Occam’s Razor by Avinash Kaushik

I am sure you have a new year’s resolution.

Perhaps it is that you’ll maximize the value you get out of your web analytics implementation (or should I say from your implemented web analytics tool since for many tools just implementing the tool is a multi year project!).

For all Analysis Ninjas this special post outlines how to get maximum value from Google Analytics. Tips, best practices, pictures, pointers, guidance.

Most people barely scratch the surface of their web analytics tools. My hope with this post is to get you beyond the Dashboard and the Search Keyword reports and the Referring Websites data.

[While this post is uniquely focused on GA these are things that you should do with any web analytics tool, be it Omniture or WebTrends or CoreMetrics or the one your mother-in-law gave you for Christmas.

Many of these features are, IMHO, easier to use in Google Analytics but they are available in other paid analytics tools.

In fact with paid analytics it is quite likely that you are also getting additional options, add-ons, drill-anywheres, etc that should exploit a lot more to get exponentially higher ROI - if you are not call your vendor and do ask for guidance to show ROI.

Net, net this post should help everyone, that's my goal.]

This a simple ten nine point checklist of Analytics Awesomeness.

Ok that was, partly, just for fun.

Let’s go. . . .

#9: Get External Context to you Performance.

If you look at your own web analytics data you know how well you are doing. 70 miles per hour. If you have access to your competitive ecosystem then you know that while you are happy to move from 60 miles per hour to 70 miles per hour in 12 months, they have moved from 170 miles per hour to 190 miles per hour!

By enabling benchmarking in Google Analytics, you can view metrics for similar sites within your category. These benchmarks enable you to identify areas of opportunity relative to the performance of your competitors.

Benchmarks are available for Visits, Bounce Rates, Page Views, Average Time on Site, Page Views Per Visit and Percent of New Visits. Sweet!

I am loving this data because I am incessantly focused on getting New visitors, and looks like I am doing 17% better than competitors. Happy birthday to me. :)

And bonus tip, you can choose the category of your web based business.

Simply click on the Open Category List link in the Benchmarking report and drill down into the relevant available three levels.

For example: Computers & Electronics -> Internet Software -> Content Management.

Make the benchmarks unique to your own business vertical / category.

In addition to the GA data I also wanted to point out the most excellent Coremetrics benchmarks report. Very sweet data there as well, please bookmark.

Helpful Google Analytics Resources:

Helpful Occam’s Razor Articles:

#8: Internal Site Search, baby! Rock it!

My undying and eternal love for internal site search data is in one word: legendary! :) I do heart it so. And it comes in at #8.

There is perhaps only one single place in your entire web analytics data where there is no question about visitor intent. Its the queries they are typing in your website’s search engine. At all other places in your data you are guessing and inferring.

Extract every little drop of nectar from this data.

It can help you understand visitor intent and identify new opportunities to: improve landing pages, cross-promote products, and adjust your creative strategy.

Helpful Google Analytics Resources:

Helpful Occam’s Razor Articles:

#7: Search, Organic, Get Good At It.

Pure web analysts tend to focus on referrers and campaigns (DM type) most of the time. There is usually someone who does “search” and that usually means Paid search. Organic search analysis is usually an orphan (or there is someone doing that some of the time).

You want your boss to love you? Focus hard core on Organic search. It is free, it is a long term investment and can be the foundation of a great Paid search strategy.

GA allows you to wonderful analysis of your Organic search data. Torture it every which way.

Don’t know how to quantify impact of SEO to get Management support? See this article by e-nor: Leverage Google Analytics to Monetize your SEO Effort .

Don’t know how to get your organic results page ranking shown in GA? See this article on my buddy Joost de Valk’s blog: Track SEO rankings with Google Analytics .

And that’s scratching just the surface.

But if you want to do SEO Analytics than you need to get out of your Analytics comfort zone (or Omniture or CoreMetrics or ClickTracks comfort zones) and use Webmaster Tools. [Google Webmaster Tools, LiveSearch Webmaster Center ]

Think of it this way. Lots of links show up when people search for you (oh the humanity of it! :)) and your Analytics tool only shows you data for people who show up at your site. Webmaster Tools (especially the one from Google) shows you a wealth of info that you would never see in your Analytics data.

There is a ton of good stuff but without a doubt my favorite for analysis:

Where do your website links show up? For what keywords do you actually get traffic for? Awesomely actionable data.

Compare your trends over time (click on that down arrow next to locations), compare your searches from different platforms (like Mobile Search) or for different kinds of content:

It is a shame that few Web Analysts actually know of webmaster tools and fewer still use the data to help their companies rank better.

You want to stand out amongst the sea of Analysts? Get good at things like this.

Helpful Google Analytics Resources:

Helpful Occam’s Razor Articles:

#6: Landing Pages, Landing Pages, Landing Pages.

I always get asked this: “If there is one thing I can start doing in terms of Web Analytics would would it be?”

My answer is always the same, Bounce Rates for Top Landing Pages:

Thing that suck are simply standing there, naked, staring at you. I challenge you to look at this report for your site and not find broken things.

Could be campaigns, could be landing pages, could be wrong calls to action, could be…. you find that out. What this report gives you is actionable impactful starting points.

Then fix, fix, fix (and use the free Website Optimizer!). As a good friend of mine says: “You never get a second chance to make a first impression”. Cheezy? Yes. True? You betcha!

Helpful Google Analytics Resources:

Helpful Occam’s Razor Articles:

#5: Goals, Goals, Goals [& Goal Values]!
[Bonus points for crimes against the universe if you have not assigned Goal Values to your Goals.]

You: What should I measure?
Expensive Consultant: What are you solving for? That’ll be $500.

:)

If that $500 makes you think of why in the first place you created the website then wipe your tears away on that ten second consultation. It was totally worth it.

Define your goals and measure ‘em.

And here’s something worth $1,000. It will be extremely rare that you’ll have one goal. No website exists for one exclusive purpose. Think hard, identify all the goals, and measure them.

Online clicks drive awareness, customer loyalty, offline sales, and yes even online sales! Make sure to track all actions on your website that help your business.

Don’t discount the importance of tracking even things like: store locator pages, email sign-ups, maps on site, feed sign-ups, leads into your CRM pipeline, request for quotes, etc. Even sexy stuff like Loyalty and Recency (you non-ecommerce website owners, are you listening?).

While you can’t create a “Goal” for Visitor Loyalty and Recency, you should create goals (as in “the goal of our BBC website is to have 50% of the visitors visit more than 15 times a month”) and then measure success using your GA Visitor Loyalty reports.

Goal values should be assigned to each goal. For example if you submit a qualified lead on www.cat.com for a Hydraulic Excavator M322D (starting at $300k) then that through careful financial analysis you can determine that each web lead is valued at $189 (based on past conversions, offline valuations etc).

Make sure you type that into Google Analytics as you set up a goal to track lead submission. You will get “revenue analysis”, you will get “per visit goal value”, you will get “$ Index” of each page view (!) and more yummy goodness. Do this!

[The above image is from a non-ecommerce website where the goals are both online and offline. Each goal has an assigned goal value allowing the owner to precisely compute the economic value generated from the website - $14,930 in Dec.]

Helpful Google Analytics Resources:

Helpful Occam’s Razor Articles:

#4: Enable Deep Ecommerce Tracking & Analysis.

With the simple radio button click on your GA account profile and addition of a simple javascript tag to you ecommerce page you are set to reap benefits of some deep deep analysis of your ecommerce data.

Sadly not that many people do this.

Just look at the wealth of information you can get, about conversions, revenue, product sales, meta categories…. and even true “pan session” insightful analysis like Visits and Days to Purchase!

With GA the ecommerce tracking is actually quite feature rich. You can use it to track other things on your website.

For example if you are a non ecommerce lead generation website then you can capture rich transaction data for those leads and even track how many visits/days it takes someone to give you a lead.

Our good friend Justin shows you how to do exactly that here: Tracking Lead Gen Forms.

Of course I have not even touched in how much more magnificent your campaign tracking will be or how every advertising and sales channel can be held significantly more accountable now.

Helpful Google Analytics Resources:

Helpful Occam’s Razor Articles:

#3: Your Only True Analytics BFF: Bounce Rate

Even by the end of 2008 I am astounded at how few people know what the KPI (yes it is a KPI!) Bounce Rate truly means and how delightfully easy it is to understand and deploy in your war against waste.

Bounce rate is simple, easily accessible and should be used as the first point filter for “initial engagement”, “quality traffic”, “crappy pages”, “efficient campaigns” and other such key desires.

In this economy if you are not truly a BFF with bounce rates you are leaving too much of your scarce money on the table.

Helpful Google Analytics Resources:

Helpful Occam’s Razor Articles:

#2: Powerfully Leverage Custom Reports.

You are unique, yes even special :), why should your reports not be? Why be saddled with 80 reports from your Web Analytics Vendor when none of them quite contain exactly what you want?

Or what about those crabby patties in Marketing who won’t look are any report and you want to give them just a two column report with just the KPI’s that they need?

For all of those reasons and more you need to use Custom Reports.

All you need are two things: 1) Some understanding of what your users want 2) Drag and drop skills. . . .

A unique report for your unique needs.

Note that I have put in one table KPI’s and Metrics that are otherwise scattered all over the place in GA. No more hunting and pecking!

Also note the other cute thing you can do with Google Analytics. You don’t have to create a massive puke of reports. You can create a tab with just metrics for the Sales Team. Another tab in the same report with Marketing Team metrics etc.

I call it creating “micro ecosystems”. You have a fewer reports, and they are significantly more targeted. Try it, you’ll love it.

Promise me, you’ll never look at a standard report ever again.

Helpful Google Analytics Resources:

Helpful Occam’s Razor Articles:

And, drum roll please. . . . the #1 thing to do with Google Analytics is. . . .

#1: Give Me Segmentation or Give Me Death!

Were you surprised at my #1 choice? I think not.

Aggregated data rarely provides the insights you need. That’s because lots of different people come to your website for different purposes and and exhibit different kinds of visitor behavior because of different drivers and motivations.

If that were not enough you are trying to get certain goals achieved through different strategies and electric shocks.

All this makes for a potent soup and your only way to truly being a Analytics Emperor is to be really really really good at using Segmentation.

Use Google Analytics to focus on visitor persona’s you care about, measure the degree of visitor engagement, identify high value sources, magnify desired customer behavior, or simply identify the “whales” who buy tons of stuff from you.

How to? See specific actionable tips in the Occam’s Razor article linked below.

Helpful Google Analytics Resources:

Helpful Occam’s Razor Articles:

There you are.

Nine amazing things you can do to truly leverage a free web analytics tool like Google Analytics (or a paid analytics tool like WebTrends or Omniture or your favorite).

Do this and you and your company will be fine. More than fine.

If you are not doing all of these things, you are simply collecting data. You won’t be fine.

Good luck!

Now your turn.

If you had to maximize the value you get out of a web analytics tool what would be your advice?On our suggested scale, how would you rate yourself? A Prince? King? Emperor? : )

Please share your insights, critique, love (and self ratings!).

PS: Before we go I would like to thank two beautiful people for their help: Matt Parry & Rachel Meyers. You rock!

Admit it, you secretly live in the fear of your Senior Management finding out that your online greatness is less a result of your online campaigns and more a result of the tons and tons your company has invested in the real world.

The real world. “Offline” to you and me. :)

We tend to often overlook the pesky offline real world. Sooooo booorrring ! (Say that with a Paris Hilton’ish brush off. :)

I think most of it is not malicious.

For one thing it is really hard to measure. For another thing your “online” presence is probably three geeks (says the proud geek!) living in Seattle and your “offline” presence is 15,786 people in the company with power concentrated in New York and Atlanta. Hard to coordinate and get “them” to listen to us and pay attention to us.

But the world is not online or offline, it is nonline (hat tip to David Hughes for that magnificent term). One of these day everyone will get that.

Meanwhile you are the smart one, know that it is important that our “online analytics” morphs into true Multichannel Analytics, i.e. non-line analytics.

Recently we had covered how to measure offline impact of online activities.

In this post, dedicated to Sandra Dowker, we’ll cover the reverse: how to measure the online impact of offline campaigns / activities.

Then you’ll have a blue print for doing nonline analytics: Offline -> Online -> Offline .

Why should care about offline at all?

Convinced?

Sure you are.

If it is not the enticing prospect of understanding the data better then it has to be craving to be loved by your greater organization. :)

So what’s the problem here?

Like in the case of the offline impact analysis, the problem here is also one of the missing primary key (see that post please if you don’t know what this is).

Our goal is the figure out how to tie your visit to our website due to a stimulus from a offline campaign. How do I know that it was a tv ad that drove you to the site or a magazine article or a banner by the side of a road etc.

Once you have collected that piece of data you can do any of the other analysis you want for that offline to online traffic stream, be it conversion rates or site abandonment rate or task completion rates or even non-ecommerce outcomes.

Let’s get going. . . .

Tips for measuring on-line impact of the offline channel:

Here are the most common offline-to-online movement channels and how to collect data for end-to-end analysis of each:

#1 : Use redirects (vanity url’s).

The grand daddy of them all. Plaster your billboard (or magazine article or tv ad) with a easy to remember url and boom (!) you got yourself some tracking.

A magazine ad with the call to action Visit www.usequickbooks.com redirects to www.quickbooks.intuit.com/tracking_code=newsweek_dec_2008 The rest of the analysis is a piece of cake (simply segment out visits with that tracking code).

A box of 12 krispy kreme doughnuts to your IT person will ensure all redirects are coded with the correct tracking code.

If you are super cool like my former employer then you will have web based interface where the Online or Offline Marketer can attached a tracking code to any url, hit save and be in tracking heaven!

Another example:

www.dell.com/tv

(today, dec 22 here in Maldives) redirects to

http://www.dell.com/content/topics/segtopic.aspx/tv?c=us&cs=19&l=en&s=dhs&keycode=6Vc94&DGVCode=TV&dgc=TV&cid=11510&lid=985367

Note the amount of specific tracking that the folks at Dell have attached do that redirect. Bravo folks!

Contrast that to HP which has a ton of tv ads running and has a working redirect that leads to no trackable information:

www.hp.com/tv

to

http://www.shopping.hp.com/webapp/shopping/home.do

Missed opportunity.

In closing it is a crime of the highest proportions if your magazine, catalog, tv, radio, bus, billboards are not using:

That was not hard right? You are on your way to measuring online impact of your offline channel!

[I can sense the Smarty Pants amongst you snickering at the prospect of data pollution because people posting your vanity url's online. That is the reason #2 above is important. Permanent redirects (301's) will pass the referrer's data to your website. You can then split out online referrals from the offline referrals (offline will have a blank referrer).

For example the Analysis Ninja at Dell will easily be able to split out visits from my blog from clicking the links above, www.dell.com/tv, and easily exclude. Now stop snickering.]

#2 : Use unique redeemable coupons / offers codes.

Pretty much all multichannel merchants now do something similar to this. . . .

The offline mediums (a magazine in this case) carry unique promotion (or offer or config) codes that must be typed into a form on the website. This allows those visits to be tracked as being from a external “motivation”.

This works well for tv, magazines, radio, catalogs and other such mediums where it is easy for people to remember the codes.

For example the ads for www.1800flowers.com in the NYC Taxi Cab I was riding close to thanksgiving asked me to use the coupon code taxi to get $5 off a $50 order. It was easy enough code for me to remember it and use it.

www.qvc.com and www.hsn.com also tend to use these types of tracking mechanisms as one of them tools in their arsenal to track people who buy on the site while watching the tv show (or later).

Catalogs for pretty much every major or minor company, including yours I am sure, is using this exact strategy as well by providing unique coupon or offer codes.

Bonus Tip: Another great strategy is to use the same coupon code between your channels. For example you are giving $65 off the Apple Ipod Touch. Your ads / catalogs / tv campaigns can say something like “call 1800 Hot Hot Now or visit Hot Hot Now . com and use the code ipod65“.

The benefit of this is that you are providing people a choice in terms of channel preference (use the phone or the site if you want) but since the code is the same you can track it delightfully later. This helps you understand channel preference by media type (tv / catalog / radio) and also by product type (electronics / food / meds etc) and by region (what’s up with people in Florida 100% using the phone channel to order their Viagra) etc etc.

#3 : Use online surveys / market research.

On this blog we have often talked about using onexit surveys (4Q or others) to understand Primary Purpose, Task Completion Rates & Segments of Discontent (see: Three Greatest Survey Questions Ever).

You can easily adopt that methodology to ask two more questions of your website visitors:

“Which of the following were the source of your visit to our website?” [The answers can be: A tv ad, A radio spot, A google (:) search, You received a catalog etc, wordsmith your options.]

and

“What is the likelihood that as a result of a visit to our website that you’ll make a purchase?” (or sell your kidney etc) [The answers will be something like More likely, Less likely etc.]

These two simple questions, drop down single choice, will help illuminate both the drivers of visits to the website (and a real chance here that you’ll explain your very high “direct visits” number here) and also the preference in terms of the purchase channel (and of course you can use this for tech support or other non ecommerce websites as well, just twist the second question a smidgen).

In our online to offline article I had talked of the possibility of using primary market research to understand outcomes that happen offline. You can also complement surveys like the one above with primary market research to understand channel use by your customers.

#4 : Correlate traffic patterns with offline ad times / patterns.

My wife was watching HSN the other day (that is how hard it is to find something to watch on tv in the age of 900 channels!) and saw Wolfgang Puck talk about his genius kitchen knives and a “extra special deal just for you right now for only $19.99 plus shipping and handling”. (Imagine that with a Austrian accent.)

So Jennie of course went to the website instead of the phone and two minutes later (well a week actually) we were proud owners of Wolfgang Puck limited edition extra special for a limited time only with a chopping board knives (with knive covers!).

Now this is not a unusual customer behavior. Offline media stimulus causes us, lemmings like, to run to the site and do stuff.

Yet it is extremely rare that a Web Analysis Ninja sit down and overlay the company media plan on top of the traffic patterns and deduce the impact online of the offline media spend.

Why not?

Sure you have to beg, plead, and practically sleep with someone to get your hands on the comprehensive media plan for the company. But take one for the team and do what you have go.

Once you have the magazine / tv / catalog / postal mailings / radio / billboard plan for your company then do the correlations with your website traffic and see what the impact is.

My tip would be not to just look at overall traffic (or All Traffic) for the site, you may or may not detect something. Correlate it with your Direct Traffic. You might see a sudden spike in traffic there. Or with Direct and Search (organic or paid) referrals. When you do stuff on tv / radio / retail stores people search (what can I say!).

I had covered exactly this strategy in this post: Excellent Analytics Tip #12: Unsuspected Correlations Are Sweet! You can find out exactly how to execute this analysis in your company from that post.

Here is a picture of the correlations that I had shared there (for one company who ran radio campaigns and the resulting impact, not just directly from visits from the vanity url mentioned in the radio ad but also from Direct Visits, Branded Search etc, very impressive, and surprising, holistic outcome):

Please see that post for detailed explanation and guidance.

Of course it is always most optimal to identify causality as well (because correlations don’t always mean causality). The keywords you drill down when you see the search spike perfectly matching your offline campaigns for example, that has intent. Or you are perhaps running the survey I just mentioned, that will show causality. Or you sold a bunch of Puck knives exactly when, or slightly after, the tv ad ran, that will indicate causality.

Do that.

UPDATE: #5 : Use the power of controlled experiments!

Jim Novo’s comment below reminded me that I forgot to add this super awesome way to measuring multi channel impact.

I had covered it as a key strategy in my multi-channel analytics post, tip #6 for measuring impact of online to offline outcomes.

Please see that post for more details. I share stories about using newspaper inserts or conducting geographically isolated experiments in retail stores (or fast food restaurants) etc.

The strategy works for on-line to off-line, and it works even more brilliantly for off-line to on-line.

.

There you go. Four Five simple things that anyone can to to get started on their journey to identify online impact of their offline strategies.

Remember your goal is to identify the complete picture: Offline -> Online -> Offline.

Those of you who have my book, Web Analytics: An Hour A Day, will know this handy dandy reference picture from page 235… it shows how to track your nonline world efforts and capture the key pieces of data to do true multichannel analytics:

You can use this multichannel value analysis framework to plan out how you will make sure you are passing the various primary keys and forth and also use all possible techniques at your disposal.

Ok now your turn.

What are the strategies that you have use to measure online impact of your offline campaigns? Have you used any strategies above? What has worked for you?

Please share your experiences, your best practices and tips.

Thanks.

PS:
Hello from the absolutely gorgeous Maldives, more specifically the Conrad Rangali Island. It is lovely here, warm, the staff are wonderful, snorkeling is great (and I am committed enough to all of you to still squeeze in a, hopefully, valuable blog post!).

Here are some pictures. . . .

This is one of the two islands that forms the Conrad:

You take a sea plane from Mali to get there, this is one of the island resorts from the air:

A typical sunset:

Lots and lots of snorkeling, absolutely wonderful, and my one high was running into a 4.5 ft shark. It passed by my nose, I was absolutely petrified, it was not [Tweet]. Here’s a small one close to the shore:

Tiny little hermit crab, hard trying to get this close to it without scaring it:

Other than that lots of personal time (also hard to photograph):

Hope your holidays are fun wherever in the world you are.

Ok, don’t forget to share your online to offline analytics tips using the form below!