Occam's Razor by Avinash Kaushik

Here is a key difference between Reporting Squirrels and Analysis Ninjas: The latter almost exclusively leverage custom reports (powered by advanced segmentation) and the former flirt with one standard report and then another and then other and in the best case scenario pull only half of their hair out.

There is nothing particularly wrong with the standard 19,000 reports in your web analytics tool. But they do represent the Vendor's best guess about what you should look at. Sometimes they even get it right.

Most of the time though your business is absolutely unique (even as it exists amongst hundreds of competitors) and it is absolutely important that you take your web analytics tool and mold it around you. The power that is given to you even in free tools like Yahoo! Web Analytics and Google Analytics can create a view of data that will help you find faster insights.

This post is inspired by a suggestion from Horia Neagu in reply to my tweet asking for blog post ideas. My thanks to Horia.

Horia's question was: How about a post entitled "10 Google Analytics Custom the Reports You Absolutely Must Set Up"?

I am not going to write about that, simply because the very idea that a report is custom means that there are probably no "ten standard custom reports" to set up.

I am going to share one recommendation and two ideas for making your own custom reports better.

This is a "teach a person to fish" type post. Sorry. :)

No Goals, No Glory.

Here's a cliché: If you don't know where you are going, any road will take you there.

Nowhere is this more applicable than when it comes to trying to find insights from your data you can action.

You report your poor heart away, no one seems to be able to take anything you give and take action.

Often it is the case that you and I have not bothered to sit down with he HiPPO / the boss's boss and tried to understand what in the name of all that is holy and pure is our website trying to do!

What are the goals?

No custom report (or advanced segment, the life giving oxygen) was ever created without an answer to that question.

So ask that question. Get an answer before you go about your customization ways.

If your leaders / clients truly want wisdom from you they will answer the question. But it does happen sometimes that begging or throwing yourself at her/him does not elicit anything of value.

In those rarest of rare cases (after you have already submitted your resume to other companies that will cherish you for the golden child you are) try to figure these one or more of these three things out:

1. Where it the company currently spending money? Email marketing? Affiliate? Paid Search? Online PR?

And what's the biggest bucket?

Now go create your custom reports because if you can help the HiPPO's figure out how to reduce cost of acquisition they will love you more than you can imagine.

2. If possible, without violating HR policies, figure out what your boss's salary bonus is tied to.

Start doing analysis that will help your boss get a raise. A great goal to have, love and promotions likely.

3. Go visit your website, yes yes the one you have not used for a while. : ) Find out the single worst thing about it (should take you less than half hour of clicking around).

Now go look for data that will help you prove that the worst thing is the worst thing. Not a bad goal to have to fix what's completely broken, and people will listen.

Three good proxies if you have no goals to start with. Ideally you'll know what your Macro Conversion is so you'll start your analysis with a bang. Super ideal would be that you know both your macro and micro conversions!

Remember: No goals, no glory. Not for you. Not for your boss. Not for your company.

Custom Reporting Tip #1: Always, Always, Always Focus On The End To End.

One problem with standard web analytics reports is that the data you need is scattered all over the place, making it harder for you to find insights.

For example I am trying to figure out which pages stink and need fixing. In Yahoo! Web Analytics the standard report only shows Page Views and Average Time on Page. How much good will that do?

Or I want to figure out which sources of traffic I should make love to or divorce? The standard Google Analytics report spreads the data I need over four tabs.

Custom reports are good at solving this problem. Drag the dimension you need (traffic sources, landing pages etc) and analyze the data by choosing the metrics that tell the end to end story.

End to end has three pieces: Input. Onsite Activity. Outcome.

Here is my favorite, custom, traffic sources report:

By inputs I mean metrics that help you understand (in line with your goals) how well the "top of the funnel" (usually acquisition) is working.

In my case that is measuring Visits (to know who is sending how much) and % of New Visits (to know who is sending how much that is of value to me – new visitors are very valuable in this case).

At a glance I have the information to start making some preliminary superficial judgments about performance.

By onsite activity I mean choosing metrics that help you understand the behavior of your visitors on your website (thus absolving your Acquisition team of any blame, perhaps!).

In my case that is measuring Bounce Rate (not so fast Acquisition team, don't get me bad traffic! :)) and Average Time on Site (as a proxy of measuring if the landing pages are engaging visitors and as a proxy of how much each traffic bucket engages with the site).

Depending on my goals I would choose different onsite activity metrics for my custom report.

By outcomes I mean, well you don't need to know do you? You read this blog! I am all about outcomes, every day!!

In my case the outcome metrics are Goal 1 (my macro conversion) and Average Value created for my website.

I could also have used $ Index or Per Visit Goal Value metrics if I were analyzing a non-ecommerce / content only website.

Remember Without a crisp articulation of outcomes every battle you fight will be lost, every day and you will live a very very unhappy life.

With these end to end metrics my custom report tells me stories that would otherwise take too long to piece together (or stories I might have missed completely).

One of the stunning realizations was just valuing Twitter traffic for example. (Click on the above report for a higher resolution report).

My twitter (social media) campaigns were doing exceptionally well. Lots of traffic (#3) overall, the second highest conversion rate (0.78%) and a Average Value that was not the best but rather sweet ($136 – which looks ever better when you compare the cost which is negligible).

Yet non focused traffic from twitter is not doing that well. 0.33% conversion and $39 average value. Pathetic.

I can now jump, like a na'vi, from row to row understand performance quickly and efficiently.

The power for a custom report that shows the end to end story.

It is so easy too.

For example here is the exact same custom report created in Yahoo! Web Analytics, just 30 seconds of drag and drop:

[Click on the image for a higher resolution version.]

Always go e2e. If you don't, you better have a good excuse!

Custom Reporting Tip #2: Create "Micro Eco-Systems".

I think I can honestly say that I have get to meet a single decision maker or a department or a company that has yet to tell me: "You know what the problem is Avinash? I don't get enough reports."

:)

We love spewing out data and pretty soon your company has 200 reports and I'll bet you $50 that not a single decision is actually based on data.

So fix it.

Create micro eco-systems.

What I mean are custom reports that do three things:

Let's understand how to do this by looking at a real life example.

My goal is to create a "search ecosystem" report that collects different important pieces of data, for three different stakeholders, all into one place.

I do that by first understanding who all the stake holders are who'll need to use the data (let's hope!) and doing a simple stake holder interview to understand what their business goals are.

Now rather than spamming everyone with a report (that no one will find, I'll have a hard time version controlling, and other such pain), I'll just put it all together in one place (at least in Google Analytics due to a simple yet exceptional feature – tabs!!).

Here is a pictorial view of the process that I'll go through:

[Click on the image for a higher resolution version.]

My first "client" is the Acquisition team, they are responsible for spending the company's money wisely. They are measured on bringing new Visitors (potential customers) to the site.

I create a tab for them that shows Visits, New Visits, Bounce Rate and Average Time on Page (not site). I add the latter two because I want them to see their end to end view and I want them to realize they hold some level of responsibility for people not just coming, but also staying.

We have just one report, each day (God willing) they'll log in and see their own personalized sweet view of the data:

[Click on the image for a higher resolution version.]

But I am not done yet.

Next up is my HiPPO. Let's call him Paul.

Paul only cares about Revenue and all things connected to revenue. He does not care about any other metric. Nothing wrong with that.

Rather than creating another report for Paul I click on *Add Tab* simply do this:

Create a person view for Paul. I throw in Visits (I have to give him some context and some Input metric) and Goal Conversion Rate (so he knows efficiency), Goal Value, Revenue, Shipping (because Paul is having us charge lots for shipping because he thinks of it as a profit center (!!), not great but remember I am personalizing).

Here's the resulting output:

[Click on the image for a higher resolution version.]

You know what the result is?

Paul actually looks at the data every other day (and a bit deeper each Friday) because it does not contain crap. It contains just what he needs to do his job (find people to reward and find people to fire).

That is what you are going for. Taking people from data apathy to data usage.

[Oh yes, yes, I noticed Revenue and Shipping are zero in the above screenshot. I wish I could show you someone's real data! Not today. But you get what the report is trying to do.]

Finally there's Amy. Another key stake holder, but a tougher nut to crack. You see her bonus is tied only to Visits, a low bar if there ever was one.

So what do you do?

You can't over smart Amy, she is too smart for you (and probably a level or two higher).

You are going to lose her if you give her too much data.

You need to entice her to start using data, and restrain your smarts – you know you want to create a impressive 8 column report!

In this case I simply add a tab. It says Amy Chang (so she knows it is her very own personal report). It has Visits and Average Time on Site. I added Time on Site as a Outcome metric, just to keep up with my outcomes obsession.

[Click on the image for a higher resolution version.]

It is simple. It is effective.

It will get her to see just the data she wants (plus one more thing :)).

And here's the sweet part…. since this his an eco-system report perhaps she (and Paul as well) might see other pieces of data in other tabs and might be intrigued enough to ask you to add more metrics.

Then and only then and only at that time and only when you are asked (am I repeating myself?) add those metrics. It is vastly more likely that she (and Paul) will use the data.

There you go… one report that has all the data each stake holder needs personalized and customized.

No one is going to come to you and say: "hey want folder is my search report" or "I don't understand what all this data is saying" – it is personalized. And when you have to make changes, it is all in one place.

Well…

Such simple little things: tabs (or cup cakes :)).

Makes it so much easier for you to create a data democracy. And a bit sad that you can't do this with most paid web analytics tools today – yes you can create a custom report but the above report would be a one huge 13 column report that:

I hope all paid web analytics vendors will incorporate this feature, for the sake for our data democracy!

That's the story. Goals. End to End. Micro self contained eco-systems.

I was hoping to teach you how to fish, rather than just tell you which 10 reports to create. Regardless of the specifics of the reports and metrics above I hope you have learned a bit more as to how to think about approaching the issue and the important things to focus on.

Custom reports are a powerful way to take what looks overwhelming in web analtyics – REPORTS and DATA – and make it didapper. It is also a wonderful way to start the journey of your company, big or small, to start using data.

Ok now your turn.

Do you have custom reporting tips to share with us? What small or big thing you have done that really really worked for you? Have you tried end to end reports? How about micro eco-systems? What strategy completely failed? Got a custom report you think everyone in the world should be using?

Please share your stories / tips / bruises / successes.

I'll send the best one a copy of my new book Web Analytics 2.0.

UPDATE: It was hard to pick just one winner so a copy of the book goes out to SteveK (for advocating common sense!) and to Ali Shah (for emphasizing sharing of context). A bonus prize also goes to MGSeeley (for bringing a smile with his adorable analytics haiku!).

How do you measure success of a online webinar?

I recently did a webinar for the Search Engine Strategies conference (I am doing the opening conference keynote at SES London and SES New York) and my Market Motive co-faculty member Greg Jarboe sent me this KPI via email:

"Your webcast was a big success. Your KPI questions per attendee was off the chart!"

I don't know why I had not thought of this wonderful KPI. So much better than # of attendees.

As always though context is king.

It could be a good thing ("you were great, engaged the audience") or a not such a good thing ("no one understood a thing you were saying, hence so many questions"). Only upon reading the actual questions could I figure out which case it was (mercifully case #1 for me!).

End of a minor web analytics lesson on going beyond obvious metrics and never, ever, never forgetting context.

Back to our story. . . an hour is too short a time to answer all the questions (even in a webinar just focused on attendee questions). So here is a small selection from the 80 questions I could not answer in the wide ranging webinar.

We will cover measuring success of SEO efforts on one web page, how to do search engine optimization for b2b websites, how to rank for highly saturated industries / categories / keywords, and which competitive intelligence tools do I use for search program optimization (and targeting display ads using search data!).

I hope you all find the answers to be of value.

#1. How do you measure SEO performance on a page level? I'd like to know how well my seo efforts for a particular pages have performed.

Every measurement question should start by taking one step back and thinking of goals.

In this case here are some obvious ones:

Now it is not hard to figure out how to measure performance! [Before you do any kind of measurement please consider going through the above exercise. It is simple, effective and works like a charm - not to mention allows to get going faster.]

Before you analyze do one small thing. Log into the Advanced Segmentation tool in your web analytics tool. Create a segment for Organic Search traffic. Sources -> Contain -> Google, Bing, Yahoo! etc. Save. Another way to cheat at this is to simply use Medium Matches Exactly Organic.

If your web analytics tool requires you to call the vendor to set up advanced segments or re-tag your site to get segments then switch. There are too many choices in the market.

Now log into whatever web analytics tool you use and drill down to the specific page you are interested in ("Top Pages Report" / "Content Title Report" etc). Apply the Organic Search segment to that report (in Google Analytics segments are on the top right, in other tools please refer to user manual).

More traffic, not that hard. Stretch the time period to six months (or some large date range – remember SEO takes time). What do you see? Nice and gradual up and to the right trends. Do your happy dance! Something's working. Now look down at the table under the graph that shows traffic sources. If you did your segment correctly you'll see just the search engines and how much each is contributing to your overall traffic. Does the distribution match your goals?

Ready for the next step? Click on Referring Keywords and now you are looking just at the keywords bringing traffic to this page. Do the keywords match the intent of the page? Do they contain keywords you were specifically targeting? No? Why not? On the other hand what are the surprises? Is the customer intent contained in the keywords telling you how to change / improve the page? Do it!

Indexing. . . I am a big fan of Google's Webmaster Tools because of the wealth of data available, if you are not using this free resource (no matter if you are a SEO or not). Bing's Webmaster Tools have also evolved a ton, please claim your account right away and dive in. [I have not had much fun with Yahoo!'s web master resources.] In either tool you are looking for how well your site is indexed (report: Your site on the web -> Top search queries -> Impressions), how well your pages are indexed and, my absolute favorite, which keywords your search results are showing up. You are checking to see if:

1. the pages you are targeting are being indexed frequently and

2. if your site is showing up for the keywords you were targeting.

You want validation that you are showing up for the set of keywords you are optimizing for (above) and that your pages are being recorded as being optimized for the right keywords (above the above :).

Success. . . I humbly believe that the biggest mistake most of us doing SEO make is that we are far too obsessed with ranking and meta this and that and how to work back algorithms etc etc. We should focus more on what was the business impact of our SEO efforts.

So in this context go back to your page report (from step 1 where you applied the organic segment) and look at the $Index [which is: (goal value + e-commerce revenue) / unique views of the page you are analyzing]. That is a crude measure how how efficient your page is being at converting. Of course look at our favorite metric bounce rate by keyword (that tells you if you can get people to give you one solitary click, the most primitive measure of SEO success).

If you truly want to kick it up a notch as a SEO please please please go to the Goal and Ecommerce / Conversions reports and apply your organic segment, stretch the time period, and report (aggressively) how well your SEO efforts are delivering value to the business.

Do it at a overall level, do it by country, do it by search engine, do it by specific keywords you were targeting. . . . and take two minutes to straighten your clothes because a new level of love and praise are about to be dumped on you by your company / client!

[Does the above seem like a lot of work even if it is straight forward? It is. I know we look for short cuts. There is no such thing in real life. But if you are willing to put in a little bit of sweat equity then you'll stand miles apart from your SEO competitors. Not a bad trade off, right?]

.

#2. Is there a fundamental difference in SEO strategies for business-to-business sites vs consumer focused ones?

[It is worth pointing out I am not a hard core SEO, that would be Todd Malicoat, I just play one one TV! Think of below as my personal lessons from the front-line of doing this work to the extent my humble skills allow.]

The basic techniques you use to do search engine optimization between b2b and b2c do not change all that much.

Ok All that is the same, no matter if you are a b2b, b2c, b2a (business to aliens, yes they do exist!). Do all that first to make sure you are not coming to play the super bowl naked.

Here are a few things that are different with b2b. . . . .

* Some very effective SEO strategies like allowing users to add reviews and comments and extend the scope of the page do not work as well with b2b as it is a differ net type of engagement and experience with your customers. Well don't give up. You have many many white papers, though leadership papers, webinars, Big B2B Association publications where you contributed and more locked up in pdf or, much worse, behind a forced "give me your login" / "create a account" page. I am going to give you a false email, why not just give me the content, AND let the search engine index it efficiently after all you want people to consume the content.

Did I say already content is king?

* One of the most common issues with b2b websites is that they often have a very specific understanding of their space when it comes to how their potential customers search for information. This results in not speaking the same language (say keywords) as their customers. When I work with b2b websites I spend a lot of time in the AdWords Keywords Tool, Insights for Search, Compete etc analyzing keywords and search behavior in my category. This knowledge goes back into re-doing content, urls etc.

This is of course a good method for b2c as well, but it is significantly more important for b2b.

* Start a conversation. There will likely be a lot fewer individuals talking about you / your industry, a lot fewer tweeting and expressing their love (or hate). I get it. But conversation on your site and away from your site is key (obvious fact). Why not host a user forum on your website for current and future customers to come together and share their thoughts / ideas / complaints / rave about your competitors (scared?)? Why not seek out the few people who do talk about the industry on twitter and engage with them? Why not start a YouTube channel with a series of how-to videos? Why not, : ), start a blog? Not just to highlight your own pomposity and press releases but to really share and lift your industry (not just your company)? Why not become the destination for industry professional?

So few people in the b2b space bother to start conversations, why not use that to your advantage? Even if you can hook 100 people is that not more than worth it?

Three small things that I would prioritize higher when I work with b2b sites.

What do you do differently when it comes to your b2b clients?

.

#3. When trying to help your rank in search engines. . . when you are in a saturated industry like health or travel insurance – how does the approach change or differ?

Two words: Long Tail!

When you say saturated most people mean that for the "top" keywords they are interested in there is too much competition. For example: "hotels in las vegas", "cheap health insurance" etc.

When there are a lot of players in the field it can be difficult to show up for the "head terms", especially if there are some strong players in the field. In these cases I have had a very positive experience focusing not on the head terms (terms for which there is a lot of traffic) but rather focusing on the long tail (usually key phrases that individually have little traffic but collectively these key phrases can deliver a ton of traffic).

So, if relevant for your business, try to rank for "california health insurance plans" or "california individual health plans" etc. Key phrases (not just words) that each have much less competition (and will likely deliver more relevant audiences).

You can use various keyword tools out there to identify these key phrases and then adapt your SEO strategy (pages, content, urls, etc) to focus on them. One way I use is to just type in competitor urls into AdWords Keyword Tool and then research what is working for them and adapt my strategy.

Targeting the long tail with SEO can be a bunch of work, hence I have recommended in the past that one effective and cheap way is to use paid search to monetize the long tail. But I can tell you from experience that it works. For example for this blog the top 10 (head) keywords bring in something like 5k visits and the long tail (around 25k keywords) bring close to 34k visits. All organic (I am not rich enough to afford paid search!).

One more bonus tip: Leverage "universal search".

Videos, pictures, downloads, offers, buttons, maps, uploaded menus, coupons, and on and on and on.

When you search for many terms relevant to me you'll see videos pop up, my book (uploaded into Google book search) show up with preview thumbnails, some of my flickr images and my twitter account and so on and so forth. For many of these searches I don't rank #1. But man do those listings (when triggered by the search engine's algorithms) stand out and grab the Searcher's attention. Often for competitor or big paid web analytics tool queries where I have a snowball's chance in a hot place of standing out.

It is ironic that most big companies (with so many assets to leverage) are pretty bad at this. So you win! :)

Also Google (I work there) Local Business Center is really good: http://www.google.com/local/add If you are a small business then this is one more important arrow to have in your quiver!

.

#4. Can you look at your competitors sites in the analysis tools you have discussed?

Yes.

But first. . . . it is important to realize that you need to have two skills before you look at competitive intelligence tools:

If you meet the above two requirements. . . . .

I love using competitive intelligence tools because they give me a perspective and context that is simply missing from Omniture or WebTrends or CoreMetrics or the clickstream tools.

In the search context here are some of my favorite tools and what I use them for.

Insights for Search:

I adore I4S because it is perhaps the most comprehensive "database of intentions" thanks to providing us all with access to worldwide Google organic search data.

Use it to understand the latest trends in your category. For example: "How is interest in the computer security category (All Categories -> Computers & Electronics -> Computer Security) and what are the top 100 search terms and the fastest rising brand names / products / searches in that category?"

Use it to identify opportunities. "What states do people search for credit cards the most? What states do people search for Visa credit cards?" Oh look the states with really high credit card searches don't have really high visa card searches, maybe we should do some offline advertising!

Use it to time your campaigns. "When should I have started SEO and PPC campaigns for Italy Tours 2010?" In April 2009!! That's when people first started looking for them. Now go plan for 2011.

Helpful article: How to use Google Insights for Search.

Ad Planner:

This wonderful tool is really built to help you do better display advertising. You log in and you have the delightful ability to do demographic (male, female, age, education, income etc) and psychographic (baby boomers, extreme sports fan, household decision makers, luxury goods consumers, moms etc) segmentation. You can hone in precisely which websites most likely contain your desired audiences. Show them relevant ads and get clicks!

But in the search context there are two things that you use this tool for.

Type in any website you want, expedia.com in my case, and checkout the site and search affinity data:

[If you don't see the image above, turn off your ad blocker.]

"The affinity score estimates how many times more likely you are to reach an audience who visits a specific site or searches for specific keywords versus an audience on the internet overall." Source.

Sweet 'eh?

Second, click on the tab that says Search by Audience and then the Keywords Searched button and now you have an ability to use search behavior to identify audience pools.

To use the examples of my beloved Indianapolis Colts (go Colts!!!). . . . I have an ability to type in a bunch of related keywords (the tool suggests most used ones) and find out which websites are most likely to be visited by people who search for these keywords:

[If you don't see the image above, turn off your ad blocker.]

At the top are keywords I typed. On the bottom are most commonly searched keywords, I can choose these if I want.

I hit ok and then sort by Comp Index, to ensure I sort the data by the highest audience concentration (audience that searches for all things Colts in this case).

I can use this search and web data to identify where audience I am most interested in exists. I can use it to find out the keyword data for those sites. I can use this to identify sizes (visitors, page views etc) of those websites.

Nice right? Actionable too!

Helpful article: How to use Google Ad Planner.

Compete.Com:

Compete is a paid tool (and it only contains US data). I really love using it because of the wealth of search data it can provide, at an affordable prices.

[I have had a complimentary Pro account for the longest time thanks to the nice people from Compete, that might bias my opinion. Other than that I have no other affiliation with Compete.]

In context of Search I use the data for. . .

1. Identifying what are the top referring keywords for any site that I am interested in:

Above data for www.clickequations.com (the paid search analytics company I am on the advisory board of). Of course when you log in with a paid account you would see rest of the data like paid and natural search split for each keyword and time and what not.

Craig will not be happy that he ranks only #12 on the keyword list! :)

I can either use this data to go after keywords that are not currently referring traffic to ClickEquations (more for me!!) or I now know what keywords I need to target to take ClickEquations down in my quest for world domination! Ha!

See how focused you can be with data?

2. Identifying share of search for a keyword:

In this case I would like to own the pear fruit market, though at the moment I only own two trees. So I go into Compete to find who my current competition is (above exact match data for query "pears"). I can get lots of details about volume, paid and organic share, what percent of traffic comes to a site from that keyword, etc etc.

Now that I have a benchmark I can go about my super awesome kick butt SEO efforts and one way I know I am winning is to check this report in a month or two (or three weeks after whenever I think I am done). If I show up here I know I am having a impact.

These are just three of the many tools I use. There are a whole lot out there that sometimes give you similar data to the above three, or often give you a lot more. Just remember that there is a lot you can learn from what is going on in your ecosystem and at your competitors.

Ok now your turn.

Got a couple tips you want to share with us about how best to do SEO for B2B sites? How would you measure success of SEO efforts spent on a page on your website? Would you use any of the four ideas I have suggested? Care to comment on how to do SEO for crowded industries or for keyword categories where one or two players seem to dominate? What is your favorite search competitive intelligence tool?

Please share your tips / best practices / comments / critique.

Thank you.