Web Analytics Q & a with Avinash Kaushik & Nick Mihailovski

Avinash: Hi everyone. Welcome to the Analytics Questionar, as we're calling this thing. My name is Avinash, and this is -- Nick: Nick. Avinash: And what we're going to do today is we're going to the moderator where there'll be a ton of questions from all of you about all things analytics. And we're going to go back and forth, and we'll try and give you a bunch of ideas and answers to these questions. So Nick, let's get started. Nick: Sure. Avinash: Why don't you cover the first one. Nick: Sure. So we have Mike F.K. from Fremont. And he asks, "Why should I consider bounce rate in our average time on site as a worthwhile metric when JavaScript is not giving me how long someone interacts on a given page before they leave?" Avinash: Cool. So actually one of the important things to realize is that bounce rate simply considers the number of pages that the person has viewed on your website, and it does not take into account 'time' as a function. So, for example, you could come to the site -- to have seen the page for ten million hours or have seen it for two seconds -- there is actually no bearing on bounce rate. Bounce rate simply measures the fact that somebody's viewed one page and left. And, in as much, there's no time interaction going on there, you're absolutely right that the last page that gets viewed by somebody on your website, no balantics tool captures how long people spent on that unless they hack around it. And that's just the way it is. Bounce simply says, "Somebody came to your site and did not make a single click." And that's really what you should have in your head. And as you analyze performance, make sure that's all you're analyzing. So if you're an e-commerce website, what it means is, somebody came to your site, saw a page, didn't click any button -- not even the 'next page'. If you're nonprofit, it means nobody made a donation, because they only saw one page, and so on and so forth. Make sense, Nick? Nick: Makes sense to me. Avinash: Cool. So let me ask Nick the next question. This comes from Mike in Raleigh, North Carolina. "Because Google Analytics requires JavaScript to be enabled, is it safe to assume that all of the data it gathers comes from actual users and not from search bots?" Ooh, good one. Nick: It's a good question. Generally, most bots don't execute JavaScript. So, that means they won't execute the JavaScript tracking code. And that way, we can accurately track what people are doing on your site. Now, some bots might execute JavaScript, but there's so few that it shouldn't actually have a meaningful impact on the metrics that you have on your site. So nothing to worry. Avinash: Exactly. Exactly. And I know that -- not just the Google Analytics team, but also all other vendors, do a very good job of staying on top of these kinds of bots -- Nick: Exactly, yeah. Avinash -- and making sure your data is cleaned up for that thing. But for the most part, with JavaScript tags, you don't have to worry about search robots. Nick: Great. Avinash: So, next question. Nick: So this one comes from Timothy in Missouri. "Can you define some parameters for the industry categories in the benchmarking report? I can't find a category that suits my site. And yet clearly G.A. has assigned me to one." Avinash: [laughing] It's a good question. As all of you know that, in Google Analytics, we provide you six different benchmarks that you can use in order to figure out how your performance is like -- visits, unique visitors, time on site, bounce rate, and couple others. And Google Analytics will automatically categorize you, but it's important that -- while you cannot sort of reassign yourself to a different category -- what you can do is, at the top part of the report there is a link called 'change category.' And so, I can go in, and I can actually change my categories. For example, if I got assigned into the 'auto parts' category, but I actually write about analytics, I can actually go back and click on that link and look at all of the industry categories that are available, and then, re-categorize myself for the duration of that session to the right one. So I'll go pick "mini computers/business analytics." And then, all the data I will look at for my benchmarks will be for my category -- the one that I just ended up choosing. So don't worry about the fact that Google Analytics has got you assigned to some category. If that doesn't work for you, just go choose your own. And, over time, as more and more people use the feature, I think Google Analytics will become better and better at assigning people to the right categories. Nick: Definitely. And then, there's so many different sites and companies, it's hard to actually put everyone in a specific category. You might actually belong to two categories. So, it's good to actually look at both, and see how you compare. Avinash: Exactly. Just go ahead and change your own benchmark. Nick: Yeah, great answer. Avinash: So, we're doing this rapid fires. So let me ask Nick an interesting question from NYNJ Express. So that's a cute name by the way. "How can I identify the page that is responsible for high bounce rate on my site? What tool in Analytics can I use to put a lens through this problem?" Nick: That's a great question. Bounce rate is a really important metric. We have a report in Analytics that shows you all the top landing pages. Again, 'bounce' is when a person comes to your site and only has one page view in a session. So we look at the landing pages, and then, we could look at all the bounce rates for all those pages. There's one report -- it's on the far right-hand side -- that will compare all the different numbers to the site average. And then, on the high-bounce rate, which is bad, it'll be colored red. And the low-bounce rate, which is green, it'll be colored green. And so, you can quickly compare, looking at colors: What is doing good? What is doing bad? But the other thing, also, to look at is, What's causing those bounce rates? Typically, it's the source of traffic that's driving, you know, unqualified visitors to your site. So once you find a page that has a high bounce, you'll then want to go into the page, figure out what are the sources of traffic to the page. And for each of those sources, apply that same report to figure out which one is sending high-bounce traffic and low-bounce traffic. Avinash: Exactly. So, the report that Nick is mentioning is in the contents section on the right-hand navigation for Google Analytics. And so, click on 'content'. And then, in the list that shows up, you'll see 'top landing pages'. That's exactly the link that you want to click. And while it says top landing pages, it's not just landing pages for the campaigns. It's the top landing pages for the entire website. It is the top entry pages. And that's the report that Nick is referring to. That's the best place to start. Once you use Nick's strategy to find the one with the high-bounce rate, click on the link -- the URL -- of the page, and you drill down. And what Nick is recommending is, choose source keywords and source referring sites and that will tell you exactly why the bounce rates might be higher on that page. So, that's a really nice one. Nick: Yep. Great. Well, let's go on to the next question here. So here's a question from Brian C from Carrboro, North Carolina. Avinash: Wow, we're popular in North Carolina. Nick: Yeah. [laughter] So how do you determine which keywords you should focus your content strategy on? Avinash: Oh, this is a really good one. There isn't a short answer. If there was a short answer, we would both be very rich. [laughter] There isn't. Actually, I recommend sort of two approaches to thinking about this problem. And the first one is, actually, I go to a tool called "Google Insights for Search." And in Google Insights for Search, you can choose any category. So let's say, I just started to sell iPod clones. So that's my new business. I just cloned the iPod in Taiwan, and I'm making it for like $20 an iPod. Nick: There you go. I'll buy one. [laughter] Avinash: And so, what I can do in Insights for Search is, I can go and say, "Over the last 12 months in the category of consumer electronics, show me what keywords are the most popular that people are using to find consumer electronics." And I could even drill down by saying, "I want the category for MP-3 players." And they'll give me a list of keywords at the very bottom on the left-hand side. It'll tell you the top ten keywords being used by people to find those products. And on the right-hand side at the bottom will show you the hottest, rising keywords. And those are the two you want to start to think about. It's like, "Wow, so when people look for MP-3 players, that's really the people I want to come to my website, because I made an iPod clone. I'm going to look at those two lists of key words to figure out what my keyword strategy should be." And click on the button called 'export.' It's in the top right. And it'll give you lots and lots more keywords. In the reporting, you'll only see top ten. So that's one. Another way to think about it is, for me to worry about, What are the best possible keywords in the universe that anybody has used, that match the content on my website? So, right? And so, there's a tool called the "Google Search-based Keyword Tool," and I can just type in my URL. And when I type in my URL, you know, "iPodclones.com" -- that's my little website I'll set up. When I type it in, Google will actually tell me all of the head and the long-tail terms that are being searched on Google.com anywhere in the world that match the content I have on my website, and how much impression share I have. So by analyzing the data, you can see, "Wow, here are all the keywords that are possibly related to me. And look, I have all zero percent impression share today." So you can take that list. It'll give you the list of URLs on your site. It'll give you the keywords. Now, you know exactly which key words to optimize for which URL. And you can use that data to go about figuring out, Which content I can execute properly for my SCO strategy. What the Search-based Keyword Tool also does is, it tells you which of those you can buy using page search if you want to. They can use the data both for paid and organic. So that's a great -- so those are my two tips. Go outside your web analytics tool. Insights for Search and Search-based Keyword Tool essentially are competitive intelligence tools from Google -- both free. And you can use that. And what we'll do in our share notes. Nick's going to write on the blog post. Nick: You should put the links on the bottom. Avinash: We'll add links on the blog post so that you can find them. Nick: Perfect. Good. Avinash: So the next question is for Nick, actually. Actually, this is a very interesting question. "Will in-private browsing in IE-8, Firefox 3.5, and Google Chrome have any effect on accuracy of unique and return visitor counting? And is there anything in Google Analytics that can be done to compensate for this?" So, before Nick answers the question, let me just say that the browser, if it's set allowing users to do certain kinds of behavior, I highly recommend you try not to hack around it. I believe in respecting the user's choice, if I started an in-private session, I think we should all respect that user's choice, and not hack around it. But with that said, Nick, what's the impact of in-private browsing or Google's 'incognito mode' on data collection? Nick: Right. So let's talk about what those modes actually mean. When you enable these different features -- incognito mode or in-private browsing -- it only stores all your information for the session that you are on the browser, and once you turn it off, everything is deleted. So that way, if somebody comes back to the browser, they can't get the cookies and information from a previous time. So what typically happens is, when you set that, the server still gets all the data. You can still track all the users. And so, from analytic side, we still collect everything within the session. It's just, once that feature is turned off, the visitor information is lost. So new and returning visitors, everybody will look like a new visitor for that session. But all the session-based metrics -- the referrals, the page views, and events -- all of that will stay accurate. Avinash: Exactly. Exactly. And if I were to summarize it, I think the unique visitor count will be inflated. The percent of distribution between you and returning visitors will be affected. But all other data is collected -- your page views and visits and all of those, internal site search, everything is fine -- and you should be able to analyze it. It's important to keep track of the trends in the in-private usage. It's a very, very, very, very tiny percentage of usage at the moment. But as I said before, if that's the preference of your users not to be tracked, then we shouldn't. So with that, I'll bring to close our question hour for this week. Nick is going to post links to our new moderator page. So please go ahead and submit your questions. And once we have a handful of them, we'll answer them and post another video. So thanks very much, Nick. Nick: Yeah, great. Thanks, Avinash.