Category Archives: Web Analytics

‘Measurement, analysis and learning’ key bottleneck for marketers

In Unica’s recently published ‘State of Marketing 2011’ study, measurement and analytics was identified as the top bottleneck for the 279 marketers polled:

This is the first year that this has registered as the top pain point. I’d suggest one reason for this, could be the maturity of web marketing and the emphasis this discipline puts on measurement.

Interestingly, the survey finds that the key issue marketers face is turning data into actions. There’s a problem with obtaining data, and there’s a problem with converting that data into valuable outcomes.

This relates to an earlier IBM study which showed that businesses who can act on business insights generally perform better. Looks like many marketers realize this but are struggling to turn this into reality themselves.

Peruse the report in its entirety:

A wonderful use of the persistent URL: Unica Netinsight

image We currently use Unica NetInsight as our web analytics tool of choice on IBM.com. One of it’s advantages is that it’s highly configurable: on practically any report you can go in and add filters to hone in on the data you want and add metrics/dimensions to expand out on the information you get back (say if you want to add Visits as well as Views to a report).

All wonderful stuff.

But what happens if you want to pass that information around? One neat feature of the tool is that every report you run has a unique ID in the URL. So you can send someone the URL of a page you’re looking at and be confident that they will see exactly the same report as the one you are viewing. If they change any of the values (say run the same report but roll back one year prior), once they rerun the report, they get a fresh URL. The structure is setup like this:

https://xxx/xxx.cgi?base=getreport&id=29877

Every time you change any value and rerun the report, that ID on the end of the URL increments. The server does the work to map that ID to all the report variables as the page is being generated.

How does this differ from other tools?

Many other tools out there (such as Google Analytics and Omniture, if my memory serve me correctly) may use the URL to specify what kind of report you are running, eg. Keyword report versus Top Pages report, but other key pieces, such as which data profile you are viewing, are stored through other means such as cookies so what one browser sees will differ from the next. You can’t be exactly sure that the URL will generate the same report for everyone.

It would be wonderful to see other applications in the business intelligence and analytics space follow this example. It may require slightly more coding on the backend to map unique IDs to reports, but from a user perspective it’s great to have the sense of security that when you pass around the URL, you can be confident everyone sees the same thing and that you can record or bookmark the URL and know when you reload it a year from now, you’ll be looking at the same report you have in front of you today.

It’s fine to plot the interest graph, but what happens next? (Social Media Week panel)

In a panel discussion today on social listening at the swanky new SF PeopleBrowsr office, the interest graph formed the basis of a lot of the discussion. I guess I’m out of touch with social media monitoring as this concept was new to me. First we had the social graph, of which I’m aware: a mapping of all your connections (say friends and family) to whom you are connected across social networks. Now with some degree of overlap, you can also plot an interest graph: this time mapping connections based on a shared interest. Susan Etlinger of Altimeter used the example of a fashion site where people build connections based on couture. You may not share these interests with your grandma, but only a small subset of your friends, and the extended network of aesthetes you meet on the fashion site.

Jodee Rich from PeopleBrowsr suggests these interest networks are of more value to businesses as it gives a truer value of an individual’s importance to them. Businesses will get more value by targeting their communications around those people who have authority in that interest area (interest graph). Context is everything. You only have authority in relation to an interest (or theme). Having 500K followers on Twitter means nothing unless those followers share the common interest which is of value to the business tracking you.

This got me thinking where my own social presence and my social and interest graphs lie. By day I work in the technology sector and I generally share with people with this interest (from within my company or external folk) on Twitter. This is where I geek-out. Now I do have the other side to my online communication: where I share pictures of my newborn, other interests like music and art and bizarre oddities I find on the web. This extra-curricula activity all happens on Facebook. And rarely do the twain meet. I know not everyone divides up their online existence to this extreme, but many will have some degree of division and in these cases businesses need to ensure that they have tools that can map across the different networks in use.

When it came to what businesses should do with all this listening intelligence they build, I felt that there were more questions than answers. Tim O’Reilly proffered that sophisticated companies will go beyond business intelligence and use social listening to shape business processes. Effectively molding products and services around what the audience says it wants. However, he also suggested that this ‘autonomic’ model of business should have some human component if I understand rightly what he later said about ‘humans going the last mile’. Computers can only go so far before some level of human intervention is required to make sense of the data and take appropriate action. I’m uncertain as to at what point human intervention really makes sense and I know this is a hot topic of debate in decision management science.

O’Reilly also states that ‘great companies have everybody listening’. Listening isn’t just the domain of marketing or comms departments, but everyone can get involved and use this input from the market to drive the company forward.

I can see a flaw in this plan: the tooling.

I have problems enabling anyone to listen who doesn’t have social media responsibilities written into some part of their function. Even if I can get them access to a social media monitoring dashboard, they’ll be looking at the predefined generic terms determined by the marketing/comms team that setup the tool. This won’t include the terms that a local office would need to monitor the conversation relevant to them. So I inevitably end up pointing them to personal social media tools like Tweetdeck, which lacking any kind of workflow, offers no scope for coordinating conversations.

Brian Solis deserves a shout-out for doing a wonderful job of guiding the conversation and even working in a ‘sexy’ Marvin Gaye reference.

Optimize SEM and SEO Lead Gen Campaigns with Web Analytics (Webinar)

Integrating search engine marketing (SEM) and search engine optimization (SEO) projects and teams is a best practice that can deliver a powerful Virtuous Cycle.  Built on the foundation of an analytics platform such as Coremetrics Continuous Optimization Platform, an integrated approach to SEO and SEO can significantly improve the ROI from your web presence.

Multiple surveys and studies have indicated that SEO projects consistently provide extremely attractive returns on investment.  Yet eCommerce and online marketing teams frequently struggle to quantify SEO ROI: both prior to the project as part of an internal budgeting process, and after the project to evaluate its success.  Using a recent case study of a global powersports company, we will demonstrate how Coremetrics Digital Agency worked with the client to optimize their lead generation engine by integrating Search Engine Marketing with Coremetrics’ Web Analytics. Building on this SEM experience we then targeted keyword phrases with the potential for the highest, measurable SEO ROI.

We will show the virtuous circle at play between SEO and SEM:

image

For instance, you can see significant improvements to your SEM campaigns by applying lessons learnt from analyzing your SEO efforts (such as which keywords drive most interactions).

Attend this upcoming seminar with Coremetrics’ John Zoglin, Senior Director, Search Marketing Services to learn more.

Date: Wednesday, December 15, 2010
Time: 1:00 PM EST | 10:00 AM PST 
Register now!

More about Coremetrics


Blog early, share late: research findings

Early birds catch the blogworms. Or so suggests research by blogging metrics maniac Dan Zarrella. You have the best chance of getting eyeballs to your posts if you get that content out before 10am US Eastern time. In a recent webinar hosted by Hubspot, Dan unleashed a torrent of findings from his surveys and research of over 170,000 blog posts.

This fine infographic does a great job of summing up general reading/feedback trends seen across the blogs studied:

Whether it’s views, links or comments, most activity happens early in the day. Saturday is a big day for commenting. Which could well be related to this activity on social networks:

Retweeting follows a similar path. It looks like most people read content early in the day, with little variance across the week. As we get nearer the weekend, people start getting social: whether that be retweeting on Twitter or sharing on Facebook (and getting around to commenting).

Judging by the success of this webinar, interest in blogging definitely isn’t on the wane, which makes me somewhat skeptical about a recent study suggesting that although corporate blogging isn’t exactly dead, it’s reached saturation point.

There was no evidence of this during Dan’s study of blogging, which had the Twittersphere ablaze for the full hour of the presentation. You’ll see there was particular interest in the tie-up between blogging and other social media: in particular those duelling siblings Twitter and Facebook. And that’s where blogging can really come into its own: as the content destination for inbound marketing tactics across Facebook and Twitter.

To my mind the Dan’s research also highlights a key difference between search- and social media marketing. For search marketing, attracting those indefatigable search bots that trawl the web for new content is a time-independent task. Just make sure you get content out in short order to win favor from the recency filter was the long and short of what I was told not so long ago by search experts here at IBM. The time of day really has little importance: algorithms aren’t more likely to read posts in the mornings.  Whereas this research from Dan bears a closer resemblance to the findings you might see around email marketing which is often deemed to be time-sensitive. Readership is near-synchronous and content is highly perishable. And if you are blogging outside the time-zone of your key audience, beware. Your content could well end up overlooked. As you may have noticed, I’m taking Dan’s messages to heart and working on getting this content out in a timely fashion. Right, now time for breakfast!

For further details on this study, check out the aforementioned post by Dan or listen to the On Demand recording of Dan Zarrella: Science of blogging

Cognos 10: what does social networking bring to business intelligence?

In my previous life as a webmaster I was called on to develop monthly web performance reports for consumption by the whole marketing organization. At one time these had been documents that were mailed around, but we decided the best approach was to build a web interface with charts and diagrams that would be updated monthly.

We showed standard metrics. Stuff like this:

image

Each month I’d send out an email with a link to the latest report with my notes on site performance each month. For instance, I’d point out from looking at the graph on the left that although traffic had dropped this month, this is a seasonal variation. For the graph on the right, I’d say I wasn’t sure why our search traffic had grown: this is something I’d investigate with the various individuals running search campaigns (meaning for 90% of the people on the email distribution, the answer would end up in an Inbox far, far away).  

How much smarter we could have been if we’d have had access to a system like Cognos 10 that marries business intelligence/analytics with social networking capabilities that allow you to add that layer of insight on top of the data.

For instance, here’s a standard chart:

image

and here’s the same chart with the addition of related Lotus Connections discussions:

image

Going back to my examples above, if I was showing yearly traffic figures, I can use this discussion area to record what I know about seasonal variations. Now if someone receiving the report didn’t agree with my evaluation, they are free to comment on it. As for the discussion I’d need to have with my search marketing folks about why the search traffic has spiked, I can set this up from the same page:

image

…with the thread of the discussion unfolding below the graphs and charts to which it relates. Anyone wishing to follow up on the status of the question can go to that page and scan the thread to see the outcome.

I should point out that the Cognos folks have taken this a step further: integrating activities as well as discussions. The data is now more ‘actionable’. Let’s say you are looking at global sales data and you notice a slump in a certain geographic region. You can use the new functionality to setup an Activity to address this, with a number of associated tasks assigned to different sales people or teams. Over time you can evaluate their actions against the performance data all from within the same interface.

And while we’re talking about the sales team, another new feature in Cognos 10 makes it easier to access reports while on the go, directly from your smart phone:

image

One feature I’d love to see in future releases of Cognos is the ability to tie conversations/activities to given points on a graph, as opposed to just having these attached to the page of a report. As an example, the popular SoundCloud music hosting service has gained a lot of traction by allowing music enthusiasts to comment on a particular point in a music track:

image

(each blue bar represents a separate comment)

Maybe something for a future release?

Delaney Turner has a post with more information on Cognos 10, including a link to an excellent interactive demo.

Also check out the Cognos product pages.

What is the ideal blog template width?

Got caught up on a discussion around what screen resolutions most people use these days to view blogs. Made me dig into the trusty Google Analytics to see if I could discover anything for this very blog.

This is what I found:

Less than 10% of the audience use 1024×768 or less. Given that the majority of visitors are using around 1280, I’d suggest you can happily design a template at 1,000px with the sound knowledge that practically none of your audience will have a horizontal scroll. This is slightly wider than the advice from BloggingPro earlier this year.

Looking at some of the top blogs out there, Mashable weighs in at around 970px wide. Politico is a royal 1,000px and Robert Scoble’s blog sits at around 960px.

Any web designers out there know how this compares with regular websites?

Your blog homepage: To excerpt or not to excerpt?

I manage a small coterie of blogs that on average receive about 20,000 page views per month. Somewhere between 16-19% of these visitors touch the homepage somewhere on their travels. The clear majority actually enter through the homepage. Bottom line: the homepage is pretty damned important.

This recently brought me onto an argument on what is the most effective way of displaying posts. Here I’m not talking about the intricacies of laying out the front page of your blog like a magazine or a photo gallery. Rather just talking the basics: what options do you have with the straight forward roller approach that originally was one of the key defining features of a blog and arguably still remains the most common blog format. It commonly looks something like this:

This opens up a big question: how much of each post do you show on the homepage?

Most blogging platforms, including WordPress, Blogger and the Lotus Connections platform we use here at IBM show the entire post on the homepage. However, you’ll also see many popular blogs abbreviating posts on the homepage with just an excerpt and a link to the individual post in its entirety, eg. Altimeter, Read Write Web and TechCrunch.

Which approach is better?

Here I my thoughts on the blog excerpting question:

Advantages of excerpting posts

  • Easy to scan many posts
  • Easy to track interest/engagement at the post level
  • Added Flexibility: decide on length of excerpt, inclusion of images
  • Works well for group blogs: you have a greater chance of seeing the multiplicity of views from different authors upfront

Advantages of listing full post on homepage

  • Remove an extra click between viewer and full blog post
  • Good for display of short posts (‘Read More…’ could link to only one more sentence)
  • Attracts more comments (visitors can often comment directly from the homepage)

As you can see, there is no clear-cut solution, but I favor the approach of displaying excerpts on the homepage with links to the full articles. Why? To appease the scanners. A homepage listing multiple excerpted posts makes it easy to move quickly through the content and see if you find something relevant. The caveat being if you are a blogger that tends to write short, pithy posts.

Am I over-fixating on a minor detail? Probably. As Mark Murnahan points out, content generally trumps structure. A well written, relevant piece will do well, whether or not it’s excerpted on the homepage. However, I would still contend that if you run a network of blogs and have to and have key objectives and targets to achieve (for instance average page views per article, number of comments per article) playing around with the homepage structure can have a dramatic effect.

How to excerpt with WordPress

If you are blogging using a standalone version of WordPress you can simply excerpt your homepage posts using the the_excerpt function within your homepage template. This will display the first 55 words and will strip out all images and HTML. If you want more control, you can use the Advanced Excerpt plugin which gives a lot more flexibility. I see the main advantages being the option to select which HTML tags to include and options over how much of the post to excerpt.