4 Questions with Analytics Guru Avinash Kaushik

by Jessica Lee, October 15, 2010

Avinash Kaushik is a best-selling author, co-founder of Market Motive Inc. and the analytics evangelist for Google. Jessica Lee, writer at Bruce Clay Inc., recently interviewed Avinash for the company blog and gained some insight on web analytics straight from the guru of analytics himself. In this special feature article, we asked three Bruce Clay, Inc. analysts to weigh in on his answers: Scott Fowles, Bradley Leese and Shiel Love. They give their perspectives and takeaways for our audience.

1. Do you think analytics metrics such as bounce rate and time on page are playing a more vital role in keyword selection versus just high-traffic volume?

Bounce rate is just a diagnostic metric. From a website visitor’s perspective, it reports on this phenomenon: “I came. I puked. I left.”

It is very good at quickly identifying places where one of two things are happening:

1. You are targeting the wrong keywords or promising wrong things in your ads, etc.
2. Your landing pages are broken, wrong calls to action/missing content/wrong page, etc., etc.

In that sense, bounce rate helps you start finding things to fix or stop doing very quickly. But all it does is find those places. You still have to do a lot of digging to find out what’s wrong.

Time on page is a tricky metric because if you have a standard implementation of *any* Web analytics tool, then they don’t provide you the time on page for visits with single-page views.

For more detail on this, see this post: “Standard Metrics Revisited: #4: Time on Page & Time on Site.”

So, it is not that these metrics by themselves are playing a more vital role, because while they are great diagnostic metrics (at least bounce rates), they are simply not strategic enough.

The movement I do see, amongst the smart and cool, is a movement to deeper outcome metrics in picking keywords. Moving beyond click-through rate, or clicks or visits, to conversion rate and number of micro conversions to – and this is the sexiest – profit per dollar spent for each keyword.

We are not just asking, “Will people come to my site if I advertise?” We are asking, “What will those people do for my business if I spend time and love to attract them?”

Scott Fowles: Avinash is saying here that all too often, webmasters focus entirely on diagnostic metrics as sole KPIs. Instead, it’s better to use those metrics in order to understand the impact that they have or the insight that they might give to the bottom line, real KPIs. In other words: traffic and rankings are no longer valid KPIs, however, conversion-rate impact by keyword is a valid KPI.

Bradley Leese: As Mr. Kaushik states in Question 2, part of the Googlebot measures page load time, which I interpret to mean that site speed impacts Google keyword placement (rankings.) Pretty safe to assume? If Googlebot can’t access your page or the page is slower than the competition, then the rankings suffer.

And then I ask myself, Is there a relationship between a bot and human who leaves before a page is loaded? Doesn’t that imply bounce rate might be a factor combined with the page speed component? Now, I know that Mr. Kaushik would bury me if I tried to make this comparison at a conference panel because I am making assumptions that cannot or will not be verified. But I am going to guess that Google has a metric that correlates the average bounce rate and time a human will wait for a page to load (BBR: bot bounce rate).

But there are a lot more factors to consider here. And those seemingly infinite layers of complexity make my head spin. An easy example is: Selecting low-traffic, high-conversion keywords versus high-traffic and low-conversion keyword sets. This seems like a no-brainer until you have to try and convince a client that lower traffic with “hopefully” higher conversion is a worthwhile investment. The wait and see approach is never popular no matter the state of the economy.

The complexity largely stems from human behavior over what Google deems “commercial” keyword phrases. In a nut shell, identifying the strategy to direct traffic to your site into conversion channels requires fair more evaluation than monthly keyword activities. This is scary for any SEO because the tools to identify those keyword differences are unique to every site.

There is no tool and there never will be. To maximize traffic, you need to make decisions that cannot be correlated to past successes. You need to study the Web analytics, test in real world environments, and adjust and adapt accordingly.

Bruce once stated that rankings are dead. And while that might seem like a long time off yet, Google removed the training wheel (non-commercial monthly activities.) What other surprises are we in store for? What happens when PageRank vanishes, too? Web analytics has always had the potential to offer unmatched customized feedback – are you ready to take the leap?

2. Do you think search engines monitor analytics metrics such as time on site and bounce rate for any given website and then alter SERP placement based on the results of those metrics?

No.

Bing, Google and Baidu don’t have analytics tools or data capture mechanisms sitting on your site to compute time on site, etc. to then help them use those signals. Signals that the search engines like Google are using are page load time, which is available via the Googlebot visit.

Scott Fowles: This is common sense as far as I am concerned. Of course Google has no way of knowing what the bounce rate or time on page is for your site, it can only track those metrics on their own properties.

Bradley Leese: This comment gave me a tiny bit of relief, until I read “page load time.” How is landing page “load time” and landing page “bounce rate” correlated? Seems like a close relationship . too close.

Mr. Kaushik uses the word “signals.” Could just be a clever business term like “synergy,” or could be technical lingo that brought waves of terror to this SEO analyst’s heart. No longer are we able to laser focus on subject relevancy and silo/site authority, we also have to consider page performance (which we already knew) and potential user behaviors.

What happens when the page loads fast and there is a 90 percent bounce rate? According to this response, bounce rate isn’t a metric – but might there be “signals” that Google evaluates when assigning a ranking position?

3. Now that Google Instant has officially launched, from an analytics perspective, has anything changed with the way we need to run analytics and what should we be looking for in the future?

Not much, honestly.

It is too early to tell what the definitive impact will be, but the types of things I am looking at are analyzing the data for top (“head”) keywords to see any traffic impact (almost none), looking for a diversity in the queries delivering traffic to the site (Is it still majorly what it was before or are Google’s predictive suggestions driving slightly different queries?) and segmenting clicks from results previous on Page 2 to see if Page 2 and beyond are still relevant or not (if not, then double down on SEO! : ).

From a PPC perspective, a valuable exercise is to analyze the impact on impressions and from impressions to clicks to see what might be happening in your own account (Rather than believing some of the FUD that exists out in our beautiful blogosphere!).

There is no new data that Google is passing to the Web analytics tools in the referring URL, hence all the data that you are getting before and after is the same. You should be able to do all the analysis above for SEO and PPC without having to do extra technical work on your site and do it with whatever Web analytics tool you are currently using.

Scott Fowles: More common sense if you understand how Google operates. Since Google doesn’t ever tell us how its software works up front, it means that over time the SEO community will begin to understand bits and pieces and come up with methodologies around those bits. Since this was just launched, it is too soon for community to have tackled this yet.

Bradley Leese: Oh, I wish this was true. Wouldn’t it be such a relief if Google Instant changed nothing? But unfortunately, we do see a giant difference in traffic resulting from user behavior. I have read all the self-assured experts poo-pooing that Google Instant will affect rankings. Really? What happens when users stop clicking on SERP listings? Won’t that affect your ranking position? Am I the only SEO who is seeing a “traffic” fluctuation? I doubt it.

The fact is, Google Instant did change more than speed; it changed how consumers and researchers alike “use” Google search. And Google doesn’t know exactly how this will positively or negatively affect its bottom line. Is it possible that the traffic that’s sent is more relevant and thus results in higher conversions? The jury is still out, but we will come to a conclusion soon.

4. What are the top five key performance indicators that search professionals should be most concerned with to show value of a campaign to their clients or management?

All my warm hugs and kisses are reserved for any search marketer that obsesses about outcome metrics: Revenue. Profit. Number of leads converted after 30 days. Enhanced visitor loyalty (for non-ecommerce sites). Repeat purchases in subsequent 60 days, etc., etc.

It pains me that more of our community does not obsess about outcome metrics; we are far too charmed by activity.

If you want to build a robust measurement program for search, you should focus on acquisition (click-through rate, lost impression share, etc.), site efficiency (bounce rates, cart/checkout abandonment rates, etc.) and outcome metrics (the ones mentioned above).

Having an end-to-end view is great in helping you find opportunities faster. But in the end, my love is reserved for those who 1) Understand what outcomes they are delivering for the client/business and 2) Obsess about delivering those outcomes.

It is not always easy, but if it was easy, could you charge $300 an hour doing it? : )

Scott Fowles: This last answer is gold. Focus on acquisition metrics rather than diagnostic metrics. Reaffirms what he mentioned in Question 1. Activity is meaningless if nobody converts. Hence the new slogan that can be seen in Google Reader, “Conversion is the new SEO.” Nobody wants to see how much traffic they have, they want to see how much money that traffic makes them.

Bradley Leese: Bruce Clay, Inc. is a research-intensive agency. We tend to obsess over the details. It might be a new requirement that you have to have OCD to work here (have to verify that.) And, this is where the screams of terror originate.

The answers are going to be similar in that all client sites are unique. What are the key performance indicators that are most relevant to each individual client? There is no tool that can automate this process, I am deeply sorry to say. It comes down to the nitty-gritty website, silo and page analytics to create a set of metrics that apply to all clients individually.

Shiel Love: Results. Results. Results. This was probably my favorite question/answer. In my time as an SEO analyst, I see it all the time where we get lost in improving technical metrics, especially when best practice SEO is always in the backs of our minds. What I think he is saying here is that there is no set standard of metrics to gauge success/failure.

Every site is different, and I think that SEO’s (especially with the changing climates of the search engines) should start paying more attention to what the end goal of any SEO campaign is. Any analytics package or anything that tracks “technical” success are just indicators.

It’s like when you take your car to the shop and they put it through the diagnostic machine. You come with a problem (car idles rough), and the machine tells you what the possible problems are.

Now, finding SEO solutions granted is a lot more open ended than “spark plug three is misfiring, replace it for the car to idle smoothly”, but I think the lesson here is not to forget that any tool used for SEO is nothing more than a diagnostic implement to discover elements of an issue that numbers themselves cannot portray.

For more information on analytics, follow Avinash on Twitter or check out his blog, Occam’s Razor by Avinash Kaushik. For more information or to register for SES Chicago 2010, where Avinash is a keynote speaker and Bruce Clay presents on several SEO-related topics, visit SES Chicago 2010.