Jul

19

What are the REAL web analytics tools?

Filed Under Practicing Web Analytics, Web Analytics Tools | 6 Comments

Well, my grumpy old men streak continues: not only do I not love the definition of web analytics, I am also finding myself fussing about what we consider our web analytics tools; the things we use as web analysts that help us do the job. Tools imply two things: mechanical advantage and outcomes (hammers increase your power; the outcome is two things stuck together by a nail), and the tools we choose to wield determine what level of mechanical advantage and what outcomes we should expect.

The trouble is that I think the majority of us look inside our toolboxes and we are blind to the best tools available for the job at hand. And, you guessed it, both our mechanical advantage AND our outcomes suffer, making us lose focus on the job at hand.

That job is cash flow.

Period.

So let’s take a look at the various schools of thought when it comes to web analytics tools, and peer into the toolbox through their eyes.

School 1 (Bachelor’s Degree): Tools are Data-Gathering Thingies

Yes, all web data-gathering thingies are web analytics tools. BUT not all web analytics tools are data-gathering thingies.

There is a lot of attention paid to tools that are relentlessly created and improved to get you more and more insight into what is happening out there. These tools render thousands of reports at this point, they can tell you what people cut and paste from your site, what users do in flash, what their income level is, and whether they have an innie or an outie. These tools are like flashlights. When you crawl into the attic, they allow you to find more and more things that were already there to begin with.

Mechanical advantage is how much the tool can show you (and how easily you can knit a data sweater), and the outcome is awareness/knowledge/data puke, etc.

If you relegate yourself to this school and think that these are the tools in your Batman utility belt, it’s no wonder nobody asks you for any thought. If you think the tools at your disposal are just dragnets that you sort through and clean up in Excel to satisfy requests for reports, that’s all your output will ever be, and that’s all your value will ever be.

School 2 (Master’s): Tools are Reports, Data, Analysis to bend ears and drive change!

On the second step to enlightenment, you start to see some other tools in your utility belt. These are tools that can stun your internal enemies and help turn naysayers into advocates. Recognizing reports, analysis and insight as tools (rather than work product) means that your actual work product will be INFLUENCE. At this level, the analyst starts to realize that they wield some power and can use these tools to turn heads. As the tools have more polish (more complete and focused analysis and conclusions), they provide more mechanical advantage.

So at this level, the outcome is influence, and with more polish and reputation, you gain mechanical advantage in the ratio of effort in per influence out.

If you’re at this level, you are making a contribution to the company by getting people in other departments thinking. Oftentimes, those people in other departments are thinking about how to make your death look like an accident, once the flaws in their thinking are exposed, but hey, at least they’re thinking! You are gaining value, but other people are still doing the work. When things don’t work, you get to tell the organization that a correction is needed. But you don’t make that correction, and you don’t get the credit when the correction works, because what the correction was wasn’t your idea. Identifying what does not work is not equal to identifying what does work.

School 3 (PhD): Tools are other disciplines (UX, SEM, SEO, IA), ACTION

At the super ninja badass level, the best tools in your arsenal finally come into focus. The best tool is always ACTION. This means you have shone your flashlight on an issue, you have brought the right people in and influenced them, and now you’re about to whip out your mighty hammer of DO THIS. At this level, you are the general practitioner. You have done the rotations in the other disciplines, and you can send the majority of your patients home healthy without redirecting them to a specialist. But you also know when to not cross the line and take on an open-heart surgery.

This is what makes successful web analytics (and business) people: the realization that your real tools are not the the tools used to identify the trend, and they are not the tools used to communicate that the trend is a bad one that needs to be fixed; they are the ones used to actually solve the problem and improve the situation.

At analytics nirvana, your tools produce cold, hard, beautiful Benjamins as an output, and your mechanical advantage is brought about by your ability to synthesize information and ideas: drawing on all of the competing priorities that go into a finished interface or architecture, and recommending (or testing) the ideas that balance all priorities for the betterment of the user and the bottom line. Very, very few people in your organization will be capable of balancing competing priorities (your company’s compensation plan virtually guarantees it), so solving one problem while also solving / minimizing downside to others is an incredibly valuable output that almost nobody in your company can do as well as you can. The only way that you can see the problem from everyone’s perspective, however, is to own their perspective by knowing how to do what they do.

Getting your PhD

I don’t think you can call yourself a web analytics ninja without a decent degree of competency in the following disciplines:

  • Usability
  • Information Architecture
  • SEO
  • Various web marketing/advertising tactics (PPC, display, email)
  • Social Media (I do not classify this as web marketing or advertising, and I don’t believe you should, either)
  • Design (it’s okay if you suck at it, but put some effort into it so you have credz when you start drawing out ideas)
  • Copywriting
  • Making sites (HTML, .NET, PHP, SQL, Javascript, etc.)

If you’re missing skills, don’t fret. While it can take considerable time for you to learn about all of this stuff, you’re going to love it because you’ll immediately see new action tools popping up all around you. Hopefully, your organization is one where people love to cross boundaries and teach different disciplines about their own specialties (picture a medical school). I don’t see this much, but if you’re around these types of people, consider yourself blessed.

If you don’t work at a company like this, either go to a company that does, or you’ll have to try your best to get the med school mentality: if the cardiologist faints in the middle of a procedure, the remaining people in the room have enough working knowledge to get the patient out of the situation alive. If you can’t say the same for your organization, that may be a BIG problem.

You might also like:

Are you a web analytics beagle or a performance wolf? (Lame title on purpose)

May

25

Google SSL keeps everything a secret

Filed Under Practicing Web Analytics, Web Analytics Tools | 1 Comment

Google announced a secure version of their search, located at https://www.google.com. This is great if you’re worried about the black helicopters, but for analytics, this is very not great at the moment.

As good buddy and coworker Brian Ussery pointed out yesterday, when you come from https, for security reasons, your referral data is NOT retained. This means that if people search and find your site on Google SSL, you’ll never know that they came from search — they’ll appear to be direct traffic to your site.

The exception to this rule is if they click on a paid search ad that has your campaign parameters (or auto-tagged GCLID parameter) present. They will then be recorded as paid search.

I’d recommend that in the next month or so, you take a look at what percentage of your paid search is coming from Google SSL to get a feel for whether your users are taking advantage of this new offering. If you’re getting a significant percentage of traffic from the https site, you may infer that users are also generating natural search traffic from the same site, while you’re blind to it.

A good way to go about this would be to write a little script that looks at document.referrer. If it is blank (or your own site), but the URI has GCLID or utm_* parameters present, drop a note in your custom variable to identify that visit as coming from Google SSL. After a few weeks, you’ll have a good idea of what percentage of your paid search traffic is coming from the secure site vs the folks coming from regular Google, or ReGoogle.

I am making up words now so I can take credit for them in 5 years.

Go take a look at Brian’s blog post, and the rest of his blog, which is pretty excellent.

May

6

How Does Google Analytics handle 301 and 302 Redirects?

Filed Under Practicing Web Analytics, Web Analytics and Search Engines | 7 Comments

I’ve seen this question a zillion times, so please link the hell out of this so it shows up in search results and helps people (Matt Cutts, please ignore the previous statement, I’m just trying to help people). I also posted this at the Google Analytics help forums and will do a post on Search Engine Land next week, so hopefully one of these will rank and the world will be a happier place.

ONCE AND FOR ALL! LET’S DO THIS! :)

Google Analytics will report the ORIGINAL referrer to the ORIGINAL requested page if a server-side 301 (or 302) redirect is in place.

Let’s shake out an example.

You own mySite.com and have a page called unicorns.html. You decide that you want to make a better page for this, so you make “DOUBLE-unicorns.html”.

Then you realize that you want to remove the original unicorns.html because it is lame compared to your double unicorns page, but you want to make sure that people who linked to that old page see the new, awesome double unicorns page. Because you’re a good, SEO-conscious person, you do a 301 redirect from “/unicorns.html” to “http://www.mySite.com/DOUBLE-unicorns.html”.

OK, breathe…here we go.

Let’s say that the site unicornUniverse.com loved your original page, and had a link on it. Now that your 301 redirect is present, what happens when someone clicks on that link at unicornUniverse.com? What shows up as a referrer?

Yes: unicornUniverse.com is the referrer! And what is your landing page? DOUBLE-unicorns.html is your landing page.

This happens because the browser never actually sees the first file: the server redirects your browser and the field of the DOM that monitors your referrer isn’t updated. This field is referred to by the document.referrer variable, and a quick cheat to check what’s in it is to type the following in your location bar:

javascript:alert(document.referrer);

If you’re a browser ninja, you might have firebug installed, and you can look at the DOM map there to see the same thing.

Let’s take another example:

What if someone searches for “super awesome unicorn page” on Google, just MINUTES after you put your redirect up? Well, Google hasn’t had time to find your redirect and update the URL in their index! OH MY! What will happen?

All is well in the world (almost, see the last paragraph below for what’s not well in the world), because when someone clicks the link to http://www.mySite.com/unicorns.html in the Google search results, they will be 301 redirected to http://www.mySite.com/DOUBLE-unicorns.html. Referrer? google.com (the document.referrer will actually read: http://www.google.com/search?q=super+awesome+unicorn+page — that is how it knows your search engine keywords). And again, DOUBLE-unicorns.html will be the landing page.

So, direct is direct, search is search, referred is referred.

Exceptions:

There are times when a 301 redirect will NOT preserve referrer information, but they are rare. The most common example is when an https page links through a redirect on a non-https page. Again, drop the javascript above into your browser to audit this on your own site if this is a concern. If that script tells you the right referrer information, Google Analytics will be cool as a cucumber.

Also, if you’re trying to do a 301 that isn’t of the server-side variety (I’ve seen some people try to state 301 in HTML or through a JavaScript redirect — don’t ask), you will definitely have problems.

Google’s documentation on this refers to javascript and tracking redirects, and is MISLEADING. It is technically correct, but it is NOT talking about server-side 301/302 redirects. In the isolated case it refers to (which isn’t well explained), it is right.

Lastly, webkit-based browsers (Chrome and Safari among them) have a known problem with opening content in new tabs and windows. If a user right-clicks a link and selects Open in a New Window or Open in a New Tab, referrer data is LOST! MIND BLOWN! But if they simply hold CTRL or command (on a mac) while they click to launch in a new tab, referrer data is preserved. DOUBLE BLOWN!

KEEP IN MIND:

If you are using 301 redirects, make sure they preserve your tracking parameters (utm_source, etc., at the end of your destination URL). Commonly, parameters that may specify display, email, or paid search campaigns may be stripped. What happens here? The URL of the referring site becomes a referrer and your campaign information is lost.

For example:

1) Your email that someone reads on Gmail should be counted as source:”email” medium:”email” campaign:”super email campaign”

if you forget to keep your URL parameters, instead of this visit being attributed to the email campaign, it will count as a referral from google.com! WHAT?!? Yes!

2) You are running a paid search campaign and have the keyword “unicorn site.” You expect source:”google” medium:”cpc” campaign:”generic unicorn terms”

if you forget to keep your URL parameters, instead of the paid search campaign being recorded, this visit will be attributed to NON-PAID search!!! Holy shnikeys!

The last thing to keep in mind is that before 301 destination pages replace your original pages in search engines (or if you use 302 redirects, and the URLs are not replaced in that case), your actual link URL from the engines and your Entry Page will NOT be the same! In the example above, Google linked to http://www.mySite.com/unicorns.html, but the entry page was DOUBLE-unicorns.html. If you’re trying to compare traffic to entry pages with your ranking reports (tisk tisk for running ranking reports!), they will not mesh up if you marry them by URL!

Good luck!

Apr

22

A Better Definition of Web Analytics

Filed Under About Web Analytics, Practicing Web Analytics, Web Analytics in Business | 9 Comments

In my presentation yesterday (thanks to everyone who came!), I mentioned a new definition of web analytics after seeing how lame the definition on wikipedia is:

Wikipedia: “web analytics”

Web analytics is the measurement, collection, analysis and reporting of internet data for purposes of understanding and optimizing web usage.

It’s not that this definition is wrong. It’s more or less technically correct, but it doesn’t focus on output, value, the weight of the actions associated (measurement, collection, analysis, reporting), and it’s just that it’s not marketable. For the purposes of wikipedia and the fact that the definition has to fit tools, people, an industry, and a practice, well…I can live with it there. But I was disturbed that “analysis” was only one of four verbs that fit the description, so I was hoping for an inspiring definition of “analysis”:

Wikipedia: “analysis”

Analysis is the process of breaking a complex topic or substance into smaller parts to gain a better understanding of it.

Fail. I’m equally disappointed in this one. While this may again be true from the analyst’s perspective, I don’t think it captures the output or the value of analysis at all, nor does it accurately describe the true conclusion of analysis. Yes, analysis may consist of breaking things down into smaller pieces, but that is not what helps us understand things. The understanding comes once you begin to realize the ecosystem that is in place: how these various smaller pieces interact and influence each other.

And the real output of analysis is communication in simple terms, not understanding. It’s the spreading of understanding, in our business. It’s wonderful for us (the practitioners) to analyze and understand things, but completely wasted if we’re not able to convey this understanding to others in a variety of simpler languages specific to the audience.

So, here’s my new and improved (at least according to me) definition of web analytics that we analysts should use to market ourselves to executives:

Web analytics is an unbiased discipline that actively finds and validates business opportunities by studying the habits and behavior of users, competitors, and trends in the “big picture”.

This describes what web analytics is in terms of output and value (and process), not just in terms of execution. The reason I don’t like the wikipedia definition is because it does not touch at all on context or purpose, and to me almost comes across sounding like on-demand operational overhead, rather than a proactive, value-generating process. The context/purpose of web analytics is its service to a business: the identification and validation of business opportunities (both for the web and in other areas). I completely disagree with the notion that the output of web analysis is always web-centric, so I see no reason whatsoever to say that the purpose of web analytics is optimizing web usage. The output can be thousands of things from offline advertising, to pricing, to shipping carriers, to CMS re-evaluations, to compensation plans, to organizational charts and workflow and process, and on and on. Most often, the output probably will be web-centric, but defining web analytics as web-centric makes us far less valuable than we are capable of being.

Sorry…sometimes I get a bit over-passionate :)

It also includes the analysis of competitors, which Avinash has covered in incredible depth (listen to him!), and of course a constant ear on the rail of the big picture. The “big picture” really describes the greater ecosystem of both the Internet and your large-scale business tides. Without paying attention to the fact that the housing market is tanking, that twitter has exploded as a news source (or untamed brand trashing arena) in your industry, or that the price of production at your company has increased 20%, all of this “webby” stuff we bury our faces in all day really doesn’t matter. We are accountable to context, and this “big picture” view is where stuff like @comcastcares comes from, or should come from, at least. Juicy stuff.

Finally, saying that web analytics is for the purpose of improving/optimizing web usage or specific offline changes is really only a small part of the story. As I wrote in a Search Engine Land post, “The Real Value of Web Analytics,” making your site a better, more conversion-prone web asset is a wonderful outcome of web analytics, but the problem is that most companies are completely paralyzed by operational inefficiencies and departments that don’t work as a team. The best possible output of web analytics for most companies is what happens when they watch themselves struggle to actually execute.

Just like how doctors can put a dye in your blood to see valve issues and leaks in your heart, you can watch your company execute and see the issues in your processes and teams, which can teach you how to improve your company. When you can improve operations, you can improve your web site at the speed of light, and annihilate the competition.

Give it some thought. And write some of those thoughts here in the comments!

Jul

22

Are you an analytics beagle or a performance wolf?

Filed Under Practicing Web Analytics | Leave a Comment

So, I can’t even begin to describe how lame I feel this analogy is, but it’s effective, I think.

Beagles are great hunters.  They have an incredible nose, they’re very smart, they have great endurance and they’re faster than you’d ever imagine.

Wolves are also great hunters.  Fast, silent, great instincts, and accurate.

So what’s the difference between the two?  While one is pointing and barking at the issue at hand, hoping that someone else will finish the job, the wolf is dining.

What I see these days is a lot of beagles.  Incredibly smart, diligent, and creative people who bark and point at problems, letting the company know, “Over here guys!  Come take a look!”  The beagles will hang out comfortably by the fire until they’re called to duty, asked by the master to find what the company is hunting for.  They’re given a scent and they’re off on the trail. The company pours some puppy chow into their bank account every few weeks and pats them on the head as they progress in skill.  If they do a great job, they might get a nicer mat to sleep on or a shiny new collar.  But someone else is called in to handle the problem, and those people end up getting ahead.

The wolf, however, actively hunts.  They don’t wait for a master to ask them to eat, they feel the hunger and get off their wolf asses and kill something.  They drag the kill back to the den and feed their young, training them how to kill for themselves, be independent, defend themselves.  They also choose and plan their approach, starting with nearby meals that will be easier to bring down.  When they need to bring down a large beast, they collaborate seamlessly, knowing there will be plenty of time to argue over who eats first after the beast is tackled.

So, if you’re an analytics practitioner, what do you do? If you only know how to follow a scent, learn how to kill.  Learn design.  Learn usability.  Learn HTML, PHP, SQL, etc. Learn the financial backbone of the business – the core drivers of success.  Stop talking about page views and start talking about profit.  Actively seek problems that you know in your gut and bring them down with the data you know how to retrieve better than anyone.  Stop settling for being told what to look for, where to go.  Start getting hungry, and take all of the credit that’s due to you, sharing the credit that’s due others.  Teach others how to kill.  Be a wolf.

If you’re a company struggling with crap analysts, make some good hires, pay for training, and reward performance.  Not report-producing performance, PROFIT-enhancing performance.  Dis-reward (I know that’s not a word, thank you very much) the soft performance measures of old.  The kid staying until 9:00 pm to produce a report that shows you 5 data points and 0 recommendations is on the chopping block.  The kid staying until 4:00 and taking a 2-hour lunch who gives you 1 data point to support 10 recommendations is going to the corner office, and she desperately needs a better manager who can challenge her and set the bar.

Are you a beagle or a wolf today?  Don’t be a beagle tomorrow.

Jul

9

Three enormous wastes of your web analytics time

Filed Under Practicing Web Analytics, Web Analytics in Business | 1 Comment

We are all guilty of wasting time, energy, and money. But when it comes to how organizations spend these three things on web analytics, or more appropriately mis-spend these resources, it can literally cost millions of wasted dollars when you consider how many people and days (or months) of work and senseless arguments it can drive.

There are a handful of things that we should stop doing immediately.  I’ve identified three here, most of which I’ll cover in greater detail in later posts.  I believe that if companies can reduce the amount of time they spend on these activities and increase the time they spend on practicing analytics, the world will be a significantly better place, your paychecks will increase, and you will want to hit inanimate objects less often.

  1. You are implementing analytics on new pages and tools last. I don’t know if I can tell you one example of web analytics implementation being a pervasive theme in content development.  Often, the product development people come up with an idea of what their content is going to do, they will come up with a handful of success metrics, and then once everything is built, analytics will be installed, usually requiring much of the more complex javascript – or even worse: flash – to be dragged back into the developer’s hands to open the hood and make sure this wonky tool can measure everything.

    It’s completely unacceptable in 2009 that a developer wouldn’t know the language of web analytics tools, the functions available for tracking in different ways, and your analysts (who will be the ones looking at all of this) not being a part of relevant conversations and directly involved in (and capable of) the physical implementation work.

  2. You care one iota about Unique Visitors and try to get the tool to count them correctly. Now this is certainly a topic I’ll want to cover in more detail in a future post, but let’s examine the facts:

    We know that the metric is ALWAYS wrong, across ALL tools    .  Period.  There is no fixing this metric when people delete cookies, change computers or browsers, or have more than one person in a household.

    We know that we’re not going to act on the data    , other than to tell our advertisers, partners, or someone else who asks for the metric.  And you don’t need a web analytics person for that – just tell a secretary to look up the number and be done with it. If you’re not sure of the accuracy, see above.

    None of your marketing or referral traffic efforts can be controlled to target or not target unique visitors    .  With the exception of retargeting campaigns, which can also fail if cookies are deleted of different browsers are used, Uniques is not a metric that you can optimize to, nor is it a metric that you pay against.  You are paying and optimizing to impressions, clicks, visits, or some other de-personalized metric that can’t tell the difference.

    The biggest advertisers in the world have realized that it’s not about uniques    , it’s about “touches.”  Finally, the beauty of it is that advertisers like Coke and Pepsi already realize that they’ll succeed by talking to the same people over and over again, because that’s exactly what it takes.  And nowhere else in the world are people less loyal to your brand than on the web: search has made sure of that.  If I’ve been to REI a thousand times before, it doesn’t mean I’m not going to search for my next parka or pair of hiking boots on Google to see what’s out there.  With zero competitive barriers present, don’t be foolish enough to think that you have any loyal customers who don’t need a constant reminder of your presence and offering.
  3. You are trying to get the numbers from any 2 of your 10 tools to match, be close, resemble each other, anything!!!So, I feel your pain on this one but that’s where it ends.  Let’s face it: we’re all spoiled online and we think that just because our analytics tool can give us an inch, we should be looking for a mile.  It’s just not the case.The truth is probably a few-fold: your implementation sucks (it does, trust me), many of the metrics used in each report are different in some way, and you’re completely mis-using the tools to begin with.

    A good friend of mine used to work for WebTrends and came up with a simple, but brilliant statement.  ”We’re looking at web trends, not web accounting.”  Another way of looking at it is this way: if you were in the middle of the desert and had to choose between a pedometer and a compass, which one would you choose?

    Of course, the answer is obvious.  You’d much rather know if you’re going in the right direction than how many steps you’ve taken, and that’s exactly what web analytics tools offer us: insight into where we are going, if we’re going the right direction, how we should change course, etc.  But instead of listening to that information, we’re off firing people because we have two numbers that are unexplainably 8% off.

Let me know what the barriers to stopping with these modes of business are in your workplace.  What’s keeping us locked on the pedometer when the compass is what we asked for in the first place?