then somebody else, it wasn’t me, someone else at SOASTA had the great idea: well, let’s compare this to 2014 and see how they compare side by side, and it was really interesting.

I think the most interesting thing to me; or a couple of interesting things. One was that I read a lot about called the Performance Sweet Spot, so it’s that point when the load time that correlates with peak conversion rate, so the percentage of people visiting your site who convert in the case of retail to a paying customer, so a conversion rate for a media site might be they subscribe or download, but for retail, which these sites were, it’s the percentage of shoppers that convert to buyers, and in 2014 for desktop, the sweet spot was around three seconds, so at three seconds, that’s when they enjoy their peak conversion rate. In 2015, oh sorry, and in 2014 for mobile it was around six seconds, something like that and then in 2015 was a dramatic shift for mobiles. They went from six seconds to two seconds and from desktop from three seconds to two seconds. So to me, that was really, really interesting, just the fact that we saw that hard shift. I’m drawing a chart in the air with my fingers! You cannot see! There’s a really great line that I’m drawing on the line graph and just seeing how everything basically moved to the left, I thought was really fascinating. So that’s probably my favorite recent finding.

Katie:

That’s super-interesting.

Tim:

Yeah, it’s a heck of a change in one year.

Tammy:

Yeah, and I think part of it is to do with I think more people are converting on mobile so we saw conversion rates increase on mobile as well quite dramatically, whereas I think in 2014 a lot of people were maybe browsing on mobile but not converting so there’s some other interesting stuff going on there. And I think that’s the fun and the challenge of trying to interpret mobile data over time is just that unlike desktop, where how we behave on desktop has been pretty static for pretty much the entire time I’ve been using desktop, but now mobile behavior, we have showrooming and we have that other thing: it’s the opposite of showrooming, I forget what it’s called! And we have people, there’s a couple of stats from Google and from a company called CY where I learned that the average transaction, the average consumer transaction takes place over 6.2 visits via 2.6 different devices. So I thought that was really interesting. I don’t know what 2.6 devices looks like but apparently that’s what the average transaction takes.

Tim:

Yeah, and I think that you’re right, especially in mobile that stuff is where we’re seeing the most, I was going to say a word but I was pretty sure I was going to mis-pronounce it, so we’re going to go somewhere else! The most volatility here between from year to year and stuff is we’re still, we talk about mobile as like something that I think everybody’s fully aware of now that they need to be paying attention to, but it’s still relatively young in the grand scheme of things, so we’re still wrapping our heads around a lot of the ways it’s used and a lot of ways that we deliver for it and a lot of ways that we optimize for it; a lot of that is still up in the air and being figured out.

Tammy:

And I think also it varies so wildly from site to site. It really depends on what your audience wants. I learned from Etsy, talking to people there, that they really focus on their mobile app because for their audience, their audience really wants the app and they were more via the app, that sort of thing. So, knowing who your users are and again, this is why I point to real user data as giving you the best picture of what’s going on.

Tim:

Absolutely, because it’s going to vary.

Tammy:

Yeah, and now that RUM tools are able to catch up and capture more of that mobile data, which was a challenge in the past. And also, just a little subtle plug here, now that RUM tools are way more affordable than they used to be; it used to be that you had to be a fancy-pants company with really deep pockets to have RUM and that’s not the case any more.

Tim:

So, one of the of the other interesting take-aways I think; I read this post of yours about the performance sweet spot, or a post I guess, I’m not sure if it is the post.

Tammy:

I’ve written a few of them, so…

Tim:

Yeah, I know, it’s come up a few times. But I read something recently about it and you were talking about too that this is a potential signal for user expectation changes and the evolution there which is just one more variability that occurs when we’re talking about how mobile or desktop or whatever is shifting from year to year. Does this signal, the shifting to the left means that users are expecting faster and faster sites? They’re not becoming more tolerant of waiting>

Tammy:

No. Exactly, you’re right. So, intuitively, this is where we get into the realm of intuition and anecdotal stuff: ultimately I do feel that user surveys have, there’s a core of truth to them. Most surveys say that people expect pages to load in two seconds or less and if you look at usability studies by groups like Nielsen Norman group and these are usability studies that go back twenty years, they point to the fact that one or two seconds is the point at which we start to accumulate too much cognitive load, cognitive debt and having to wait for the next page to load and so my feeling is that we’re going to see expectations continue to evolve in that direction and as sites themselves get faster, and I do believe, the question that I get asked a lot is, are sites getting faster or slower? And my answer is, both. I think that the best sites are getting faster; I think that most of the rest of the web is websites are getting slower for a number of reasons but I think that if we keep tracking user metrics for those best sites and see those sites consistently deliver one or two second page loads, we’re going to see business metrics follow that trend, where load time and business metrics and user expectations are all going to align around one or two seconds.

Katie:

Yeah, that kind of reminds me of an interesting point that Jeff Veen made when we had him on the podcast is, we’re making the sites faster, we’re creating more space but then we just keep filling it with more stuff then so it kind of ends up slowing it down ultimately even though we’ve made the web faster generally.

Tammy:

It’s kind of like a companion piece to Wirth’s Law. We can make the web faster but we’re always going to find new junk to step in pages to make it slower. If you told somebody back in 1995 that the average web page would be over 2MB’s and contain more than a hundred resources, they would have been, but how? How can you even do that? And then their modem would have died!

Tim:

And along a similar vein, if you would have told them that the speeds of, we have 4G, 5G networks coming around and all these other different connection speeds and stuff, they would have assumed that everything would have been nearly instantaneous, which I think signals there’s a tendency to look at the next network optimization or the rising capabilities of the browser or the device getting more powerful over time is this sort of, well these things are going to fix and address these issues for us. But it’s not necessarily a case like we don’t get to come out of this scot-free. We actually have to do a little bit of work to make it happen as it turns out.

Tammy:

Exactly. I recommend people read as an eye-opener, read the Pew Internet Report and read Akamai’s quarterly State of the Internet Reports, because they actually tell you about connection speeds for non-urban centers and it’s really, really eye-opening. You realize that actually, yeah a lot of people live in New York and LA and San Francisco and other countries that have great internet, but most people don’t and internet is really, really poor; so much so that the FCC changed its definition of high speed last year to reflect the fact that under the old definition, it gave this artificially inflated sense that oh yeah, most of the country’s on high speed. But as soon as you actually had a more realistic truth-y definition, then you realize actually, most of the country doesn’t have a high speed and putting the onus on network providers to actually give people the internet that they need. I do think of the internet as a basic need now, when you have so many Government services that are being taken online and elderly people who are semi shut in or invalids, school requiring resources, so many people who are mobile only for economic reasons; they just can’t afford to buy desktops and pay for monthly internet on top of the mobile phone that they need. I really think it’s a bit of a moral requirement on site owners to actually make sure that their services are available to people without eating up all their bandwidth.

Katie:

Yeah, I’m totally with you on that.

Tim:

Amen! Do you think that we’re…so I’m just curious, following up on that. How well do you think we’re equipped as the web to deal with these sorts of situations like this? The versatility in the network, the fact that we’re going to have this rise in people coming from areas that are going to be very challenged in terms of connectivity and things like that, and I guess the reason I bring it up is just because it seems like it’s a fairly timely thing; there’s been Facebook in some articles, there’s been Google’s app, there’s been all these different things that have come out that have tried to provide some sort of shortcut to performance that almost seem to be implying that the web in its current state can’t deal with this stuff and I’m just curious for your take on that, and I know that maybe that’s a slightly leading question, but we can do that!

Tammy:

Well OK, so getting back to the beginning of your question, how equipped are we to deal with it? I think very equipped. Other countries elsewhere in the world, and again you’ll see this if you read Akamai’s State of the Internet Report, which you work at Akamai, so you’ve got that I’m sure!

Tim:

I’ve read that a few times, yeah.

Tammy:

You’ve probably got it committed to heart! But if people read it, you’d see that North American countries rank not great; other countries do really well and then I do think that sites in North America, especially e-commerce and media sites tend to be really bloated and it’s I think a weird mirror that we hold up to ourselves and we can go into this metaphorical rabbit-hold talking about that. But we’ve got to stop thinking about the web is more is more, and again, go down a little minimalist rabbit-hole there as well. So yea, I think it’s very do-able but it’s going to require people to actually take stock of the big picture: to think about internet speed and to think about who the real users are to actually take an honest, frank look at their pages and put that all together. When you actually survey all of those things, the solution seems to be a bit of a no-brainer, but counting on network improvements to be a catch-all I think is really focusing on just a tiny part of the issue.

Tim:

Awesome. Well, Tammy, I think that was fantastic. There are so many awesome, awesome things I think…

Tammy:

Oh, good!

Tim:

…that any body who’s listening…

Katie:

You do so many awesome tests and experiments and you have so many good findings to share that are like you said, the concrete data that people can refer to, be like OK, this has been a proven, actionable thing.

Tammy:

Yeah, I feel like I have the best job ever; it’s so fun!

Tim:

It doesn’t sound bad. It doesn’t sound bad at all. I know, I had to refrain myself from just being like, OK, Tammy, spend the next five minutes listing off all sorts of case study metrics, because…

Tammy:

Well, it’s in my book!

Tim:

I was going say, you’ve got a book coming out about this, so what is the book and when does that come out?

Tammy:

The book is called Time is Money: The Business Value of Web Performance. it’s being published by O’Reilly who are just great; I love them, and it’ll be coming out in January but there’s a preview edition, two chapters of the book, two of my favorite chapters actually, are available right now for free download for a somewhat limited time, I believe it’s for the next month, and the download link for that is soasta.io/timeismoneyebook and yeah, it’s filled with a lot of the stuff I’ve been writing about for the past six years and some new stuff that I haven’t written about before so it’s kind of all in there and I’m really excited to have all this stuff underneath one umbrella, finally.

Tim:

Yeah, that’s going to be awesome. Fantastic.

Katie:

Congrats!

Tammy:

Thanks.

Tim:

And then how else do anybody listening here who wants to keep up with what you’re doing follow you? Twitter, blog…

Tammy:

Yeah, both of those things. I would love it if people checked out my blog SOASTA’s blog, it’s a group effort. It’s performancebeacon.com and we publish lots of case studies, just real posts, meaty stuff. The goal of it is to just help people to understand performance and to take meaning out of it that they can then actually apply to their working lives and you can also follow me on Twitter: I’m @tameverts.

Tim:

Great. So, thank you again, Tammy, for coming on; that was fantastic.

Katie; Yeah, thank you so much.

Tammy:

Oh thanks for having me, it was so fun: I love your podcast, you guys are doing a great job.

Tim:

Oh thanks, that’s nice to say, that’s awesome.

Thank you for listening to this episode of the Path to Performance podcast. You can subscribe to the podcast through iTunes or on our site, pathtoperf.com. You can also follow along on Twitter @pathtoperf. We’d love to hear what you thought, so feel free to drop us a note on Twitter or leave a raving and overly-kind review on iTunes. We like to read those. And if you’d like to talk about being a guest or sponsoring a future episode, feel free to email us at hello@pathtoperf.com.